WORLD BANK TECHNICAL PAPER NUMBER 260 Payment Systems Principles, Practice, and Improvements • t David B. Humphrey *** HG 3881 .H853 1994 c.2 IIH 1111 HI J RECENT WORLD BANK TECHNICAL PAPERS No. 188 Silverman, Public Sector Decentralization: Economic Policy and Sector Investment Programs No. 189 Frederick, Balancing Water Demands with Supplies: The Role of Management in a World of Increasing Scarcity No. 190 Macklin, Agricultural Extension in India No. 191 Frederiksen, Water Resources Institutions: Some Principles and Practices No. 192 McMillan, Painter, and Scudder, Settlement and Development in the River Blindness Control Zone No. 193 Braatz, Conserving Biological Diversity: A Strategy for Protected Areas in the Asia-Pacific Region No. 194 Saint, Universities in Africa: Strategies for Stabilization and Revitalization No. 195 Ochs and Bishay, Drainage Guidelines No. 196 Mabogunje, Perspective on Urban Land and Land Management Policies in Sub-Saharan Africa No. 197 Zymelman, editor, Assessing Engineering Education in Sub-Saharan Africa No. 198 Teerink and Nakashima, Water Allocation, Rights, and Pricing: Examples from Japan and the United States No. 199 Hussi, Murphy, Lindberg, and Brenneman, The Development of Cooperatives and Other Rural Organizations: The Role of the World Bank No. 200 McMillan, Nana, and Savadogo, Settlement and Development in the River Blindness Control Zone: Case Study Burkina Faso No. 201 Van Tuijl, Improving Water Use in Agriculture: Experiences in the Middle East and North Africa No. 202 Vergara, The Materials Revolution: What Does It Mean for Developing Asia? No. 203 Cleaver, A Strategy to Develop Agriculture in Sub-Saharan Africa and a Focus for the World Bank No. 204 Barghouti, Cromwell, and Pritchard, editors, Agricultural Technologies for Market-Led Development Opportunities in the 1990s No. 205 Xie, Kiiffner, and Le Moigne, Using Water Efficiently: Technological Options No. 206 The World Bank/FAO/UNIDO/Industry Fertilizer Working Group, World and Regional Supply and Demand Balances for Nitrogen, Phosphate, and Potash, 1991/92-1997/98 No. 207 Narayan, Participatory Evaluation: Tools for Managing Change in Water and Sanitation No. 208 Bindlish and Evenson, Evaluation of the Performance of T&V Extension in Kenya No. 209 Keith, Property Tax: A Practical Manual for Anglophone Africa No. 210 Bradley and McNamara, editors, Living with Trees: Policies for Forestry Management in Zimbabwe No. 211 Wiebers, Integrated Pest Management and Pesticide Regulation in Developing Asia No. 212 Frederiksen, Berkoff, and Barber, Water Resources Management in Asia, Volume I: Main Report No. 213 Srivastava and Jaffee, Best Practices for Moving Seed Technology: New Approaches to Doing Business No. 214 Bonfiglioli, Agro-pastoralism in Chad as a Strategy for Survival: An Essay on the Relationship between Anthropology and Statistics No. 215 Umali, Irrigation-Induced Salinity: A Growing Problem for Development and the Environment No. 216 Carr, Improving Cash Crops in Africa: Factors Influencing the Productivity of Cotton, Coffee, and Tea Grown by Smallholders No. 217 Antholt, Getting Ready for the Twenty-First Century: Technical Change and Institutional Modernization in Agriculture No. 218 Mohan, editor, Bibliography of Publications: Technical Department, Africa Region, July 1987 to December 1992 No. 219 Cercone, Alcohol-Related Problems as an Obstacle to the Development of Human Capital: Issues and Policy Options No. 220 Kingsley, Ferguson, Bower, and Dice, Managing Urban Environmental Quality in Asia No. 221 Srivastava, Tamboli, English, Lal, and Stewart, Conserving Soil Moisture and Fertility in the Warm Seasonally Dry Tropics No. 222 Selvaratnam, Innovations in Higher Education: Singapore at the Competitive Edge No. 223 Piotrow, Treiman, Rimon, Yun, and Lozare, Strategies for Family Planning Promotion No. 224 Midgley, Urban Transport in Asia: An Operational Agenda for the 1990s (List continues on the inside back cover) WORLD BANK TECHNICAL PAPER NUMBER 260 Payment Systems Principles, Practice, and Improvements David B. Humphrey CI.U.U. .LlllJ:-'.L V V CllltllL.:, I *** HG3881 .H853 1994 c 2 • Humphrey, David B Payment systems : · principles, practice and improvements I • fOR PMEN1 R£CONS1W \C1\0N ~NO Ot'J£L0 Utt; '- I l'J'J::> SECTORAL UBRAR\ INTERNATIONAL BAN,< The World Bank Washington, D.C. Copyright© 1995 The International Bank for Reconstruction and Development/nm WORLD BANK 1818 H Street, N. W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing February 1995 Technical Papers are published to communicate the results of the Bank's work to the development com- munity with the least possible delay. The typescript of this paper therefore has not been prepared in accor- dance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibili- ty for errors. Some sources cited in this paper may be informal documents that are not readily available. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility whatso- ever for any consequence of their use. The boundaries, colors, denominations, and other information shown on any map in this volume do not imply on the part of the World Bank Group any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to the Office of the Publisher at the address shown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to copy por- tions for classroom use is granted through the Copyright Clearance Center, Inc., Suite 910, 222 Rosewood Drive, Danvers, Massachusetts 01923, U.S.A. The complete backlist of publications from the World Bank is shown in the annual Index of Publications, which contains an alphabetical title list (with full ordering information) and indexes of sub- jects, authors, and countries and regions. The latest edition is available free of charge from the Distribution Unit, Office of the Publisher, The World Bank, 1818 H Street, N. W., Washington, D.C. 20433, U.S.A., or from Publications, The World Bank, 66, avenue d'Iena, 75116 Paris, France. ISSN: 0253-7494 David B. Humphrey is a professor of finance and Smith Eminent Scholar in Banking in the College of Business, Florida State University. Library of Congress Cataloging-in-Publication Data Humphrey, David B. Payment systems : principles, practice, and improvements / David Burras Humphrey. p. cm. -(World Bank technical paper, ISSN 0253-7494; no. 260) Includes bibliographical references. ISBN 0-8213-3111-6 1. Banks and banking, International-Communication systems. 2. Clearinghouses (Banking)-Data processing. 3. Payment-Data processing. 4. Electronic funds transfers-International cooperation. I. Title. II. Series HG3881.H853 1994 332.1-dc20 94-40044 CIP CONTENTS Foreword vi Abstract vii Preface viii Introduction . . .. . .. .. .. ... . . .. . . .... . ... .. .. . . . . . . . .. .. .. .. . . . 1 Chapter 1: Payment Principles 3 What is a Payments System? 3 Clearing and Settlement Functions 4 Barter Has High Transactions Cost . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Commodity Money Facilitates Specialization and Trade .......... . 5 Currency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Checks and Electronic Payments . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Clearing and Settlement in the U.S . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Chapter 2: Examples Of Mature Payment Systems ..................... . 10 The U.S. Payments System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Payments Systems in Europe 12 Differences in Payments Use Across Countries ................ . 13 Payments in Newly Emerging Socialist Countries . . . . . . . . . . . . . . . . 17 Payments and the Money Supply . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 What Determines an Effective Payments Instrument? 19 Costs of Different Payment Instruments 20 Scale Economies in Payments Processing 22 Issues in Assessing Payment Systems 23 Chapter 3: Payment Practices: Small Value Payments 30 Small Versus Large Value Payment Instruments . . . . . . . . . . . . . . . . . 30 Cash . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Costs and Benefits of Using Cash for Payments . . . . . . . . . . . . . . . . . 33 Checks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 Specialized Checks: Certified Checks, Money Orders, Travelers Checks 40 GIRO . .. . .. . .. . . . . . ... .. ... . .. .. . . . .. . .. .. .. . .. .. . . 41 Credit Cards 43 Direct Debits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 Risk on Small Value Payment Networks . . .. . . .. .. . . .. .. .. . . . . 44 The Future of the Check in Developed Countries 45 Lessons Learned: How To Improve Small Value Payments Efficiency . . 46 iii Chapter 4: Payment Practices: Large Value Payments . . . . . . . . . . . . . . . . . . . . 48 Gross and Net Settlement Networks . . . . . . . . . . . . . . . . . . . . . . . . . 48 A Description of Large Value Wire Transfer Networks ............ 50 Different Types of Payment Finality . . . . . . . . . . . . . . . . . . . . . . . . . 52 Telex and Other Large Value Transfer Systems . . . . . . . . . . . . . . . . . 53 Overdraft Risk on Large Value Payment Networks . . . . . . . . . . . . . . . 55 Causes of the Daylight Overdraft Problem . . . . . . . . . . . . . . . . . . . . . 57 Factors Affecting the Distribution of Overdrafts Among Banks ...... 58 Current Risk Controls 58 New Risk Controls 60 Market Response to Overdraft Pricing . . . . . . . . . . . . . . . . . . . . . . . . 61 Payments Netting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 Lessons Learned: How to Reduce Risk For Large Value Payments 65 Chapter 5: Short-Run Improvements To A Payments System . . . . . . . . . . . . . . . 67 Short-Run Improvements for Small Value Payments . . . . . . . . . . . . . . 67 Cash . . . . • . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Checks 68 Short-Run Improvements for Large Value Payments 68 Chapter 6: Long-Run Improvements To A Payments System . . . . . . . ....... . 70 The Importance of User Need When Considering Major Changes in Payments ............................. . 70 Long-Run Improvements for Small Value Payments ............. . 70 The Replacement of Cash Payments ........................ . 70 Checks: Paper Versus Electronics ......................... . 71 Debit Transfer (Checks) Versus Credit Transfer (GIRO) ........... . 72 Long-Run Improvements for Large Value Payments ............. . 73 Funds Transfer Versus Message Transfer Systems .............. . 73 Gross Versus Net Settlement ............................. . 74 lntraday Credit Versus No Overdrafts ....................... . 75 Sender, Settlement, or Receiver Finality 76 Chapter 7: Rapid Economic Change And The Payment System 78 The Russian Payment System and the Transition to a Market Economy . . . . . . . . . . . . . . . . . . . . . 78 The Role of the Payment System in Russia . . . . . . . . . . . . . . . . . . . . 78 The Payment System Under State Socialism . . . . . . . . . . . . . . . . . . . 80 Retail Payments: Cash and Checks . . . . . . . . . . . . . . . . . . . . . . . . . 81 Wholesale Payments: Debit and Credit-Based Payment Orders . . . . . . 82 Settlement of Payments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 Electronic Payments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 Impediments to Payment System Development 83 iv Plans for Payment System Improvement . . . . . . . . . . . . . . . . . . . . . . 84 Encourage Check Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 Improve Payment Orders and Plan for Electronic Payments . . . . . . . . . 85 Develop a Same-Day, Large-Value Transfer System for Rubles . . . . . . . 85 Improvement Efforts By Private Clearing Organizations . . . . . . . . . . . . 