The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Business Review !Federal R eserv e Bank o f Philadelphia July • A ugust 1 9 9 4 ISSN 0 0 0 7 - 7 0 1 1 Business Review The BUSINESS REVIEW is published by the Department of Research six times a year. It is edited by Sarah Burke. Artwork is designed and produced by Dianne Hallowell under the direction of Ronald B. Williams. The views expressed here are not necessarily those of this Reserve Bank or of the Federal Reserve System. SUBSCRIPTIONS. Single-copy subscriptions for individuals are available without charge. Insti tutional subscribers may order up to 5 copies. BACK ISSUES. Back issues are available free of charge, but quantities are limited: educators may order up to 50 copies by submitting requests on institutional letterhead; other orders are limited to 1 copy per request. Microform copies are available for purchase from University Microfilms, 300 N. Zeeb Road, Ann Arbor, MI 48106. REPRODUCTION. Perm ission must be obtained to reprint portions o f articles or whole articles. Permission to photocopy is unrestricted. Please send subscription orders, back orders, changes o f address, and requests to reprint to Publications, Federal Reserve Bank o f Philadelphia, Department o f Research and Statistics, Ten Independence Mall, Philadelphia, PA 19106-1574, or telephone (215) 574-6428. Please direct editorial communications to the same address, or telephone (215) 574-3805. 2 JULY/AUGUST 1994 MANAGING THE PUBLIC DEBT D. Keith Sill The government is issuing less long-term and more short-term debt. In fact, the Treasury is moving toward borrowing primarily at maturities of less than three years. Can the government save money on interest payments by altering the aver age maturity of the debt? What other things, besides interest costs, need to be considered? Keith Sill looks at various aspects of managing the public debt. THE AUTOMATED CLEARING HOUSE SYSTEM: MOVING TOWARD ELECTRONIC PAYMENT James McAndrews The option of making payments through the Automated Clearinghouse (ACH) has been available for quite a while. Yet, in spite of the advantages of paying via the ACH—speed, lower processing costs— many individuals and companies have been slow to exercise the option. Why? James McAndrews outlines other consid erations, such as perceptions of control and adopting a standard communication format, that need to be weighed when comparing these two methods of pay ment. FEDERAL RESERVE BANK OF PHILADELPHIA Managing the Public Debt A X JL s the Clinton administration and Con gress wrestle with government spending and deficit reduction, the size of the public debt and interest payments on it are much in the news. The administration, in its 1993 budget plan "A Vision of Change for America," claimed that the government could save about $11.5 billion over the next four years if it issued less long term debt and more short-term debt to finance deficits, because short-term debt generally has a lower interest rate than long-term debt. In May 1993, the Treasury Department announced * Keith Sill is an economist in the Research Department of the Philadelphia Fed. D. Keith Sill* that it would begin reducing the amount of long-term debt that it issued. As a result, the Treasury now offers 30-year bonds semiannu ally (instead of quarterly) and has eliminated issues of seven-year notes. The Treasury is m oving tow ard borrow in g prim arily at maturities of less than three years. By altering the average maturity of the debt the government hopes to save money on inter est payments. Does the average maturity of the debt really matter? Should governments issue short-term debt or long-term debt or maintain a balance between the two? Are there other considerations besides interest costs that are important to consider in choosing an average maturity of the debt? 3 BUSINESS REVIEW JULY/AUGUST 1994 great deal in the postwar period, ranging from THE PUBLIC DEBT IN THE U.S. The deficit, or the excess of government a high of 124 months in 1946 to a low of 29 expenditures over its revenues, is the amount months in 1975 (Figure 1). Also, during this of new borrowing the government must under period the average maturity of the debt de take in a year; the debt is the accumulation of all clined when the debt-to-GDP ratio declined past deficits. At the end of 1993, the interest- and rose when the debt-to-GDP ratio rose. In bearing portion of federal government debt 1992 average maturity was about 70 months. held by the public stood at slightly over $2.9 The Treasury's recent changes will shorten the trillion. The federal government ran a deficit of average maturity of the debt some 12 months $254.7 billion in 1993, a number much smaller (to 58 months) by 1998. So, even though the than the size of the public debt. If the govern Treasury is reducing the average maturity of ment persistently runs deficits, the public debt the debt, it will still be about twice as high as the accumulates. If the government runs budget postwar low in 1975. surpluses, the public debt declines. Most of the government debt is in the form DOES DEBT MATURITY MATTER? How does a change in the average maturity of Treasury securities such as Treasury bills, Treasury notes, and Treasury bonds. In 1992, of the public debt affect the economy? Econom for example, such securities accounted for about ic theory says that under certain circumstances 86 percent of private-sector holdings of inter the average maturity of the debt is irrelevant for est-bearing public debt. The remaining 14 economic welfare. In this case debt manage percent was composed of private-sector hold ment policy is neutral with respect to the econ ings of savings bonds and holdings of certain omy. types of securities issued by agencies of the U.S. gov FIGURE 1 ernment such as the Feder al Housing Administration Debt/GDP Ratio and Average Maturity and the Federal Deposit of Debt Insurance Corporation. Months Percent The size of the public 140 IZU debt relative to the size of 120 the U .S. econom y has 100 show n fairly d ram atic 100 80 m ovem ent since W orld War II. If we look at the 80 60 public debt relative to Gross Domestic Product 60 40 (GDP), the ratio has varied 40 from less than 20 percent to 20 over 100 percent (Figure 1). n 20 The maturity of a securi 46 50 54 58 62 66 70 74 78 82 86 90 ty is defined as the length of time until payments from the security expire. The Note: The public debt used in computing debt/G DP ratio is the total interest-bearing average maturity of the public securities held by private investors. public debt has varied a Source: Treasury Bulletin, various issues, Department of the Treasury. 4 FEDERAL RESERVE BANK OF PHILADELPHIA Managing the Public Debt D. Keith Sill This debt neutrality proposition depends on trality proposition is a useful starting point whether households and investors can trade in from which to consider debt maturity policies. securities in such a way as to completely offset The extent to which departures from neutrality any actions that the government takes regard occur is an empirical matter.1 ing the mix of debt and taxes that it uses to finance its expenditures. If households can trade in securities so as to undo the financing mix put in place by the government, any partic 1The empirical results on the effects of debt management ular financing mix will be irrelevant in the sense policies are mixed. Two representative studies are present that household consumption and savings deci ed in the 1992 volume by Agell, Persson, and Friedman. The study by Agell and Persson finds that debt management sions are unaffected by how the government policies have little consequence for relative asset yields. finances its spending. (See A Case o f Debt The study by Friedman finds a much more significant im Neutrality for an example of this neutrality pact of debt management on asset yields. proposition.) H ow ever, this strong A Case of Debt Neutrality neutrality result relies pri marily on three assump Assume that the three assumptions for debt neutrality hold. Suppose tions, some of which clearly the government issues debt in the form of one-year and two-year discount do not hold in reality: (1) bonds, each of which pays $1 at maturity. Assume further that the current households correctly recog price of the one-year bonds is $0.95 and the current price of two-year nize the link between the bonds is $0.90. For simplicity we will allow fractions of a bond to be government budget con bought and sold. If the government issues one additional one-year bond straint and household bud and uses the proceeds ($0.95) to buy back 1.055 units of two-year bonds get constraints as well as (since $0.90 times 1.055 is $0.95), there is now more one-year debt and less the relationship between two-year debt, and government spending and taxes are unchanged. Households, in aggregate, have purchased one additional unit of onecurrent debt and future tax year debt for $0.90 and financed that purchase by selling back to the es, (2) tax rates do not affect government 