86 Lessons Learned ........................... ........... 87 Effects of the Payment System on the Money Supply in Russia . . . . . . 88 Payment Order Credit Float and Reduced Liquidity . . . . . . . . . . . . . . . 89 Lower Balance Requirements and Expanded Liquidity from Real-Time Gross Settlement or Payments Netting . . . . . . . . . . . . . . . 90 Chapter 8: Harmonization of Large Value Payments Across Countries and a Benefit/Cost Analysis of Settlement Finality . . . . . . . . . . . . . . . . . . 92 Achieving Settlement Finality in the EEC . . . . . . . . . . . . . . . . . . . . . . 92 A Benefit/Cost Analysis of Settlement Finality . . . . . . . . . . . . . . . . . . 93 Ways to Reduce Collateral or Idle Balance Requirements . . . . . . . . . . . 97 A Method for Transforming a Net Settlement Network into 98 Chapter 9: Conclusions, and What to Look for in Restructuring a Payments System 100 Bibliography 103 Biographical Information about the Author . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 V FOREWORD Payment systems enable the financial sector to serve the needs of the real economy. Their improvement is a priority in the transformation of economies, developing banking systems and emerging money markets. The development of payment systems has major implications for the accounting and legal framework, the telecommunications infrastructure, the institutional capacity in the central bank and the commercial banks that act as major providers and users of the payment system, and for monetary instruments and management. It is likely that designing and installing modern payment systems will, in some cases, require fairly large investments in software, hardware, telecommunication system and human capital, including institution building. The World Bank has, therefore, become increasingly involved in financing and advising on the improvements and development of payment systems in a number of transitional ad emerging countries in cooperation with the IMF and participating central banks. Before any significant investment can take place, it is vital that "due diligent" vision, scope and design work has been done in planning the payment systems project. The way payment systems are designed and constructed can affect dramatically the development of money and capital markets and influence the conduct of monetary policy. This is somewhat analogous to developing road systems; the layout of roads and highways very much influences future traffic and the development of an economy. The reform of payment systems often requires changes in the central bank accounting structure and may lead to reorganization of the central bank itself. This is why a payment systems project is regarded as a policy matter, not a simple technology matter, in most mature market economies. The policy design of the national payment systems will inevitably differ from one country to another, reflecting specific economic, geographical and historical conditions. Nevertheless, one could identify a set of common policy and design questions or issues. This paper by Professor Humphrey was originally prepared to address those issues under the direction of Andrew Sheng and Setsuya Sato of the World Bank and was later presented at the Training Course on Payment Systems Issues organized under the initiative of the Financial Sector Development Department in March 1994. Given the quality of the paper and the debate which followed, we decided to publish this book to a wider audience. We hope this proves to be of interest both to those who specialize in payment systems issues, as well as those with a broader interest in financial matters during transition and development. Gary L. Perlin Director Financial Sector Development Department Finance and Private Sector Development The World Bank vi ABSTRACT This book is intended as an introduction to payment systems. An efficient and smoothly operating payment system is a necessary precondition for business development, both domestically and internationally. Our goal is to provide a relatively complete understanding of the benefits, costs, risks, and problems associated with mature payment systems, such as those now operating in the United States and Europe. This framework is then used to outline areas where the payment systems of less developed countries and newly emerging socialist countries (such as Russia) may be restructured and improved. The focus is on the economic incentives and institutional arrangements that determine the use of and problems with different types of payment instruments and how these incentives may be adjusted to improve the overall operation of a payment system. Both smaller-value consumer (retail) payments and larger-value business (wholesale) payments are covered and short-run and long-run suggestions for payment system improvement are noted. vii PREFACE This book is intended as an introduction to payment systems. An efficient and smoothly operating payment system is a necessary precondition for business development, both domestically and internationally. The goal is to provide a relatively complete understanding of the benefits, costs, risks, and problems associated with mature payment systems, such as those now operating in the United States and Europe. This framework is then used to outline areas where the payment systems of less developed countries and newly emerging socialist countries (such as Russia) may be restructured and made more efficient. The focus of this book is on the economic incentives and institutional arrangements that determine the use of and problems with different types of payment instruments. Importantly, adjustments to these incentives are discussed with a view toward improving the overall operation of a payment system. Smaller-value consumer (retail) payments and larger-value business (wholesale) payments are covered in some detail and both short-run and long-run suggestions for payment system improvement are noted. It is also shown that no single path has been followed in the development and use of payment instruments across countries. While most countries have access to the same set of noncash payment instruments, two distinct paths have emerged over time. The debit transfer (e.g. check) path using provisional payments has been most closely followed in the U.S., Canada, and India, while the credit transfer (e.g., GIRO) path using final payments has been emphasized most strongly in Germany, Sweden, and Switzerland. The benefits and costs associated with these two payment development paths are an important part of this book since many developing and newly emerging socialist nations are faced with the difficult choice of which path to emphasize when considering improvements to their payment system. viii INTRODUCTION The purpose of this book is to outline some basic principles associated with payment systems, to discuss in detail the payment practices which exist on mature payment systems in developed countries and, using this framework, to point out areas where the payment systems of many countries may be improved. The focus of this analysis is on the economic relationships and incentives associated with the use of different payment instruments and on how these relationships and incentives may be altered to improve the operation of a payment system. Operational details of specific payment arrangements are also discussed as these arrangements often determine the economic basis to favor one payment instrument over another. Our goal is to provide a relatively complete understanding of the benefits, costs, risks, and problems associated with mature payment systems so that less developed and newly emerging socialist countries may, in moving to restructure their own payment systems, have a firmer basis upon which to recommend improvements and to allow The basic principles regarding the co-evolution of payment systems, economic specialization, and the development of markets are presented in Chapter 1. Although there is no immutable law governing the development of payment systems as a country grows and becomes more developed, there is a great deal of similarity in how payment systems evolve so that the experience of one country often is a good guide to future developments in others. Toward this end, the payment systems and payment instruments of the U.S. and Europe are outlined in some detail in Chapter 2. In addition, we also attempt to explain why payment instrument use sometimes differs between countries, show how payments are related to the domestic money supply, note the private and social cost of different payment instruments, and finally discuss a number of issues that often arise in developing a more efficient or safer payments network. More detail on specific payment instruments is presented in Chapters 3 and 4 which, respectively, concentrate on small value and large value payment networks and note the problems associated with payments on these networks. In particular, it is concluded that check-based debit transfer small value payments are inherently inferior to GIRO-based credit transfers and that systemic risk on large value networks can be controlled at low cost through simple changes in current payment practices. Having built up a base of information regarding the structure, operation, and problems associated with payment systems, likely areas of both short-run and long-run improvements are outlined in Chapters 5 and 6. As might be expected, many short-run improvements to a payment system can be performed at low cost yet be quite effective, such as the use of dedicated couriers in place of the mails to deliver paper-based payment information on high volume routes. In the context of long-run improvements, greater use of electronics on a check-based debit transfer network can transform it into an essentially GIRO- -2- based credit transfer network over time. Thus the choice between checks versus GIRO payments need not be long lasting as the former can evolve into the latter, which is the preferred long-run result. To provide some perspective on many of the payment system issues and problems discussed earlier, Chapter 7 outlines the serious difficulties created for payment systems when a country is undergoing rapid and radical economic change. Although we focus on the Russian payment system and the problems it faces in adapting to a dramatically different economic structure, similar difficulties are created in other countries where economic structure is in transition from a command to a market economy. In Chapter 8, the move to harmonize risk controls, operating hours, and legal structures for large value payments in the 1 2 nation European Economic Community (EEC) is outlined. This harmonization will facilitate the economic and financial market integration of Europe and foreshadows a similar harmonization of the world's major payment networks in the future. The benefits from risk reduction are shown to be likely greater than the costs. Finally, some of the major points deserving of reemphasis are noted in Chapter 9, along with a brief description of what one should be looking for in restructuring a payment system. Chapter 1 : PAYMENT PRINCIPLES Payment systems are often taken for granted. However, an efficient payments system is important to ensure the smooth operation of markets. Markets are merely the "place" where transactions occur. Payments systems are important because they permit specialization to occur in production and help determine how efficiently transactions are made and settled. This, in turn can affect the level and rate of economic growth and the efficiency of financial markets (to the degree they exist). The goal of this and the following chapters is to describe some basic elements of payments systems. We will present some basic principles of payments systems and outline their evolution over time in this chapter. In Chapter 2, we shall describe (for background) the current composition of payment systems in the U.S. and Europe, try to explain why payment instrument use sometimes differs between countries, show how payments are related to the one su I , note the private and social cost of different payment instruments, and finally discuss a number of issues that often arise in deve oping a more e payments network. What is a Payments System? At its most basic level, a payments system is merely an agreed upon way to transfer value between buyers and sellers in a transaction. As illustrated in Figure 1-1, a payments system facilitates the exchange of goods or services in an economy. While there are some differences in payment systems among countries at the same stage of development, the greatest differences are among countries at different stages of development. Figure 1 Buyers and Sellers, Payors and Payees Transfer of goods or services Goods flow Value flow Transfer of value through a payment system -4 - At its simplest level, a payment system is distinguished by what "things" are used as money to transfer value in an economic exchange of goods or services. This ranges from using commodities for money, to using currency, to using checks, to finally using electronics to transfer value. In the developed world, currency, checks, and some electronics are used simultaneously. In the less developed world, commodities--which includes precious metals--and currency are dominant. Clearing and Settlement Functions. Importantly, the payment system functions of clearing and settlement occur regardless of the type of money being used to effect payment. When commodity money or currency is used in a transaction, the act of payment itself involves the exchange of value along with clearing and settlement. This is because commodity money or currency represents final payment. That is, when commodity money or currency are exchanged, this type of money is readily acceptable and can be immediately used in another transaction. The same is not always true when the money used in a transaction is a check. Legally, at least in the U.S., when a customer tenders a check in exchange for a good or a service, a transaction is deemed to have taken place. But the actual exchange of value or final payment is contingent upon being able to collect the check from the bank (savings and loan, credit union, or savings bank) that the check is drawn upon. The person or firm receiving a check (from the individual who wrote it) first deposits it at his bank. The bank now has to collect the check--this bank is called the collecting bank. The collecting bank sorts this check (using the paying bank's routing transit number printed on the bottom of the check) along with other checks deposited that day, provisionally credits the account of the person depositing the check, and then this check and others are presented for payment at the check writer's bank--the paying bank. Upon presentment of the check for payment, the paying bank has the right to inspect it for proper signature, date, etc. and then will look to see if the checkwriter has sufficient funds in his account to cover the check. If the funds are insufficient, the check is returned to the collecting bank and, later, to the depositor. About 1 % of all checks are returned for insufficient funds in the checkwriter's account. This depositing, sorting, presentment, inspection process is the clearing function. Once it is determined that there are sufficient funds in the checkwriter's account, the paying bank will transfer "good and final funds" to the collecting bank, who then removes the provisional hold placed on the deposited check so that the depositing customer can use the funds he initially deposited. The transfer of "good and final funds" between banks used to involve the physical transfer of gold or currency but today this involves a simple accounting operation with the U.S. central bank--the Federal Reserve--that transfers funds between different (noninterest earning) commercial bank reserve accounts held there. This transfer of "good and final funds" is the settlement function. -5 - In the U.S., the process of clearing and settlement of a check is typically done overnight so that the majority of the value of deposited checks is available to be used the next day--1 day funds. The actual number of days involved can take longer, however, depending on the time of day the check is deposited and the distance between the collecting and paying banks. Barter Has High Transactions Cost. How have payment systems evolved? As might be expected, payment systems differ most according to the stage of development of the economy. In a primitive society, transactions involve the direct physical exchange of goods or barter. Barter represents the payments system and certain specific goods--not standardized commodities--represent "money" or value. The more effort and/or skill it takes to produce something, the more it will trade for. This is not unlike Marx's labor theory of value where the "worth" of some particular good being produced is proportional to the number of labor hours it took to produce it. to occur, there has to be what is called a double coincidence of wants. That is, the "seller" must want what the "buyer" is willing to give up and vice versa so that both parties are satisfied in a transaction. For example, a "seller" in a primitive economy may want a canoe paddle but is only willing to give up a bushel of wild potatoes in exchange. To complete this desired transaction, the "seller" must find a "buyer" who wants to give up a canoe paddle and is willing to take a bushel of wild potatoes in exchange. In a primitive society which only has a limited set of items that are to be exchanged, barter is an adequate way to make these exchanges. Note that the key element here is the phrase "a limited set of items to be exchanged" which implies little specialization of labor or production in the economy. Commodity Money Facilitates Specialization and Trade. When there is more specialization of labor in an economy two things occur. First, the economy becomes more developed because greater specialization leads to greater productivity or output per worker. This raises incomes and permits more items to be consumed. Second, increased specialization leads to a greater need for trade. With specialization, each member of society will not produce all or even most of what it needs and wants so these other goods need to be acquired through trade. However, the search costs involved to fulfil the double coincidence of wants requirement for barter trade to take place becomes more difficult when the economy starts to specialize. This raises transaction costs. As a result, some form of money typically replaces barter as the dominant form of exchange. For something to be used as money, it has to be generally acceptable as payment in transactions, constitute a safe store of value, and represent a standardized unjt of account. Precious metals have in the past, and even today, served in this capacity as -6 - commodity money. But peppercorns, tobacco, rum, salt, and other standardized commodities have also served effectively as money. The great benefit of money in replacing barter is that one no longer has to satisfy the double coincidence of wants restriction for successful trade to take place. That is, all a seller has to do is to find someone who wants to buy a bushel of wild potatoes, not someone who wants to buy a bushel of wild potatoes and simultaneously wants to sell a canoe paddle as occurs with barter. The seller, after receiving money for his wild potatoes, can go to sumeone else and buy the canoe paddle he wanted originally. As a result, money can lower transactions costs and facilitate trade, which in turn permits greater specialization and greater productivity in production. Currency. Finally, we evolve to the situation we see today where currency-- which is money by fiat or law of the state--largely replaces money made of precious metals. Historically, most countries were once on a gold standard. That was a period of time where the face value of currency was required to be backed 100% by gold of equal value . Thus the domestic money supply was at that time always redeemable for gold and the money supply itself could only expand if gold reserves expanded. Needless to say, major gold discoveries in California and Alaska led to large increases in worldwide stocks of gold which in turn led to periods of worldwide price inflation. Today, most domestic currencies are not backed by gold, and the stock of currency is determined by how much a nation decides it will print. In the U.S., it only costs about $26 to print 1,000 pieces of currency regardless of the denomination. This difference in cost of production and legally enforced face value of currency is called the benefit of seigniorage. In some Latin American countries, seigniorage is used to finance government expenditures because of an inability to collect the taxes imposed. In effect, there are two ways to finance government expenditures. First, taxes can be used to transfer purchasing power to the government. This reduces the real value of personal consumption in an environment of relatively stable prices. Second, if taxes can not be used effectively, then additional currency can be printed to finance government expenditures. This leads to inflation which also reduces the real value of personal consumption but in an environment of rising prices. The problem is that rising prices tend to distort investment decisions and promote expenditures on goods and services which tend to hold their value rather than add to the productive capacity of the country to promote future economic growth. Checks and Electronic Payments. In the U.S., checks are also an important means of payment along with currency. A much more recent development has been the partial replacement of checks by so-called electronic payments. These take the form of automatic deposit of payroll, Social Security, or retirement benefits directly into a checking - 7 - account at a bank. For recurring monthly payments, it is also sometimes possible to have a checking account debited electronically for the monthly amount involved. These types of electronic payments are processed through an Automated Clearing House or ACH. As well, there are limited opportunities to electronically debit a checking account at the point of sale--a grocery store or a gas station--rather than write a check. Such an arrangement is called a Point Of Sale network or POS. However, the really important aspect of electronic payments concerns the networks that are used to transfer very large sums of money each day--the wire transfer networks. $2 trillion is transacted each day over these large value electronic networks in the U.S. These networks are quite important for the smooth operation of financial markets that operate in various world financial centers. Clearing and Settlement in the U.S. The relation between the different types of payment instruments--barter, currency, checks, electronic payments--and the clearing and s the U.S. a ment s stem over time are shown in Table 1-1. Similar sorts of relationships would exist for other developed countries as well. Table 1-1 Clearing and Settlement in the United States Payment . Tiine Period Instrument: .