1.055 units of two-year debt (which raises the $0.90 needed the relativ e prices that to buy the one-year debt). Aggregate consumption by the households is households face (such taxes unchanged initially. At the end of the first year, the government has to are called nondistortionary), raise $1 in taxes to pay off the new one-year debt that it issued. But and (3) the set of investment households can use the proceeds ($1) of their purchase of one-year debt portfolio choices available to pay the higher taxes. Hence, at the end of the first year households can to households is unaffected maintain the same level of consumption as before the average maturity of by the government action. government debt was shortened. At the end of the second year, house If these assumptions are holds have $1,055 less coming in because of their sale of two-year bonds violated, a change in the way back to the government. But government liabilities have fallen by $1,055 government spending is fi because less two-year debt is outstanding. The government could thus nanced will change relative lower taxes by $1,055, and again, household consumption at the end of the prices in the economy and second year would be no different than it was prior to the government action. hence redirect resources. In Since households are able to undo the change in government financing, this case the financing mix any particular mix of debt and taxes the government uses to finance its is not neutral, and a change spending will not affect household consumption and savings decisions. in the average maturity of Households will merely readjust their portfolios in response to the the debt can affect the econ government action. In this situation the debt structure is neutral; it has no omy. Nonetheless, the neu real effects. 5 BUSINESS REVIEW JULY/AUGUST 1994 INTEREST RATES AND DEBT MATURITY slope of the yield curve changes over time. In Bearing in mind the debt neutrality proposi 1954 and 1990 the yield curve had an upward tion, why might the government try to lower slope, indicating that the interest rate on long the interest costs of its debt? If taxes distort term debt exceeded that on short-term debt. In economic activity, lower interest costs mean 1965 the yield curve was approximately flat: less distortion, since tax revenues are used in long-term debt paid about the same interest part to pay interest on the debt. The interest rate as short-term debt. In 1980 the yield curve rate that the government must pay on its bonds was downward sloping, indicating that the often changes with the time to maturity of the interest rate on long-term debt was lower than bonds. If the government's objective in manag that on short-term debt. Is there a "normal" shape, or slope, to the ing the public debt is to minimize interest costs, perhaps altering the average maturity of the yield curve? If we compare short-term and long-term interest rates over time, we see that debt can achieve it. Term Structure of Interest Rates. The yield generally long-term rates exceed short-term curve conveniently summarizes the relation rates, suggesting that the normal shape of the ship between the term to maturity of govern yield curve is upward sloping. The steepness of ment debt and the interest rate (Figure 2). This the yield curve, which is measured by the gap relationship between yield and maturity is called between the two lines in Figure 3, varies quite the term structure of interest rates. The hori a bit, but there are few episodes in which the zontal axis shows the time to maturity of the yield on short-term government bonds exceeds security, and the vertical axis shows the interest that on long-term government bonds. A Theory of the Term Structure. Econorate, which is measured by yield to maturity.2 Notice that the relationship between yields and ma FIGURE 2 turities changes over time. For one thing, when com Treasury Yield Curves paring the yields for 1954, 1965, and 1980, we see that the yield curves shifted up over time, reflecting a gen eral trend of rising interest rates. Next, we see that the 2Yield to maturity is defined as the interest rate that answers the following question: if an in vestor were to buy a bond and hold it until it matured, what av erage annualized return would he get over the life of the bond? For example, if an investor were to pay $100 for a bond that pays $121 in two years, the yield to maturity would be 10 percent. This follows from the fact that $100 x 1.10x1.10 = $ 121. 6 Maturity (years) Source: J.H. McCulloch and Heon-Chul Kwon, "U.S. Term Structure Data, 1947 1991," Ohio State University Working Paper 93-6. FEDERAL RESERVE BANK OF PHILADELPHIA Managing the Public Debt D. Keith Sill conclusion follows if investors expect the yield on one-year bonds to average 5 percent per year, but don't know for sure. If investors know that one-year interest rates will rise above 5 percent per year in the future, the yield to maturity on the 10-year bond should be above 5 percent. Absent a risk premium, an upwardsloping yield curve means that investors be lieve future short-term interest rates will rise, while a downward-sloping yield curve sug gests that traders believe future short-term interest rates will fall. The risk premium can arise because inves tors typically do not like bearing risk. Long term bonds are risky because future interest rates are uncertain and because uncertainty about future interest rates translates into un certainty about future bond prices. That un certainty could work in investors' favor, or it could work against them. The manner in which long-term bonds act as 3An excellent survey of theories of the term structure is the 1990 article by Robert Shiller. a hedge against future income uncertainty de termines whether the risk premium is positive or neg FIGURE 3 ative. For example, sup pose investors could hold a Interest Rates of the United States bond whose price is high 1946 -1992 when income is unexpect edly low and whose price is low when income is unex pectedly high. Investors would be willing to pay a premium for such a bond because it offers them some insurance against their un certain income: in a year when income is low, the investor could cash in the bond and receive a capital gain (since the price at which he sells the bond is higher than the price at 46 50 54 58 62 66 70 74 78 82 86 90 which the bond was pur chased), helping him to Source: J.H. McCulloch and Heon-Chul Kwon, "U.S. Term Structure Data, 1947 maintain his level of con 1991," Ohio State University Working Paper 93-6, and the Federal Reserve System. mists have developed and tested several theo ries to explain why the term structure of inter est rates behaves as it does over time.3 One such theory is called the expectations theory of the term structure, which states that the yield to maturity on a long-term bond is equal to a weighted average of expected future short term interest rates plus a risk premium. It seems reasonable to suppose that the yield on long-term bonds is related to expected future short-term interest rates. Suppose investors know that the interest rate on one-year bonds will average 5 percent a year over the next 10 years. In this case the risk premium would be zero, and investors will buy and sell 10-year bonds until their yield to maturity equals the average of those expected one-year rates, or 5 percent. Absent a risk premium, the same 7 BUSINESS REVIEW sumption. Such a bond would have a negative risk premium. A bond with a positive risk premium would be one whose price is low when income is unexpectedly low. In this case, when the investor receives low income and cashes in the bond, he will take a capital loss. Thus, an investor would have to receive some compensation, in the form of a higher return, for investing in such a security. In this case the risk premium would be positive.4 Although theory suggests that the risk pre mium on long-term bonds can be either posi tive or negative, the normal, upward-sloping shape of the yield curve suggests that the pre mium is positive.5 Minimizing Interest Costs. If the yield curve is upward sloping, should the govern ment borrow long or short to minimize interest costs? First, suppose there's no uncertainty about future short-term interest rates (which 4An alternative theory about why the yield to maturity on a long-term bond may differ from the average of expect ed one-year interest rates is called the preferred habitat theory. This theory, which was developed by Franco Modigliani and Robert Sutch (1966), states that investors have preferred maturities that correspond to their invest ment horizons. For example, if you were investing for a child's college education, you may choose to invest in a long-term bond rather than a series of short-term bonds. The premium (negative or positive) associated with a par ticular maturity then depends on the supply and demand for funds at that maturity. Suppose that lenders prefer to lend with a short-term commitment and borrowers want to borrow long term. Then there would need to be a positive premium on long-term debt to get lenders to loan funds for a longer period than they would otherwise want to. 5In the absence of a risk premium, the usual upward slope of the yield curve suggests that short-term interest rates are expected to rise. In actual practice, short-term interest rates are usually just as likely to rise as to fall. If the risk premium were indeed zero, this suggests that bondmarket traders are making persistent errors in forecasting interest rates, which seems unlikely. On the other hand, if the risk premium is positive, the yield curve would tend to have a normal upward slope, and persistent errors in fore casting future interest rates need not occur. 