·. .Clearing \ . Settlement . . Early history Barter Not needed Payment is settlement Gold Currency Recent past Currency Local clearing Reciprocal balance and gold Checks houses shipments Present Currency Local/regional Reciprocal balance and national Checks clearing houses reserve accounts ACH,POS Wire transfers As noted above, when the payment instrument itself represents "good and final funds", as is the case with barter trade, commodity money, and fiat money or currency, then there is no need for clearing the payment between banks nor in having a central bank provide settlement. The payment itself is settlement. -8- Paper checks developed as a substitute for currency transactions and have played an important role in the U.S. payments system since 1865. Banks in the same local area would get together once a day to exchange checks drawn on one another. Checks were cleared, and "good and final funds" transferred, through a network of interbank reciprocal or correspondent balances. In order for one bank's checks to be honored when deposited at another institution, a correspondent balance was debited and credit given to the check depositor. Correspondent balances were established and maintained through the shipment of gold (and later currency) between banks. Shipping costs were basically proportional to the value of the currency or gold being shipped. When reciprocal balances were insufficient, credit risk was incurred by the bank cashing a check. Out of this process evolved a system of check-exchange charges that were proportional to the par or face value of the check being deposited. When exchange charges were deducted, checks were paid at less than par value (non-par banking). Efforts to minimize exchange charges often led to circuitous routing of checks, which impeded the flow of funds and commerce. Payments and financial transactions, reflecting the concentration of real economic activity, were restricted largely to regional money centers. Financial transactions among these regional centers were subject to domestic exchange rates. That is, it may have cost someone in Chicago $1.05 to purchase something which cost $1.00 in New York. The existence of domestic exchange rates reflected the cost and risk of clearing and settling a funds transfer between money centers. No centralized settlement facility existed at that time and gold or currency was the settlement medium used. The establishment of the Federal Reserve in 1913 altered the U.S. payments system in two important respects. First, Federal Reserve member banks could transfer funds by wire using an essentially single, centralized reserve-account balance. Thus banks had less need to maintain a complex network of interbank reciprocal balances to clear and settle checks. Although reciprocal balances are still used today to clear checks of smaller banks, book-entry accounting using wire transfers between reserve accounts has all but eliminated the need to ship gold or currency between regional banks to settle check-expenditure flows. Because of the existence of a centralized settlement facility, domestic exchange rates ultimately disappeared, although other transactions costs remained. The second change in the payments system was the virtual eliminatic,n of non- par banking. This elimination was accomplished by a requirement, written into the Federal Reserve Act, that all checks cleared through the Federal Reserve be settled at par value. In sum, Federal Reserve participation in the payments system increased payments efficiency (1) by providing a centralized mechanism for settling check and other transactions, and (2) by -9 - eliminating the incentive for circuitous routing of checks that had developed under non-par banking. The historical evolution of payments in the U.S. is instructive because the U.S experience paralleled payment system developments in other countries. While there is no immutable law governing the development of payment systems, as a country grows and becomes more developed there appears to be a great deal of similarity in how payment systems evolve. As such, the experience of one country often is a good guide to future developments in others. Hence a detailed knowledge of how mature payment systems are structured and operate in developed countries is essential before one can reliability pinpoint areas where less mature payment systems may be improved in other countries. Toward this end, the payment systems and payment instruments of the U.S. and Europe are outlined in greater detail in the following chapter. Chapter 2: EXAMPLES OF MATURE PAYMENT SYSTEMS In this chapter, we shall provide a more detailed summary of the current composition of the mature payment systems in the U.S. and Europe, from which useful comparisons can be made to other countries. We shall also try to explain why payments systems sometimes differ between countries and illustrate how they are tied to the domestic money supply. The estimated private and social costs of different payment instruments are also noted, although not all of these costs are subject to measurement and quantification. Finally, we discuss a number of issues that often arise in developing a more efficient or safer payments network. The U.S. Payments System. The structure of payments in the U.S.--by number of transactions, by total value transferred, and by average value--are shown in Table 2-1 . In this comparison, cash has been deleted. No one has an accurate idea of the use of cash in U.S. domestic payments. This is because anywhere from one-third to one-half of the value of all dollars outstanding are thought to be held (and used) outside the country. Table 2-1 United States Payment Structure (Percent composition and average dollar amounts) Transaction Transaction Average Payment Instrument Volume Value Value Check 80% 13% $ 1,150 Credit card 16 0 45 Credit transfer or GIRO 2 1 2,300 Direct debit (ACH, POS) 2 1 5,000 Wire transfer 0 85 4,100,000 Source: Computed from Bank for International Settlements, December 1993. Some information on the nature of cash transactions is, however, available. As a first approximation it has been estimated that cash accounts for over 80% of all transactions. But, since the average value of a cash transaction is small, perhaps only - 11 - averaging $5 per transaction, its importance in terms of the total value of all expenditures is also going to be small, probably less than 1 %. This refers to all types of cash transactions-- business as well as personal. Special surveys tell us that cash accounts for 36 % of the value of U.S. family expenditures. Thus cash is more important for personal and consumer transactions (which are small in value) but less so for business transactions (which are for much larger values) . The same can be said for most other developed countries. Returning to Table 2-1, where different types of paper and electronic payments are being compared, at least four things stand out. First, the noncash component of the U.S. payment system is seen to be overwhelmingly paper based. Checks account for fully 80% of noncash transactions volume while electronic methods--represented by credit cards, ACH, POS, and wire transfers--accounts for 20%. A second point evident in the table is that although electronic payments account for a larger proportion of the value of all noncash payments, the results are dominated by the ue electronic a ment networks. These are Fedwire, run by the U.S. central bank (the Federal Reserve), and the privately owned Clearing House lnterban Payment System (or CHIPS), run by a group of large New York banks. Payments on these networks are tied to business and financial market transaction needs and average over $4 million per transfer. As a total, they account for 85% of the value of all U.S. transactions. A third point to be seen from the table is that virtually all noncash U.S. transaction volume is from a debit transfer--not a credit transfer. In a debit transfer system, payment is legally deemed to have been made prior to the receiver obtaining "good and final funds". Merchants who accept checks as payment have to present them to the bank they are drawn upon to obtain final payment. As noted earlier, if the money is not in the customer's account, the check is returned to the merchant who then has to try to get the customer to pay him again. Although this problem affects only about 1 % of the 60 billion checks written in the U.S. each year, returned checks represent some 600 million payments and are costly to process. Even for large value transactions over wire transfer networks, payments are made and accepted even though the funds in these accounts may be insufficient to cover them. These deficiencies are made up by the end of the business day and are thus called "daylight overdrafts". This is opposed to overnight extensions of credit which carry a stiff penalty and are thus infrequent. On Fedwire , daylight overdrafts were, free prior to 1994 but are now priced. In terms of size, daylight extensions of credit are around 8% of the total value of all daily payments, or around $150 billion of the $2 trillion dollars transferred each day. - 12 - Obviously, this is a serious problem and one which has implications for monetary policy and control due to the large unintended credit extensions involved. This payment risk problem will be discussed in more detail in Chapter 4. While checks and even wire transfers represent debit transfers, many ACH and POS payments represent credit transfers. With a credit transfer, funds are deemed to be in a customer's account before the payment is made. Consumer credit transfers in the U.S. are quite small, less than 2% of all noncash transactions. This is just the opposite of Europe. Payments Systems in Europe. The structure of payments in Sweden is reasonably typical of payments in most European countries and so this country's payment structure is displayed in Table 2-2 . The essential difference between Europe and the U.S. lies in the increased use of credit transfer payment instruments and a reduced reliance on checks, which are debit transfer instruments. Only an average of 18 checks per person are written each year in Sweden while the per capita figure for the U.S. is over 200 per year. Table 2-2 European Payment Structure (Percent composition and average dollar amounts) Transaction Transaction Average Payment Instrument .· Volume · . Value . Value Check 9% 2% $ 1,200 Credit card 9 0 200 Credit transfer or GIRO 77 19 1,100 Direct debit (ACH) 5 1 520 Wire transfer 0 78 37,500 Source: Computed from Bank for International Settlements, December 1993. In the U.S., debit transfer checks accounted for 80% of all noncash transaction volume (Table 2-1). In Sweden, checks only account for 9% of noncash transactions while credit transfers--either paper or electronic GIROS--account for 77 % of all transactions. This distinction is important since it is generally agreed that a credit transfer system is less risky than a debit transfer system. For one thing, there are no return items, as there are with checks. For another thing, there are no account overdrafts--either daylight or overnight--as - 13 - there are with wire transfers in the U.S. A second difference between the U.S. and Europe is the dominance of paper- based noncash transactions in the U.S. while in Europe the structure is reversed with a slight majority of noncash payments being made electronically. A reason for this difference is offered below. Overall, the main differences between the mature payment systems of the U.S. and Europe are: (1) more credit transfers in Europe {reducing risk); (2) more electronic transfers in Europe (reducing cost); and (3), less reliance on credit cards and lower values for wire transfers. Differences in Payments Use Across Countries. There are some differences across countries in their use of different payment instruments for transactions. These differences are illustrated, in simple terms, in Table 2-3. Cash is relatively heavily used in personal transactions in Japan, Switzerland, and in the newly emerging socialist countries of the Commonwealth of Independent States {CIS). The use of cash is probably most directly tied to its convenience, low usage cost, and widespread acceptability as final payment for relatively low value transactions. After this, a desire for privacy in payments information and the relative level of national crime rates would seem to be important influences on cash usage. Table 2-3 Differences in Payment Use Across Countries . . . . Reliancedn /•• cdbnfly +• <. • Cash Japan low crime Switzerland Privacy CIS legal structure Checks United States, Canada, Disaggregated banks served United Kingdom general public GIRO Sweden, Germany, France, Centralized post offices served Europe general public Wire transfers United States, Europe, Japan Centers of trade and finance These latter two influences in particular would seem to be important reasons for the high rates of cash usage observed in Switzerland and Japan but relatively low rates of cash usage inside the U.S. In addition, in Sweden, a fee on small value check and credit card transactions also induces users to use cash for retail transactions. In the case of the newly emerging socialist countries, cash is heavily used because of the lack of viable - 14 - alternatives, exacerbated by a legal structure which does not well-support the efficient and confident use of alternatives to cash in personal transactions. In the U.S. and Canada, checks are relatively heavily used compared to other countries. It is felt that this reliance on checks resulted from a relatively disaggregated banking system in the U.S. and the fact that banks have historically served the general public. This reasoning also helps to explain why Europe relies on credit transfers rather than debit transfers, a contrast which is next described. But first, it is instructive to investigate how cash has substituted with checks in the U.S. with the development of Automated Teller Machines or A TMs. Recently, when it became easier to obtain cash from bank demand or savings accounts in the U.S., idle per capita cash balances in the hands of the public (excluding illegal activities) apparently fell--that is less cash was being used. This occurred with the establishment of ATMs, which are essentially just cash dispensers. Close to 80% of ATM transactions are used to obtain cash; the remaining transactions comprise customer deposits and account transfers. The bank per transaction cost of supplying cash from an ATM is about one-half that of using a human teller in a bank branch office. But, since ATMs are more convenient-- they are available 24 hours a day, including holidays--they are now used about twice as often to obtain cash as was a human teller. Thus the net benefit to banks of a substitution of A TMs for branch offices, at least for cash dispensing, has been close to zero. The lower cost per transaction at an A TM has been largely offset by an increased number of cash withdrawal transactions. While consumers have benefitted--through increased convenience and lower holdings of idle cash balances--banks have experienced little decrease in overall costs through this new development in service delivery technology. About the only way A TMs benefit banks is through revenues raised from network interchange fees assessed when the customer of one bank uses another bank's ATM to obtain cash. The expected substitution of ATMs for bank offices has occurred but the longer-term hope is that banks can deliver a much wider range of services, not just essentially cash dispensing. Credit transfers are dominant in Europe and are called GIRO payments. (The word GIRO is derived from the Greek and reflects the flow of funds around a ring, making a circle.) GIROs typically are operated by a national Post Office rather than banks. Retail electronic payments, in the form of Automated Clearing House (ACH) payments, debit cards, or GIROs have long been touted as cost effective substitutes for checks but the expected substitution of one for the other has been unexpectedly slow in the U.S. - 15 - European countries have been much more successful in implementing electronic payments than has been the U.S., probably for two reasons: (1) by having a system of nationwide postal and banking offices and (2) because of more appropriate pricing to users. But perhaps the main reason why the substitution of electronic payments for checks has been slow for the U.S. is historical. The U.S. has a very disaggregated banking system. In most other developed countries, the top ten banks would account for over 90% of deposits or assets of the banking system. In the U.S., it takes around 3,000 banks to account for 90% of deposits or assets. Citibank, for example, which is the largest U.S. bank, only accounts for 2% of the total national banking market. Although U.S. banks have historically served as a primary source for cash safekeeping and payment services for the general public, banking services are still local or regional, rather than national in scope. While a truly national bank with branches across state lines existed over 150 years ago, it was a monopoly. In the early 1800s, the Second National Bank of the United States controlled one-third of all U.S. deposits. However, the public feared the concentration ead for more com etition along with restrictions on branching across state lines. In this respect, regionalized banking in the U.S. is very similar to the current banking structure in Europe where domestic banks compete with each other within each nation but cross-border competition from banks in other countries is relatively small. Such a structure will likely change in a united Europe, just as nationwide banking will, when it comes, induce significant changes in competitive structure in the U.S. Historically, in many European countries, the postal service--not banks--provided necessary cash safekeeping services to the general public. It was then a relatively small step for this single, nationwide institution to offer regional and nationwide bill payment services for their savings account holders. This evolved into a GIRO. The U.S. does not have nationwide banking and, as a consequence, it has not developed a significant GIRO network. But there are other reasons--dealing with bank and user incentives--that help explain why the expected shift from paper to electronic payments has been slow in the U.S. First, the only direct providers of paper or electronic payment services are banks or other depository financial institutions. They are the only providers because they are the only institutions who have direct access to the central bank's final payment and settlement service which is needed to accommodate a debit transfer system. Because depository financial institutions are the only suppliers of check and electronic payment services, every time they shift a customer from using a check to using an electronic payment, they also succeed in shifting transaction business from one part of the bank to another part. On balance, a bank's transactions volume does not increase since electronic payments merely cannibalize check volume. Such a redistribution of payments - 16 - business within a bank would still be useful if strong scale economies existed for electronic payments. While there is some evidence of such scale economies, banks have not priced their check and electronic payments products in a manner to make these cost differences evident to their customers. The cost of providing check services are typically covered by requmng a minimum account balance to be held in a transaction account rather than only assessing a direct fee per check written. This amount typically varies depending on whether or not interest is also paid on the account. But even if all bank costs were reflected in direct fees, these fees would not account for the float benefits obtained by check users and not given to users of an electronic point of sale (POS) system where customer accounts are debited immediately. Although relative bank costs may favor electronic payments, the relative prices seen by customers would still favor checks because of the actual or perceived float benefits. As will be illustrated below, there is an inequality between the relative social and private costs of check and electronic payments which serves to mute the incentive of users to switch from paper to electronics. This is especially true for businesses, who write 40% of all checks in the U.S. but receive 90% of all float benefits. Even though corporations benefit the most from check float, they are also interested in electronic payments. While this interest stems partly from the fact that electronic payments are cheaper to process than checks, corporations see even greater savings from having all accounting, invoice, and payment information in electronic form. Currently, a dual media system is run where internal accounting information is electronic while the actual invoice, billing, and payment process is paper based. This is an expensive and error-prone arrangement which, if eliminated, can reduce costs. Such cost reductions would be real, as opposed to the zero-sum game now being played to obtain float benefits from check payments. But, since one firm's float gain is another firm's float loss, any one firm is typically unwilling to give up its own float benefit by unilaterally sending electronic payments while continuing to receive checks. In fact, those business electronic invoice and payment networks that have been discussed would be structured to retain the current distribution of float benefits. This is achieved by delaying the date of when electronic payments are effective to mirror those delays which currently exist in the check system. While this "solves" the float problem associated with electronic payments, it does nothing to resolve the difficulty in getting a truly large number of firms to agree on a standard format for electronic payment and invoice information. Finally, we come to wire transfers. Wire transfers represent very large value - 17 - payments, where safety, security, and speed are quite important attributes. These payment networks are concentrated in various world centers of trade and finance for this is the group of users typically served by these large value networks. Payments in Newly Emerging Socialist Countries. The newly formed countries of the former Soviet Union and other countries attempting to move from a planned/command system to a market economy currently have rather rudimentary payment systems. First, almost alt personal transactions take the form of cash. Checks are only rarely used and GIROs, credit cards, or electronic payments not at all. Second, inter-enterprise transactions, or payments between firms, whether large value or of ordinary size, are almost entirety paper- based credit transfers called payment orders. 