8 JULY/AUGUST 1994 implies that the risk premium will be zero). The expectations theory implies that future short term interest rates will be higher than current short-term interest rates. In this case, even though short-term rates will be higher in the future, it does not matter whether the govern ment borrows short or long— the interest cost will be the same. A simple example will help to make this clear. Suppose the one-year interest rate today is 5 percent, and the one-year interest rate one year from today will be 10 percent. The govern ment decides to borrow $1000 and repay the borrowing at the end of the second year. If the government borrows using one-year debt, at the end of the first year it must repay $1050. If the government rolls over the debt, at the end of the second year it will have to pay interest on the $1050, so that total interest and principal due at the end of the second year is $1155. What would the government's cost be if it used two-year debt instead? Since there is no risk, investors would demand the same return on the two-year bond as on the sequence of oneyear bonds. Using the expectations theory, the yield to maturity on the two-year bond is the average of the one-year interest rates, which is 7.5 percent. At the end of two years, the total cost of borrowing for two years is the same ($1155), regardless of whether the government borrows short or long.6 If we introduce uncertainty, the picture be comes more complicated. Now bond-market traders form expectations of future interest rates. Further, the introduction of uncertainty brings the risk premium into the picture. If the risk premium is positive, on average the gov ernment will have a lower interest cost by 6The exact formula for the two-year rate gives an interest rate slightly lower than 7.5 percent because of the effects of compounding interest. In an environment with no risk the formula for the implied two-year rate is given by (l+ i2 r)2 = (1+.05)(1+.10). FEDERAL RESERVE BANK OF PHILADELPHIA Managing the Public Debt borrowing short term. But the lower average interest cost comes at the price of higher uncer tainty concerning the final payment. Suppose again that the government borrows $1000 today. The one-year interest rate today is known to be 5 percent, but the one-year interest rate one year from today is not known. Sup pose bond traders believe that there's a 25 percent chance that the interest rate next year will be 8 percent, a 50 percent chance that it will be 10 percent, and a 25 percent chance that it will be 12 percent. Consider first the strategy of borrowing short term. At the end of the first year the govern ment will owe $1050 with certainty. If it rolls over the debt, at the end of the second year, there's a 25 percent chance that the government will owe $1134, a 50 percent chance that it will owe $1155, and a 25 percent chance that it will owe $1176. On average, the government will owe $1155. Now suppose the g overnm ent decides to borrow using two-year debt. What will its cost be in this case? If we assume that the expecta tions hypothesis is true and that there's a pos itive risk premium, the yield to maturity on a two-year bond will be the average of today's one-year interest rate and the expected oneyear interest rate in the second year, plus the risk premium. The average of the short-term rates is 7.5 percent (the average of 5 percent and the expected 10 percent). Thus, the yield to maturity on the two-year bond is 7.5 percent plus the risk premium. Let's assume the risk premium is 0.2 percent, so the yield to maturity is 7.7 percent. Then, the interest and principal that has to be repaid at the end of two years is $1159.93 ($1000 x (1.077) x (1.077)) = $1159.93). Should the government borrow long term or short term? In the example, the expected inter est cost to the government of borrowing short term is $155. If the government borrows using two-year debt, the interest cost will exceed $155. This result seems to favor short-term borrowing. However, by borrowing short D. Keith Sill term the government faces a risky outcome. In the example, there's a 25 percent chance that borrowing short term will cost $176, which exceeds the cost of borrowing using two-year debt. On average, the cost of borrowing short term will be lower than the cost of borrowing long term, but the lower interest cost comes at the price of a risky outcome.7 BENEFITS OF LONG-TERM DEBT We have seen that if the government tries to manage the public debt to minimize interest costs, it can lower its interest cost, on average, by borrowing short term rather than long term, but at the price of bearing greater risk. Aside from this interest rate minimization issue, are there other factors that the government should consider when planning the average maturity of its debt? Debt Maturity and Insuring Against Risk. Economic theory suggests that debt of different maturities may offer investors different oppor tunities to insure against economic uncertain ty. We will frame this discussion in terms of a simple economic model in which consumers live for two periods.8 We can think of the first period of life as the working years and the second period of life as retirement. In the first period, consumers work and invest in an asset that is risky in the sense that the return is unknown to investors at the time of invest ment. Investing in the risky asset is like buying corporate stocks to save for retirement. How 7We have neglected to mention transactions costs. By having a shorter average maturity of debt, the government rolls over the debt more frequently and thus pays more in transactions costs. For example, if the government bor rowed for 10 years, it could make one transaction by issuing one 10-year bond, or it could make 10 transactions by issuing 10 one-year bonds. The higher transactions costs must also be considered in assessing the extent to which the government saves money by issuing short-term debt. 8This argument is based on an article by Douglas Gale. 9 BUSINESS REVIEW ever, in any period only one of the two gener ations alive at that date bears the risk of the investing, namely the retirees. Everyone would be happier if some of the asset risk could be transferred from the retirees to the workers. This intergenerational risk-sharing can be accomplished by introducing government debt into the economy. Suppose the government introduces one-period debt into the economy. This debt offers young investors a safe asset to invest in. Since investing in public debt carries no risk, it allows the young to attain, with certainty, some amount of consumption when they retire. If investors don't like risk, they may be better off if they have the opportunity to guarantee some amount of consumption when they retire, compared with investing all of their savings in the risky asset. By issuing one-period bonds, the govern ment allows intergenerational risk-sharing in the following sense. Buying a one-period gov ernment bond is like buying a claim on the next generation. When the young buy bonds, they hold them until they retire; the bonds are then paid off by the government. But the govern ment pays off the bonds by transferring re sources from the new young generation of workers to the retirees. Thus, by transferring resources from the young to the old, the debt serves to guarantee retirees some level of con sumption. Debt of maturity longer than one period would be more risky for these investors be cause of capital gains and losses that can occur when economy-wide rates of return change. But under certain circumstances this riskiness of long-term debt could be advantageous to investors even if one-period debt is not. Sup pose that investors observe that the return to the risky asset is high, and further, they expect the return on risky assets to be high next period. In this situation the current price of a twoperiod bond will be low.9 Similarly, if the return to the risky asset is currently low, the price of two-period bonds will be high. How 10 JULY/AUGUST 1994 ever, if the price of bonds is high when the return to investment is low, the two-period debt is a better hedge against the risky invest ment. Why is this so? The argument is much the same as that in our discussion of the risk premi um. Take the case of investors who purchased both two-period bonds and the risky