1 These transactions are very cumbersome and quite time-consuming, often taking weeks to complete. Third, the politically determined geographic dispersion of industry in different regions of the former Soviet Union has resulted in the new republics engaging more heavily in inter-republic trade than would be the case if trade patterns were originally determined by market forces. Accordingly, inter-republic payments are a large element of the current system for inter-enterprise large-value payments. A market economy requires much more from a payment system than does a planned or command economy. In a command economy such as that of the former Soviet Union, the payment system was used primarily as a way of accounting for the movement of goods between enterprises. While funds were transferred with a paper-based payment order, there was no real need to make this process a speedy one. The float created had little economic value since all payments were essentially from one part of the government to the other and the "funds" necessary to make these payments could be created at will. Because inter-enterprise payments were essentially intra-governmental payments, there was also no need to carefully determine counterparty (other enterprise) risk to a payment order transaction even if it was a debit instrument (like a check). As these nations move toward a market economy with private ownership of firms and/or less governmental support of public enterprises, two things are happening. First, payment system float has taken on economic value because enterprises will now "own" the funds they spend, rather than having their revenues allocated to them in the government budget each year. Float absorbs limited working capital and thus becomes a business cost, which impedes inter-enterprise trade. This impediment can be reduced if the clearing and settlement process is speeded up. Typically, payment documents are mailed to each party in an inter-enterprise transaction chain. The simple substitution of courier delivery in place of the mail can substantially speed up the clearing process and reduce float while preserving l' Until mid-1992, paper-based debit transfers, called payment demand orders, were the most widely used but have been quite restricted since that time, with reliance on credit-based payment orders. - 18 - the current paper trail for documentation, security, and control. Second, with payments now going between relatively independent enterprises rather than between different parts of the government, credit risk becomes important. Unlike the government, independent enterprises may default on their financial obligations. Thus the assessment of counterparty creditworthiness becomes a new and necessary task to perform in the transaction process. And, since counterparty credit risk is greater the longer it takes to clear and settle inter-enterprise transactions, float reduction can reduce counterparty risk in addition to saving working capital. In the longer-run, payments efficiency in newly emerging socialist countries can be improved by adopting and implementing those aspects of mature payment systems in the U.S., Europe, Japan, or Canada that would be most cost-effective and consistent with the developing legal and institutional structure of these countries. For example, checks or paper- based GIRO payments can be introduced for personal transactions. For large-value transactions, speedier telex transfers for inter-enterprise transactions may be substituted for slower paper-based payment orders and a centralized wire transfer network using central bank "reserve balances" could be used for interbank transfers and settlement. Such improvements will not be easy to implement, however, because often the necessary underlying legal and institutional structure does not yet exist. Payments and the Money Supply. One aspect of a payments system only touched upon above is the relationship of the payment system to the money supply, and hence monetary policy. Payments are a flow of value that takes place within an economy that transfers value from buyers to sellers in a transaction. But this flow needs to come from someplace. Where it comes from is from the supply of money in an economy. The money supply sits around idle until part of it is used for a payment in a transaction. If daily payments are a large percentage of the stock of money that exists, then the "turnover" of money or its velocity is said to be high. The relationship between the money supply and economic activity is expressed in the well-known relationship of MV = PT or the money supply (Ml times its velocity (V) is equal to the price level (Pl times the number of transactions (T). PT represents some aggregate level of economic activity, such as GNP. The efficiency of the payments system is reflected in the money turnover rate, which indicates how many times the money supply has to be reused in order to meet the transaction and payment demands associated with a given level of aggregate economic activity. As a result, if the efficiency of the payment system improves, payments will take a shorter time to be cleared and settled before the funds being transferred can be reused to finance another payment. Consequently, improvements in the efficiency of the payments system would permit a country to reduce its domestic - 19 - money supply, assuming that economic growth and other things remain constant. The principle un~erlying the relationship between payments system efficiency and the money supply is shown by the equality MV = PT. Holding PT (or GNP) constant, an increase in payments system efficiency raises velocity (V) allowing the money supply (Ml to be reduced to support the same level of economic activity. Another way to illustrate the relationship between payments and the money supply is to show the payment instruments associated with the various components of the money supply in the U.S. This is presented in Table 2-4. The narrowest concept of the money supply is called M1 and comprises the value of the stock of coin, currency, and transaction or demand deposits. Obviously when cash is used in a transaction, it comes from the stock of outstanding currency and coin. Similarly, when checks, ACH, GIRO, and wire transfer payments are made, a transaction or demand deposit is being debited or credited. Table 2-4 ~ments and the Money Supply Cash Currency outstanding (M1) Check, ACH, POS, GIRO, and wire transfer Demand deposits (M1) (No associated payment instrument) Savings deposits (M2) CDs (M3) Treasury bills (L) Broader definitions of the money supply incorporate the value of stocks of other deposits or assets. These broader definitions--called M2, M3, and L--recognize that these other deposit and asset stocks can be liquidated and transferred to transaction accounts to also be used in making payments. Indeed, virtually any asset, given it can be easily liquidated, can be thought of as potentially contributing to the domestic money supply. In practice, however, only the most liquid assets perform this function very often and so only they are commonly adopted as defining the domestic money supply. What Determines an Effective Payments Instrument? The most important characteristic of a payments instrument is wide acceptability in transferring value in trade and exchange. Acceptability is determined, at least in part, by the total cost of using a particular instrument in a transaction. This can involve the direct costs of production, clearing, and - 20 - settlement as well as the indirect costs related to convenience and risk of loss through theft or fraud. An appropriate legal structure is also needed to ensure wide acceptability of noncash payment instruments. In this context, the rights of payors and payees must be clear and enforceable so that all parties to a transaction have an appreciation of the risks involved in using different payment instruments. For example, in the U.S. the Uniform Commercial Code, along with regulations issued by bank regulators and agreements among banks in local check clearing houses, all govern the clearing and settlement process for checks. They stipulate the size of a checks, what information should be on it, where this information is to be located, who has the responsibility to inspect and verify this information prior to payment, within what time period payment should be made, and who is responsible for loss of a check and/or fraud. Without these "protections", it is very unlikely that checks would be a very popular instrument in the U.S. In fact, one of the reasons why electronic payments have not been very popular in the U.S. has been the relative lack of a clear legal structure governing these types of payments. It was not that long ago, for example, that the tax collecting agency (the Internal Revenue Service) would only accept as proof of payment the original canceled check that was involved in a transaction. Now they will accept a bank supplied copy of the microfilmed check, permitting a greater acceptability of check truncation efforts, and will accept bank statements showing electronic debits using ACH or POS transfers, which has helped make these electronic payments more acceptable to users. Transactions costs are also important for the acceptability of a payments instrument. Transactions costs are composed of direct production costs plus implicit costs, such as float. From a social standpoint, the cost of various payment instruments should be priced so that the user of the payment instrument--the decision maker--faces the full cost involved. But in a competitive market framework, about the best we can expect is that the costs of the various payment instruments will reflect the full costs of their production, which will not always cover the full social costs involved. Because of the failure of the market to price each payments instrument in terms of both the full private (production) and social (float) costs of use, the decision process of choosing the most efficient or lowest cost instrument for a given transaction will be flawed. Costs of Diffsrsnt Payment Instruments. In general, the total user cost of different payment instruments is the sum of its direct production/processing cost plus its indirect float costs. Estimated U.S. payment instrument costs per transaction are shown in Table 2-5. This relationship is easiest to explain by taking specific examples, starting with the per transaction cost of using cash. It has been estimated that the direct per transaction average cost of using cash is 4 cents per transaction . This includes the printing/minting cost of currency and coin, plus the theft, safekeeping, and processing costs incurred by - 21 - establishments that accept cash payments. It also includes the costs incurred by financial institutions that take cash deposits by retailers and the costs of retiring and replacing currency in circulation once it has been worn out through excessive use. Table 2-5 User Cost of Different Payment Instruments (United States, cost per transaction) Production and Payment Instrument Processing Cost Float Cost User Cost Cash $.04 $.05 $.09 Check .79 -.83 -.04 ACH, POS, GIRO .29 .00 .29 Wire transfer 7.33 -.02 7.31 Source: Humphrey and Berger, 1990, Table 2-1. But there are indirect costs of using cash--costs that one would not think about unless cash was used for a very large value payment. This is the opportunity cost of holding idle cash balances prior to the time cash in used in a transaction. This opportunity cost is the lost opportunity of earning interest on cash holdings had these funds been instead deposited at a bank. It was estimated (using current interest rates) that it costs cash users about 5 cents in lost interest to use cash in a transaction. Accordingly, the total cost of using cash is, on average, 9 cents per transaction. Note that here the indirect cost of using cash is positive (reflecting lost interest income). Just the opposite arrangement holds for checks. Here the production/processing costs of 79 cents per transaction is more than offset, on average, by a negative 83 cent indirect cost--the cost of check float. Check float exists because it takes time--one, two, or three days, depending on if the