asset to save for their retirement. At retirement, these investors will want to sell their bonds (which have become one-period maturity bonds) to the new young generation. If the return on the asset turns out to have been low, new investors, seeing that the return to the asset was low, expect a low return to their investment in the asset (remember that we are assuming a posi tive correlation in investment returns). There fore, the new investors will want to buy bonds from the retirees, bidding up the price of those bonds. These retirees get a capital gain (an appreciation in the bond price) that in part compensates them for the low return on the risky asset. No such capital gain would be realized if the retirees had purchased oneperiod debt instead of two-period debt. This argument is not limited to two-period bonds. Thus, the economy could be better off if investors had the opportunity to invest in long-term debt securities because long-term debt might provide better insurance against the uncertainty associated with risky assets.1 * 0 9Returns on bonds and returns on the risky asset will be linked by investors' demand for the two alternatives. If the expected return to the asset rises, while the uncertainty associated with the asset return remains unchanged, then investors have an incentive to shift their investment funds toward the risky asset and away from bonds. As investors shift funds out of bonds, the price of bonds falls and the return on bonds rises. 10Referring to the discussion of debt neutrality on page 5, the reason that debt maturity matters in the example just given is that trading in government securities offers inves tors opportunities that they otherwise would not have and so assumption (3) is violated. FEDERAL RESERVE BANK OF PHILADELPHIA Managing the Public Debt Confidence Crises. Another argument in favor of governments' issuing long-term debt can be made. Long-term debt can raise inves tors' confidence that the government will be able to meet its obligations in the event of a crisis. A 1990 paper by Alberto Alesina, Alessandro Prati, and Guido Tabellini devel ops this argument using a case study of the public debt in Italy. The Italian debt-to-GDP ratio is close to 100 percent, and the Italian government has to pay a steep premium to borrow long term. Alesina and associates show issuing long-term debt may be beneficial to the government, even though it is more costly than short-term debt, because long-term debt can help to avoid confidence crises. A confidence crisis could occur if govern ment bondholders thought that the govern ment might have difficulty making payments on the debt. Suppose the government finances its borrowing by issuing only one-year debt. In that case, a large quantity of the debt comes due each year, and the government must borrow a large quantity each year, both to finance any current deficit and to roll over the existing debt. If investors thought the government might have difficulty repaying its debt obligations, they could all demand repayment of their debt holdings. The government would find itself unable to borrow to roll over existing debt. The government would have to either raise taxes substantially to pay off debt holders or default on the debt.1 1 On the other hand, if the government issued long-term debt and had an evenly concentrated amount of debt coming due each year, it could diminish the likelihood of a confidence crisis. By issuing long-term debt to finance deficits, 11If taxes are distortionary, economic theory suggests that governments should try to smooth taxes over time. Distortionary taxes and tax smoothing are the reason that the maturity structure of the government debt matters in this model. D. Keith Sill the government has a smaller quantity of debt that comes due each year. Therefore, this strat egy may raise investor confidence in the gov ernment's ability to meet its obligations, and runs on the government debt may become less likely. In the case of the Italian debt, Alesina and associates note that by issuing long-term debt and reducing the risk of a debt crisis, the government could lower the risk premium on the entire maturity structure of the debt and, therefore, lower debt-servicing costs.1 * 2 DEBT POLICY AND FISCAL INCENTIVES We have examined several different theories that point out some of the costs and benefits of both short-term and long-term debt. An opti mal debt maturity structure takes these factors into account, as well as the incentives that current government policy places on the poli cies of future governments. Time-Consistent Policy. Economists have considered how the maturity of the public debt can be used as part of a strategy to implement a fiscal policy that is optimal over time. In a dynamic environment, fiscal policy takes the form of a plan for both the present and the future. If today's government forms a fiscal plan, that plan has implications for future tax rates, future government spending, and future borrowing. But can we guarantee that some future government will find it optimal to stick to the plan that we develop today? In general, the answer is no, so we say that the plans are not time-consistent. The issue of time-consistent plans is dis cussed in more detail in a 1985 article by Herbert Taylor in this Business Review. For our purposes a simple example will help clarify the idea. The United States incurred a large debt when it 12The confidence crisis story is less applicable to the U.S. than to countries such as Italy. In the U.S. the default premium on government debt is considered to be virtually zero. 11 BUSINESS REVIEW fought the war for its independence. The gov ernment was able to borrow because it prom ised to repay the debt after the war. However, once the war was over, many Americans advo cated defaulting on the debt because repaying creditors would require an increase in taxation; thus, the government had an incentive to devi ate from the policy im plemented earlier. Alexander Hamilton, the first Secretary of the Treasury, argued against this time-inconsisten cy, realizing that in the future the new govern ment would likely need to borrow again. Had the government defaulted on the war debt, borrowing in the future would have been more difficult and costly. Debt Maturity and Optimal Fiscal Policy. In general, successor governments will have an incentive to deviate from an optimal fiscal policy put in place by today's government. But economic theory suggests that the maturity structure of the public debt can help provide incentives for future governments to stick to a fiscal plan developed today. This happens because a government that inherits a public debt has reduced flexibility: it must pay interest on the inherited debt and either pay off debt coming due or roll it over.1 If a government 3 inherits a large quantity of public debt that comes due during its time in office, its incen tives, say, with respect to taxation, may be different than if the inherited debt is long term and thus not all coming due during the govern ment's tenure. Suppose today's government believes high er taxes and higher inflation reduce economic welfare. The government might then form a fiscal plan that tries to set current and future taxes and inflation in a way that increases society's well-being. A strategy for the public 13We are assuming that the costs of defaulting on the debt are so high that future governments do not consider defaulting as a policy option. 12 JULY/AUGUST 1994 debt could be a key part of this calculation, since debt allows governments to smooth taxes over time and to reduce the temptation for future governments to deviate from the fiscal plan. The maturity of the public debt can be used to lessen the government's incentive to try to use inflation to reduce the value of its debt.1 4 Consider the case of a government that inherits a stock of long-term, fixed-rate debt. The gov ernment recognizes that since the debt was issued in the past, the interest payments on that debt are fixed in dollar terms. This gives the government an incentive to increase the rate of inflation so that it can pay off its inherited debt in cheaper dollars. This inflation acts like a tax, and the nominal debt comprises part of the tax base. The real value of the payments that investors receive from their bond holdings de clines when the price level rises.1 5 By reducing the average maturity of the debt, current governments can reduce succes sor governments' incentives to increase infla tion. A government that inherits short-term debt will gain little by increasing the inflation rate. When the debt is short term, it is rolled over frequently, giving the government little opportunity to pay off the debt in cheaper dollars. In addition, any attempt to raise infla tion will be quickly reflected in higher interest rates on short-term debt; investors will de mand to be compensated for higher anticipated inflation. In effect, a greater quantity of short 14This discussion is based on the work of Guillermo Calvo and Pablo Guidotti. 15This argument applies to debt with a fixed nominal face value, which is the predominant form of debt issued by governments. The government has an incentive to raise inflation even if the gains from doing so are illusory in the sense that bondholders, at the time they purchased the bonds, demanded an inflation premium in the form of a higher interest rate. FEDERAL RESERVE BANK OF PHILADELPHIA Managing the Public Debt D. Keith Sill term debt lowers the inflation tax base available to the government and, therefore, lessens the incentive to use inflation to raise revenue. CONCLUSION Deciding on a preferred maturity structure of the public debt involves many consider ations. On the one hand, the maturity structure of the debt may be largely irrelevant for the economy if departures from the neutrality prop osition are small. On the other hand, if the departures from neutrality are significant, then the choice of a debt maturity structure may be guided by factors such as interest cost minimi zation, risk-sharing arrangements, confidence crises, and reinforcing incentives for future policymakers. Economists have not yet reached agreement on the questions of whether there is an optimal maturity of the debt and, if so, what factors are involved. The U.S. Treasury is engaging in a strategy to reduce the average maturity of the public debt. Our analysis suggests that on average, this strategy should reduce the costs of borrowing, but the government also takes on more risk, since future interest rates are uncertain. The shorter average maturity may also weaken the incentives future governments have to use in flation to raise tax revenue. REFERENCES Agell, Jonas, Mats Persson, and Benjamin M. Friedman. Does Debt Management Matter? Oxford: Oxford University Press, 1992. Alesina, Alberto, Alessandro Prati, and Guido Tabellini. "Public Confidence and Debt Management: A Model and a Case Study of Italy," in Rudiger Dornbusch and Mario Draghi, eds., Public Debt Management: Theory and History. Cambridge: Cambridge University Press, 1990. Calvo, Guillermo A., and Pablo E. Guidotti. "Credibility and Nominal Debt," IMF Staff Papers 37 (September 1990), pp. 612-35. Gale, Douglas. "The Efficient Design of Public Debt," in Rudiger Dornbusch and Mario Draghi, eds., Public Debt Management: Theory and History. Cambridge: Cambridge University Press, 1990. Modigliani, Franco, and R. Sutch. "Innovations in Interest Rate Policy," American Economic Review 56 (May 1966), pp. 178-97. Shiller, Robert J. "The Term Structure of Interest Rates," in Benjamin M. Friedman and Frank H. Hahn, eds., Handbook o f Monetary Economics, Volume I. Amsterdam: Elsevier, 1990. Taylor, Herbert E. "Time-Inconsistency: A Potential Problem for Policymakers," this Business Review (M arch/April 1985). Tobin, James A. "An Essay on the Principles of Debt Management," in Fiscal and Debt Management Policies. New Jersey: Prentice-Hall, pp. 143-218. (An early, excellent reference on managing the public debt) 13 Philadelphia/RESEARCH Working Papers The Philadelphia Fed's Research Department occasionally publishes working papers based on the current research of staff economists. These papers, dealing with virtually all areas within economics and finance, are intended for the professional researcher. The papers added to the Working Papers series thus far this year are listed below. To order copies, please send the number of the item desired, along with your address, to WORKING PAPERS, Department of Research, Federal Reserve Bank of Philadelphia, 10 Independence Mall, Philadelphia, PA 19106. For overseas airmail requests only, a $3.00 per copy prepayment is required; please make checks or money orders payable (in U.S. funds) to the Federal Reserve Bank of Philadelphia. A list of all available papers may be ordered from the same address. 1994 94-1 Loretta J. Mester, "Efficiency of Banks in the Third Federal Reserve District" 94-2 George J. Mailath and Loretta J. Mester, "A Positive Analysis of Bank Closure" (Supersedes No. 93-10/R ) 94-3 Dean Croushore, "The Optimal Inflation Tax When Income Taxes Distort: Reconciling MUF and Shopping-Time Models" 94-4 Arthur Fishman and Rafael Rob, "The Durability of Information, Market Efficiency, and the Size of Firms" 94-5 Shaghil Ahmed and Dean Croushore, "The Marginal Cost of Funds With Nonseparable Public Spending" (Supersedes No. 92-2/R ) 94-6 James McAndrews and Rafael Rob, "Shared Ownership and Pricing in a Network Switch" 94-7 Shaghil Ahmed and Dean Croushore, "The Importance of the Tax System in Determining the Marginal Cost of Funds" (Supersedes No. 92-15/R ) 94-8 Joseph P. Hughes and Loretta J. Mester, "Bank Managers' Objectives" (Supersedes No. 93-17) 14 FEDERAL RESERVE BANK OF PHILADELPHIA The Automated Clearinghouse System: Moving Toward Electronic Payment James McAndrews* I n the late 1960s, when a group of California banks first suggested the idea of an electronic system to make low-value, recurring pay m en ts— the A utom ated C learin ghou se (ACH)—some people predicted that the ACH would overtake checks as the main way of making payments. Why then are relatively few payments made through the ACH? After all, the competition between checks and an elec tronic form of payment seems like a race be- * James McAndrews is a senior economist in the Research Department of the Philadelphia Fed. tween a turtle and a hare. Checks (the turtle) have to be physically moved by hand, truck, and air from place to place to reach the checkwriter's bank, while electronic payments (the hare) move in a flash over telephone wires. Electronic payments have a speed advan tage over checks. In addition, each electronic transaction is cheaper to process than a check. Nonetheless, other considerations give the triedand-true technique of payment by check an edge in the contest with its electronic rival. First, the difficulty of finding cheap and effec tive ways to electronically communicate which particular bills have been paid by ACH has slowed its acceptance. Second, creating and 15 BUSINESS REVIEW maintaining the ACH has required large, fixedcost investments by both banks and corporate users, which can offset the per-item cost advan tage of ACH processing. Third, a person who pays by check can benefit from the time be tween when the check is written and when funds are finally transferred from her account; this is called float. Expectations that ACH payments would overtake checks were too optimistic, but many specific uses of ACH have proven successful. Most notably, an estimated 30 percent of the U.S. work force now have wages and salaries directly deposited into their bank accounts by ACH payment. WeTl explore why this and some other uses of ACH have been successful and discuss the history, organization, and cur rent developments in ACH to get a sense of direction for the future of this payment system. To do so, we first need to understand how ACH works. BASICS OF THE AUTOMATED CLEARINGHOUSE The Automated Clearinghouse (ACH) is an electronic system that connects banks so that they can transfer funds between accounts in different banks. While it was not always so, today's ACH system is all electronic: banks use computers linked to a computer at the process ing center and relay payment information over telephone lines.1 The ACH system was designed for small, repetitious payments such as payrolls, mort gage installments, insurance premiums, and utility bills. Repetitious payments are well suited to ACH because they allow the one-time costs of setting up the authorization for pay ment to be spread over multiple transactions.2 The ACH was also designed so that, like checks, ’See the March 1986 and the April 1986 issues of the Economic Review of the Federal Reserve Bank of Atlanta for information on the history and problems of implementing ACH. 16 JULY/AUGUST 1994 ACH transactions can be returned, for exam ple, due to insufficient funds. To allow for returns, money actually changes hands from one to four days (depending on the nature of the transaction) after the payer's bank is noti fied of the ACH transaction to be settled.3 ACH transactions can be one of two types: credit or debit. A credit transaction is initiated by the payer: the customer of an electric compa ny, for instance, relays to her bank her account number at the electric company along with the electric company's deposit account number and bank. Each month the customer can then phone the bank and initiate an ACH credit transaction that will transfer the amount of her bill to the electric company. In the case of a set billing amount, the customer can arrange for that amount to be sent automatically every month. Alternatively, a debit transaction is pre authorized by the payer but is initiated by the payee. In this case the customer signs a form authorizing the utility to debit her account each month. The utility sends her a bill and then initiates payment for the bill at some agreedupon date. Including the costs of accounting, mailing, processing, and transportation (but not includ ing the benefits of control of timing or the information costs of a payee's attempting to determine who paid their bill in a credit trans action), the cost of an ACH transaction is esti mated to be roughly half the cost of a check 2The costs to enroll a person in a federal government direct deposit program were estimated to total $6.94 in 1981; $1.32 of this cost was incurred by the depository institution, and the rest was incurred by the federal agencies using the program. See William Dudley, "A Comparison of Direct Deposit and Check Payment Costs," Staff Studies No. 141, Board of Governors of the Federal Reserve System, November 1984. 