payment is mailed or presented in person --before a customer's account is actually debited after a check has been accepted for payment in a transaction. In Canada, the banking system charges the (business) check writer for the opportunity cost of the time it takes to clear a check and obtain "good and final funds". As - 22 - a result, check float is minor and, importantly, the decision maker, in choosing to use a check, bears this cost directly. In the U.S., however, the receiver of a check bears this cost, effectively giving the check writer a small discount on the value of his purchase or bill payment. As a result of float, the check writer sees a small negative total cost of minus 4 cents for each check written. Here the float benefit more than offsets the real resource production cost of a check. The existence of check float can distort the tradeoff between check and electronic payments on the demand side. This is important because, as seen here, the production/processing cost of a consumer electronic payment--ACH, POS, or GIRO--is less than the cost of a check. But because check float is borne by the receiver of a check rather than the user who chooses to use a check, the user cost favors check use even though in terms of total real resources electronic payments are cheaper to produce. Relative prices should affect relative use of the different payment instruments. As one might expect, the relationship between payment use and social cost is weaker than that between use and user cost. The Spearman rank correlation coefficient between actual use of 9 different U.S. payment instruments--a broader selection than that shown in Table 2- 5--and the estimated social cost of these 9 instruments is .22. The Spearman value between use and user cost is larger at .65. This illustrates the distorting effects of float in the U.S. payments system where there is a divergence between social and private costs. In order of importance, this holds for checks, credit cards (not shown), and cash. While GIRO payments are shown here to have zero float costs, because there are no GIRO payments in the U.S., in actual fact GIRO payments in Europe cost the payor one day's float. But this is the only cost; the GIRO system in Europe typically absorbs all production and processing costs so the user only incurs the loss of 1 day's interest on the value of the payment being made. The costs shown for wire transfers, at over $7 per transaction, appear to be quite high but, relative to the value of the payment being transferred--which averages $4 million--it is exceedingly low. Wire transfers are typically made on a same-day basis so float is almost nonexistent, which is a good thing for otherwise a wire transfer would have an extremely large float cost for the user. Scale Economies in Payments Processing. While Table 2-5 shows the estimated cost of different payment instruments based on their current level of use in the U.S., it says nothing about how these costs may change depending on the amount of payment instruments processed. Put differently, what are the scale economies associated with some of these payment instruments? Virtually the only payment scale economy information available refers to check and electronic payment processing by the U.S. central bank. Commercial bank processing cost information is private and thus has been unavailable to researchers. - 23 - Fortunately, the technology of processing payments is essentially the same for both central bank and commercial bank suppliers of these services. The Federal Reserve is a large supplier in the U.S. payments market--it processes about one-third of all paper checks, two-thirds of all electronic wire transfers, and almost all of ACH electronic transfers. As such, scale economy results for Federal Reserve payments processing operations will likely provide a good approximation of the cost economies which may be realized by banks and other depository institutions in their payment operations. The results of empirical studies suggest that most of the high volume processors in the industry face essentially constant average costs in the processing of paper checks. The exception would be for relatively small processing volumes where scale economies can exist. While large scale economies have been estimated for ACH electronic processing, this result is likely due to the fact that ACH volumes are currently so small--only around 2 % of that of checks in the U.S. Thus the large scale effects that have been estimated for these electronic payments will be used up and the cost curve will flatten out as volumes grow and the market matures. Finally, slight scale economies have been estimated for wire transfer operations in the U.S. and for SWIFT (Society for Worldwide Interbank Financial Telecommunication) operations in Europe. Strictly speaking, SWIFT is a message transfer service, not a funds transfer or payment network. SWIFT messages tell banks to shift funds between already funded customer accounts at the same bank--thus they do not transfer funds between customer accounts at different banks. Interbank transfers are made almost entirely through Fedwire and CHIPS. Issues In Assessing Payment Systems. Having discussed in moderate detail the benefits and costs of payment systems and outlined the current structure of two mature systems, we now briefly discuss seven important characteristics one should be aware of in assessing--and perhaps improving--the efficiency and safety of a payment system. These seven characteristics range from deciding on whether or not the most appropriate payment system should be based on a debit or credit transfer arrangement, to determining who gains and/or loses if the current payment arrangements are altered. This last element can be the most important one in terms of being able to effectively change or improve a payments system. We now briefly discuss each one of these seven points. 1 . The choice between a debit versus a credit transfer payment system is essentially one between checks versus electronic payments. While the U.S. and other nations' payments systems have evolved from currency to checks to electronic payments, the technology exists today to bypass the check stage entirely. In general, a credit transfer - 24 - system, either electronic or paper-based, is by far the most efficient and least risky type of payment system to have. It is efficient because electronic transfers can be cheaper to process than paper checks and because the production costs of a credit transfer system do not involve return items, which are expensive to process. A credit transfer system is less risky because a payment can not be made unless there is money is in an account to be transferred. Thus the risk of loss or fraud is much lower in a credit based system. The operation of debit (check) and credit (GIRO) transfer systems are illustrated schematically in Figures 2-1 and 2-2. While debit transfers are almost always paper-based in the U.S. and elsewhere, small numbers of electronic debit transfers do occur over ACH networks. Similarly, while most GIRO payments in Europe are electronic, paper-based GIRO payments do occur as well. As shown by the arrows in Figure 2-1, a debit transfer involves going from the payor who writes a check, to the payee who receives the check, to the payee's bank where the check is deposited, to the payor's bank who finally pays the check if there are sufficient funds in the payor's account. If the funds in the payor's account are insufficient, then the check is returned by reversing the path it took to be collected and presented for payment. This process illustrates both the risk to the payee in accepting a check as payment and the extra cost involved if the check has to be returned. Figure 2-1 Illustration of a Debit Transfer Check: A debit transfer 3. Check is collected PAYOR'S - and settled PAYEE'S BANK --- BANK 4. Payor's account u ,, is debited 2. Check is deposited PAYOR - - PAYEE 1. Check is written Solid line: Route of forward collection, points 1 to 4. Route of returned items, point 4 to point 1. - 25 - A credit transfer follows a different route. As seen in Figure 2-2, the payor instructs his GIRO to transfer funds from his account to the payee's account. If the funds in the payer's account are insufficient, the transfer never takes place as the instruction is rejected. Thus return items do not exist within a GIRO framework. As a result, when funds are received in the payee's account, they are good funds. Importantly, the entire transaction occurs within a single organizational structure--the GIRO--so the funds transfer and settlement process is both simpler and less expensive than with a debit transfer system where interbank payments occur along with payments between payor and payee. Figure2·2 Illustration of a Credit Transfer GIRO: Acredit transfer .,.. ::;:: . .... .. •. . . >. PAYOR'. ·· .. . AND . ·:·"""j,Avelifsr: .· · " - . . ~---~IROcredits ~e's eccount ....,...,........,........,........,........,........,., PAYOR PAYee·· Solid line: Route of forward collection points 1 to 2. There are no return items. Given the extra cost and risk associated with a paper-based check system, some observers have suggested that the various newly emerging socialist countries--which are now heavily reliant on cash for all but the largest value transactions--could skip over checks entirely and move directly from a currency based system to a system of electronic payments. If this occurs, it is likely that an electronic system based on credit transfers--similar to the European GIRO networks--would be the best model to follow. 2. Regardless of the type of payment system chosen, it is important that prices should cover all the costs. Most of these costs will be incurred in the clearing process, which involves the handling, sorting, and physical or electronic delivery of the payment instructions and/or the instrument itself. The settlement process is also important but is relatively low cost. - 26 - If float is involved, as with the use of checks, then procedures should be developed to charge the check .!::!§.fil for this cost, as has been done in Canada. Only when all the prices for the various payment instruments cover their full costs will the payment instrument user be able to make the best choice, and thus efficiently allocate the resources used to make payments in a country. 3. It is also useful to know where some of the "hidden" costs are in a payment system. In a country which relies on check payments for business transactions, the largest hidden cost is probably check float (except for Canada which has essentially solved its check float problem by pricing to business users of checks). Check float is important in the U.S. (as well as China, which also uses checks) because the larger is check float, the greater will be the working capital needs of payers in the business sector. Of course, the greater are the working capital needs, the greater are the transactions costs involved in making payments. Float can be looked upon as a tax on the payments system. The revenues raised through this tax go to the banking system, to other businesses, or to the general public depending on who the large value payers and payees are. It is a zero-sum game that eats up real resources as the various parties to a check transaction spend real resources in an attempt to maximize their check float benefits. Thus, overall, check float raises the cost of doing business. A second hidden cost to the payment system is inflation. Inflation tends to draw funds out of the banking system and into cash, which is respent more often in order to minimize the holding of idle balances which lose purchasing power. Thus inflation leads to increased turnover of payments and provides incentives to invest in activities that hold their value, but these are typically more speculative than productive. This, clearly, has been the result of inflation on the payment system in Latin America. A third hidden cost to the payment system is foreign exchange convertibility. Foreign exchange convertibility is essential for international trade and investment. When it is difficult to make international payments, trade and investment suffer. Consequently, it is best to have an efficient mechanism which can easily transfer funds between countries when the currency unit differs. This is usually done today through the holding of reciprocal balances between banks in different countries. 