3Another electronic payment system, the Fedwire— the Federal Reserve System 's high-value funds transfer network—does not allow returns and transfers money the same day the bank is notified. FEDERAL RESERVE BANK OF PHILADELPHIA The Automated Clearinghouse System: Moving Toward Electronic Payment transaction.4 A recent survey, conducted for the National Automated Clearing House Asso ciation, shows that, for the respondents to the survey, the total bank processing cost of an ACH item averaged 5.7 cents, while the total bank processing cost of a check averaged 10.5 cents.5 Bank processing costs for check-writing or for ACH bill-paying services are reflected in the fees, explicit or implicit, that banks charge their customers. The payer of a check, howev er, may derive benefit from float (float is the value of money between the time the payee's bank account has been credited and the time the payer has money removed from her bank ac count), which means the payer may prefer a check even though the cost of processing an ACH transaction is lower than the cost of pro cessing a check.6 ACH HISTORY AND ORGANIZATION The idea for an automated electronic clear inghouse for interbank payments was devel oped in 1968 when the San Francisco and Los Angeles clearinghouse associations formed a committee to study how to create an electronic clearinghouse. This led to the first automated clearinghouse, operated by the Federal Reserve Bank of San Francisco, in 1972. During the 1970s other regional automated clearinghouses were also formed. The Federal Reserve System supported these private-sector developments 4See David B. Humphrey, The U.S. Payments System: Costs, Pricing, Competition and Risk (Monograph Series in Finance and Economics, nos. 1 and 2, New York University, 1984). 5See Direct Payment Market Analysis, prepared for the National Automated Clearing House Association, Herndon, Virginia, by the Payment Systems Institute, January 1994. 6Scott E. Knudson, Jack K. Walton II, and Florence Young, in "Business-to-Business Payments and the Role of Finan cial Electronic Data Interchange," Federal Reserve Bulletin, April 1994, pp. 269-78, calculated that the value of float to businesses in 1993 ranged from about $0.86 to $1.12 per check. James Me Andrews by operating most of the clearinghouses. The private-sector clearinghouses developed the rules and procedures for making ACH transac tions. In 1974 the American Bankers Association formed the National Automated Clearing House Association (NACHA). Its charter was to develop an interregional network, establish uniform rules nationally, and expand the types of transactions then available. By 1978 the national network, managed by the Federal Re serve System, was operational. Today it pro cesses transactions for well over 20,000 depos itory institutions. Three regional clearinghouses, Arizona, New York, and Hawaii, process their own regional ACH transactions. Visa, the credit card associ ation, created an ACH that began competing with the Federal Reserve's system on a national basis in 1991. In 1995 the Federal Reserve will consolidate its ow n ACH activ ity into a single clearing house facility, which, along with the improved computing equipment now available and the revised software for ACH, is expected to re duce costs. With the consolidation of the Fed eral Reserve ACH system into one national clearinghouse, the private-sector ACH opera tors recognized the need to establish a national clearinghouse to adequately compete with the improved system of the Federal Reserve. (Visa has operated its system nationally since 1991, but the majority of its users are from the West.) As of April 1994, the private-sector ACH ex change (PAXS), consisting of the Visa ACH and the New York and Arizona clearinghouse asso ciations, offers its members a national ACH service. (See Private vs. Public ACH.) ACH GROWTH The number of ACH transactions has nearly tripled since 1986 and their value has more than tripled.7 (See Figure, p. 19.) Most of this growth has come in private, rather than government, transactions because the federal government 17 JULY/AUGUST 1994 BUSINESS REVIEW Private vs. Public ACH The private-sector ACH exchange (PAXS) began offering national ACH processing in competition with the Federal Reserve System in April 1994. If the banks whose customers are party to the ACH transaction are both using PAXS, settlement will occur using the Visa settlement system. If only one of the banks involved is using PAXS, the transaction is settled using the Fed's system. A notable difference between the Fed's system and Visa's is that the Fed's system settles by an exchange between the parties of the gross amount of funds owed, while Visa's settles by the parties' exchanging only the net amount of funds owed. A bank makes a payment to another bank by sending funds, typically balances on deposit at the Fed, to the other bank. In a gross settlement system the banks that are party to offsetting transactions must hold sufficient balances to exchange the gross amounts of the underlying obligations when payment is made. In a netting system the parties take advantage of offsetting transactions, and only the party that owes the larger amount needs to send funds. Netting by using PAXS reduces the amount of deposits that banks need to hold at the Fed. For example, if First Bank owes Second Bank $50,000 in one transaction and Second Bank owes First Bank $200,000 in another transaction, gross settlement means First Bank will have to hold at least $50,000 in deposits, and Second Bank will have to hold at least $200,000 in deposits before the transaction settles.3 Netting means only Second Bank would have to hold deposits when the transactions are settled, and only $150,000, the net amount owed when payment is made. Banks find it desirable to reduce the amount of deposits they must keep at the Fed because they could place those funds in alternative investments that pay higher returns. Banks that send and receive large numbers of ACH transactions would be attracted to a netting service.b But netting can also expose the parties to increased risks, precisely because fewer reserves are available in a time of liquidity crisis.c If a large member of a netting group were to fail to settle on a given day, the other members of the group would be forced to quickly find extra reserves, or they too would be unable to settle. This might lead to a cascade of failures to settle. Unlike the Fed, which can stem such a cascade by creating bank reserves, a private-sector settlement system must plan on some other way to stem the spread of such failures to settle, should such occur. Under certain circumstances, the Visa system relies on unwinding.1 In an unwinding, if a particular bank should 1 fail to settle at the time appointed for the bank to deposit the amount it owes other banks in the system, Visa would remove that bank and all its associated transactions from the day's settlement. After doing this unwinding, Visa would recast settlement with the other banks in the system. If a settlement failure occurs a second time, Visa would not attempt net settlement again, sending all the transactions to the Federal Reserve for settlement instead. While unwinding transactions in a large-dollar-value settlement system poses a significant risk of systemic failure, it's less of a problem for transactions in a small-dollar-value system such as ACH. For example, the July 1992 transactions data for the Third District (described in footnote b) indicates that the system could have settled via unwinding each day had the largest net debtor failed. aThis simple example presumes that banks do not overdraw their accounts at the Fed during the day. Although banks may overdraw to a limited extent, the Fed encourages them not to exceed those limits. See George R. Juncker, Bruce J. Summers, and Florence M. Young, "A Primer on the Settlement of Payments in the United States," Federal Reserve Bulletin, November 1991, pp. 847-58, for a discussion of the settlement process of the Fed. bThe reduction in reserve account balances in moving from gross to net settlement appears to be substantial. For the 10 banks that were the largest users of ACPI in the Third Federal Reserve district and all the banks in the U.S. that had ACH transactions with at least one of these, the average daily gross payments for July 1992 were $1.34 billion; multilateral net payments were only $214 million, just 16 percent of the gross payments. cSee Patrick Parkinson and others, "Clearance and Settlement in U.S. Securities Markets," Board of Governors of the Federal Reserve System, Staff Study 163 (March 1992), for an extended discussion of the risks in settlement systems. dIn addition to unwinding in a settlement failure, Visa uses an extensive array of risk-control devices. The description of Visa's system is taken from "Proposal to the Board of Governors of the Federal Reserve System for a Net Settlement Account for the VisaNet Automated Clearing House System, May 1990," provided to me by Visa. 18 FEDERAL RESERVE BANK OF PHILADELPHIA The Automated Clearinghouse System: Moving Toward Electronic Payment James McAndrews was an early convert to ACH. Today, ACH ment, and to pay for purchases from private firms. transactions are concentrated in four types. Direct Deposit of Payrolls. As mentioned Government Payments. The federal gov ernment has been a leader in using ACH. The earlier, an estimated 30 percent of the U.S. work Social Security Administration, the Department force uses ACH for direct deposit of their of the Treasury, and the Department of Defense wages, an increase from just 4 percent in 1984. have expanded the use of ACH considerably. This makes direct deposit the most common Over 85 percent of Social Security benefits are use of ACH and the one with which people are paid by ACH. The federal government began most familiar. Direct deposit of payroll is an using ACH for direct deposit of payrolls as example of a credit transaction. Both parties often prefer direct deposit. Stud early as 1975. In addition to being a leader in paying Social ies have found that direct deposit is less costly Security benefits and wages through ACH, the than check payment, making ACH attractive to U.S. Treasury and the Department of Defense employers.8* Furthermore, an employer could both have programs under way to convert to ACH most payments to their vendors and 8Dudley, 1984 (see footnote 2 for complete reference), contractors. In the Treasury's program, called Vendor Express, over $61 billion in payments and David B. Humphrey and Allen N. Berger, "Market were made through the ACH, compared with Failure and Resource Use: Economic Incentives to Use Dif ferent Payment Instruments," in David B. Humphrey, ed., $44 billion made by check (and $209 billion by The U.S. Payment System: Efficiency, Risk and the Role o f the wire transfer— used for high-dollar-value pay Federal Reserve (Kluwer Academic Publishers, 1990), pp.45ments) in fiscal year 1992. In the Department of 86 . Defense program over 28 FIGURE percent of major contract paym ents are currently ACH Transaction Volume made by ACH. The fed eral gov ern Number of Transactions Dollar Values ment's use of ACH is likely to expand, since one of the recommendations of Vice President Gore's National Performance Review for the federal government is to use ACH to reimburse expenses for its employees, to make payments to other agencies of the govern- 7Transaction volume for the ACH was about 2 billion payments in 1992, with a value of $7.8 tril lion (see Figure). This compares with 57 billion checks written in 1991 for a dollar value of $66 tril lion. 1986 1987 1988 1989 1990 1991 1992 Note: All dollar values in trillions Source: National Automated Clearing House Association 19 BUSINESS REVIEW JULY/AUGUST 1994 cause hardships for its employees if it attempt ed to exploit float by drawing checks on an outof-state bank, for example, to take advantage of the time between when the checks were written and when the money was actually transferred to their employees' accounts. The long-term nature of the employment relationship miti gates employers' incentive to exploit this ad vantage of checks. Employees frequently pre fer ACH because, with direct deposit, they avoid trips to the bank to deposit paychecks. Often, such trips can take place only at lunch time when banks are congested, so direct de posit avoids a waste of time and energy. Consumer Bill-Paying. Paying bills by ACH is potentially a major convenience for people. According to NACHA, of the almost 2 billion ACH transactions made in 1992, 800 million were consumer bill payments, twice the num ber of 1989.9 However, this represents a small share of the approximately 20 billion bill pay ments made annually by consumers. About half of all ACH bill payments are for insurance premiums, while the remaining payments are evenly split among mortgage loans, utility pay ments, and auto and other loans. Customers can benefit from this service be cause it reduces time and postage in preparing and sending bill payments. Companies can benefit by reducing processing costs through handling fewer checks and obtaining payment in a timely fashion, which allows the company to better manage its cash needs. Corporate-to-Corporate Payments and Cash Concentration. The main business use of ACH has been for "cash concentration." This ACH transaction allows units of a widely dispersed company to send money to a central deposit account. By doing this the company can econ omize on deposits, rather than having a large amount of money in several accounts across the country. ACH has allowed companies to do this much more quickly and at a lower cost than was possible by check. The payment of bills by one company to another, while a potentially beneficial use of ACH, is still overwhelmingly done by check. According to NACHA, in 1992, of the more than 10 billion trade payments made, fewer than 10 million were made by ACH.1 0 9These data on transaction volumes were supplied by NACHA. 10These data on transaction volumes were supplied by NACHA. 20 WHY HASN'T ACH GROWN MORE RAPIDLY? While government use of ACH and the use of ACH for direct deposit of payroll have seen impressive growth, the growth of ACH in con sumer bill payment and corporate trade pay ments has been slow. Partly this reflects tech nological advances that have increased the speed of check collection and processing. Be cause of technological developments in auto mated reading and sorting machines, a great deal of electronic sophistication is now used in processing checks. More rigid collection times and disbursement times for check availability, mandated under the Expedited Funds Avail ability Act of 1988, along with improved tech nology, have reduced delays in check collec tion, reducing float and increasing the accept ability of checks to payees. In the case of consumer bill-paying and cor porate trade payments, ACH payments also have some unique features that have made it difficult for ACH to gain ground on checks. Whose Bill Is It? Paying bills using ACH has encountered two primary difficulties in gain ing acceptance. F irst, in settin g up a preauthorized debit transaction, in which a company directly debits the customer's ac count, the customer gives up the freedom to delay payment if she is temporarily short of funds. Instead she must make sure the money FEDERAL RESERVE BANK OF PHILADELPHIA The Automated Clearinghouse System: Moving Toward Electronic Payment is in her account before the company initiates the ACH debit transaction, which may require greater attention on the customer's part; this effort may not be worth the savings in time and postage of writing and mailing checks. In addition, for many people, writing a check acts as a reminder to verify checking account bal ances and to put money into their checking accounts to ensure that the balance is large enough to cover the checks they have just writ ten. The second difficulty of making bill pay ments occurs with credit transactions. Many companies do not accept bill payments by ACH— only a check will do. The bill-payment service of Germantown Savings Bank in Phila d elp hia v iv id ly illu stra tes this point. Germantown Savings has had a telephone bill paying service since 1979, and many customers use it. Over 45,000 payments are made each month through the service to more than 21,000 different firms. However, only 172 of these firms accept payments by ACH. For all the other firms, Germantown Savings must write checks to pay the bills. Why do many firms choose not to accept bill payment by ACH? Because they are not likely to reap any cost savings by doing so. A credit transaction could arrive on any day of the month, and to easily find out the account num ber of the customer making payment, the com pany needs a computer link with its bank so that the bank can transfer the invoice information in machine-readable form. Even then, the firm must learn how to interpret the standardized account information encoded into the ACH payment, information which, with check pay ment, would arrive with the check on the com pany's own customized invoice. Furthermore, the company must maintain a system to read this information and update its accounts in response to the payment information. This process of learning and maintaining systems to read new ways of conveying information can be costly: witness the difficulty the U.S. has had in James Me Andrews attempting to convert to the metric system of measurement. Thus, the total cost to the com pany of an ACH transaction may end up being higher than the cost of accepting payment by check. As a result of these control and information cost considerations, consumer bill payment has not made as much progress in displacing checks as many had hoped. Some new attempts at solving these problems are being made, espe cially in utility-bill payments. Utility payments are repetitive, and most important, only a small amount of information needs to be sent along with payment—the account number of the cus tomer. Several of the regional automated clearing house associations have taken the lead in pro moting the use of ACH for the payment of utility bills. In particular, the Hawaiian ACH Association, the Mid-America Payment Ex change in Omaha, and the Mid-America Auto mated Payment System in Cleveland all have conducted marketing efforts to promote the use of preauthorized automated utility-bill pay ment. The Hawaii program has probably been the most successful: more than 20 percent of the customers of the Honolulu Board of Water Supply, for example, now pay their bills through ACH, compared with a national average of less than 3 percent of utility bills paid by ACH. Furthermore, about half of all signups between September 1990 and September 1991 took place during the three months of an advertising cam paign. Pacific Bell has created another innovation in paying utility bills through ACH. Its system allows a consumer to call a telephone number, review the amount of her bill, and then instruct the company to debit her account on a particu lar day. This system solves the problem of the consumer's feeling that she doesn't control the timing of her payment in a debit transaction, while it preserves the merchant's preference, when choosing between debit and credit trans 21 BUSINESS REVIEW JULY/AUGUST 1994 actions, for the easier-to-process debit transac tions. Your Format Is Not Talking to My Format. For corporate trade payments a different prob lem arises.1 Whenever payment is separated in 1 time from delivery of the purchased item, in formation must accompany the payment to match it to the corresponding delivery. Often the information can be quite complex. Correc tions to the invoice may be needed, or multiple invoices may correspond to a single payment. Until the early 1980s ACH was unable to con vey such potentially large amounts of informa tion along with the payment instructions. The problem facing ACH was choosing a standard format for the information accompa nying a payment. For example, suppose a firm wishes to send payment information to another firm. To do so, the first firm (or its bank) must translate its own internal format for the infor mation into a standardized format, and the receiving firm must then translate the stan dardized information into its internal format. If the standard chosen by the ACH is cumber some or not widely used by firms in other applications, the translation step could be cost ly and would ultimately inhibit the use of ACH. At the time that the ACH was created, firms were just beginning to engage in electronic data interchange (EDI). NACHA recognized the need to send information along with payment and, in 1983, created a type of transaction, called the corporate trade payment (CTP), to include both payment and invoice information. However, the CTP turned out to be incompat ible with the direction of the emerging stan dards used in EDI; therefore, a receiving firm's computer could not understand the message sent. NACHA soon understood that the CTP format was flawed. By 1985 NACHA devel oped another format for information, the cor porate trade exchange (CTX), which was com patible with the new standards in EDI and which has proven much more successful, its use growing, according to NACHA, by 153 percent in 1992.1 Undoubtedly, the difficulty 2 in finding a standard and the slow acceptance of the agreed-upon standard have slowed the acceptance of ACH for trade payments. n See Scott E. Knudson, Jack K. Walton II, and Florence Young, 1994 (see footnote 6 for complete reference), for a comprehensive overview of the issues raised in this section. 12See Bernell K. Stone, One to Get Ready: How to Prepare Your Company fo r EDI (CoreStates Bank, 1988), for more on the development of the CTX transaction. 22 DEVELOPMENTS IN POTENTIAL ALTERNATIVES TO ACH Check-processing technology has improved over time, with the use of lockboxes spreading and with progress in developing check trunca tion using digital imaging. Developments in electronic payment systems have occurred as well, including point-of-sale systems and cor porate credit cards. All of these potentially could compete with ACH. Lockboxes have become an important meth od for firms to collect payments. With a lockbox, a company directs its customers to send their payments to a post office box. The firm's bank then collects the mail, deposits the enclosed checks, and then sends the firm information about who paid. The information sent by the bank to the firm is sometimes sent electronical ly by having the customer enclose with pay ment a document that can be read with an optical scanner. When payment is received, the bank scans this document and then sends this machine-readable information to the firm. This form of collection quickens the availability of checks and, in some cases, takes advantage of electronic processing. Check truncation, which involves taking a digital image of the check at the bank of first deposit and, thereafter, simply sending the digital image electronically to the payer's bank, FEDERAL RESERVE BANK OF PHILADELPHIA The Automated Clearinghouse System: Moving Toward Electronic Payment reduces the transportation cost of a check. If the imaging technology becomes more developed, and if use of the system grows enough to allow the high capital costs of the system to be spread over a large number of items, the processing cost could be less than that of processing a paper check. And truncation would offer con sumers the benefits of familiarity and control that they enjoy with check-writing, although their checks would not be returned to them. However, the challenge of creating a successful system is similar to that of developing the system for corporate trade payments by ACH: agreeing on, developing, learning, and main taining technologies to convey payment and invoice information electronically; and ensur ing that such technologies are sufficiently easy to use that they will be adopted on a wide spread basis. Point-of-sale (POS) systems are becoming more popular in grocery stores and gas sta tions, places where both cash and checks are used, and for payments that are less repetitive than those primarily suited to ACH. The POS systems often settle their interbank balances by ACH, so to that extent they are complementary to ACH. Corporate credit cards that feature monthly bills providing detailed information (sometimes available in electronic form) that a business needs to monitor and account for the purchases it makes are now being offered by banks. Such a service eases many problems in making lowvalue purchases by reducing the time and effort it takes to process and verify invoices. The transactions feature of corporate credit cards is another competitor for payments that are less repetitive than those best suited for ACH. James McAndreivs CONCLUSION While the Automated Clearinghouse has not eliminated the check as a means of payment, specific uses of ACH have been successful. In particular, direct deposit of payrolls and the government's use of ACH have shown signifi cant growth. In consumer bill-paying and corporate trade payments, obstacles to greater use of ACH are gradually being overcome. The obstacles to more widespread use of the ACH include the difficulty of agreeing on, developing, learning, and using new ways of communication that can easily convey invoice information, and issues of consumer control of the timing of payment. These obstacles are being overcome in spe cific types of payments: utility payments re quire little additional information to be con veyed, are repetitive, and are an area of healthy growth in the use of ACH; corporate trade payments are also an area of growth, in part because of the adoption of the CTX transaction format, which is compatible with other, more commonly used forms of electronic communi cation. However, many corporate payment invoices are complex and require that a large amount of information accompany payment, and these payments, even if repetitive, are less likely to be made by ACH. New methods of payment are now being developed that have specific advantages over ACH for certain kinds of transactions. Thus it appears that the future will hold not just one type of payment method but many, including paper checks, lockboxes, check truncation, point-of-sale systems, and the Automated Clear inghouse. 23 FEDERAL RESERVE BANK OF PHILADELPHIA Business Review Ten Independence Mall, Philadelphia, PA 19106-1574