4. Large value payments can be quite important even though the transactions volume is small. Almost all large value payments are entirely business to business or government to business payments. While relatively few in number, business and government payments account for almost all large value payments in a country. Thus an efficient business sector requires an efficient and safe method to make large value payments. - 27 - In practice, the need for large value payments rises with the level of regional, national, and international business and financial market activity. Because of the large values involved, special security procedures are needed. These involve dedicated communication lines, special passwords that are changed frequently, established verification procedures, backup operational facilities, and even message encryption. Such arrangements are not cheap but, relative to the amounts of funds involved, are often well worth the extra effort. Large value payment networks can be debit or credit transfer systems. The latter is less risky than the former since unintended overdrafts can occur with a debit based system. This creates systemic risk in a payments network. Large value transfers can be made on a gross or net payment basis. In a gross payment system, each transaction is followed with a separate payment even though a series of offsetting transactions may have occurred between the two parties to a payment. If a series of offsetting transactions is netted down to a single payment for the net amount due, trien We t'lave a 11et pay111e11t svste111. hnpo1tai11tl 1, the pl!l'f'l'T'leAt she1:1I~ Fefleet tt:ie wRdarl'fiAQ contractual relationship between the parties involved. If the contract treats each transaction separately, even if offsetting transactions have been made, then only gross payments should be made. Net payments should only occur if the underlying contract is itself in net terms. This issue is quite complex and will be covered in more detail in Chapter 4. 5. Another important issue concerns the degree of central bank involvement in the payment system. The involvement of the central bank in the U.S. payment system is not typical of that of most countries. Most nations have a very aggregated or concentrated banking and financial system and so the banks handle all payment clearing operations themselves. In these countries, the central bank typically only provides a settlement service. This is because only the central bank seems to have the "deep pockets" and public trust necessary for the efficient provision of this service. As well, the central bank often holds idle bank balances for reserve requirement and monetary control purposes and can make interbank transfers among these accounts at a lower cost than if each bank were to hold an idle reciprocal bilateral clearing balance with each of the other banks for clearing and settlement purposes. In the U.S., in contrast, the central bank provides both clearing and settlement functions. The Federal Reserve is, in fact, a major provider of payment clearing services to the banking system which, in turn, provides payment services to businesses and the general public. This shift in responsibility in the U.S. has probably occurred because of the fragmentation of the U.S. banking system, which is itself a political response to an underlying concern regarding the concentration of financial and economic power. • 28 • While most central banks will not be directly involved in the clearing functions of the payments system, it is appropriate that they be given oversight and rule-making responsibility for controlling risk and unintended credit expansion during the clearing process. This concerns the responsibility to minimize, eliminate, or collateralize daylight and overnight overdrafts that can occur during the clearing process and unintentionally expand the money supply, creating systemic risk. 6. A typically unrecognized issue concerns the portion of the domestic money supply tied up in idle balances needed for the payment system. This effect on the money supply is directly related to the structure of the banking system. At one extreme there is a completely centralized monobank arrangement where all balances are held with only one institution so all clearing and settlement among customers occurs internally to the single (mono) bank using customer account balances or deposits. Here there is only one tier in the payments system and no interbank balances need to be held to clear and settle payments. More commonly, there exist a number of banks within a country: each bank holds balances with a central bank for interbank clearing and settlement in addition to the customer account balances used for clearing customer payments at each bank. While this is also a centralized arrangement, clearing and settlement occurs both internally to banks and externally between banks using the central authority for final settlement of interbank payments. Two tiers exist in this case and the need for interbank balances can be quite extensive if there are large numbers of banks. The level of interbank balances can be reduced through the consolidation of the banking system. This can occur directly through mergers and acquisitions among banks so that more payments are cleared internally within each (larger) bank, or it can occur indirectly through additional tiering. Additional tiering of the banking system arises when there are numerous small and large banks and/or there are multiple economic or money centers in the country. Small banks usually find it cost-effective to clear and settle their non-local, low volume payments through a larger bank that already holds the necessary clearing and settlement balances either with other large banks in other regions of the country and/or with a central authority. Such "correspondent clearing arrangements" are quite common in the U.S. and the U.K., which both have numerous banks and multiple economic centers, and allows smaller banks to conserve on their holdings of interbank balances for clearing purposes (over and above those balances that might be held for reserve requirement/monetary control purposes). Multiple tiering, along with actual mergers among banks, may represent the path chosen by Russia as its banking system becomes more market-orientated over time. Once a country's banking system moves away from a monobank structure, consolidation among banks and multiple tiering of clearing and settlement can free up for other uses that part of the money supply tied to interbank clearing balances in the payment system. - 29 - 7. Lastly, it is important to find out who gains or loses from the way payments are currently handled in a country. This is necessary in order to determine '1Vho is going to be for or against any suggestions to restructure or otherwise alter the way payments may be currently handled. It is entirely possible that vested interests--either within the banking system, in business, or government--may not be in favor of improving the efficiency of their payments system because they will lose revenues and/or control of various payment functions. One case that comes to mind concerns efforts to speed up the clearing and settlement of check transactions in China. Check float in China can last up to 30 days and, for certain types of large value payments, a private or state owned firm's bank account is debited some 30 days before the funds are actually moved to another account, giving the bank a 30 day float benefit. It is thus not surprising to find out that banks in China were not enthusiastic supporters of efforts to shift large value check payments to virtually same-day large value wire transfers in some centralized processing facility. The solution to this conflict ,s to recognize now tNe bahR:lng syste111 lilay lose it r ti ris, revto erre"~emeAt aAe tr'( te sgmet:io,111 compensate them for this revenue loss rather than just take it away and give it to some other sector of society. Finally, it is often the case that a nation's payment system can be substantially improved with only minor changes in how it operates. Advanced electronic payment arrangements are nice but not necessarily cost-effective solutions. It is useful to remember that a payments system functions, to a large degree, on trust and that new innovations in the payments area, while technologically feasible, are not always embraced in the marketplace. Thus an evolutionary approach is best when it comes to altering the payment arrangements of the general public. However, more rapid change is possible for the business sector because the costs and benefits of improved payment arrangements are typically larger and more visible. Chapter 3: PAYMENT PRACTICES: SMALL VALUE PAYMENTS Small Versus Large Value Payment Instruments. Payment instruments and payment systems are most conveniently divided into two groups: small value payments and large value payments. This is because there is a clear hierarchy in the use of payment instruments which is related to the average value of the transaction being undertaken. There is another distinction as well: small value payments are typically "retail" merchandise and bill payments initiated by consumers to businesses or business wage payments to workers while large value payments usually are "wholesale" payments from business to business or from government to business and are functioned through direct interbank transfers, both domestic and international. Thus the type of instrument used, who it is used by, and what it is used for, are all related to the average value of the payment being made. As the average value of a transaction rises, the method of payment moves from cash, to credit card, to check, to ACH or GIRO, and finally to a wire transfer. Cash, composed of coin and currency, is typically used for very small value transactions. Cash is very convenient, has a low opportunity cost, and the cost of losing a small amount is not great. Importantly, cash represents final payment and fraud (counterfeit) risk is usually slight. Although data on the overall importance of cash payments is quite sparse, the usual value of a cash transaction in the U.S. is on the order of a few dollars to perhaps $25 and has been estimated to comprise from two-thirds to over eighty percent of the number of all payment transactions. 2 This is similar to the estimated transaction use of cash of 7 8 % in the Netherlands, 86% in Germany_ and 90% in the United Kingdom (Boeschoten, 1992). Of course, in many developing countries, due to a lack of acceptable alternative payment instruments, cash transactions would comprise an even larger percentage of the number of payments. As the average value of a transaction rises, the opportunity cost of holding non- interest earning cash balances also rises and holding larger cash balances correspondingly increases the cost of loss or theft. As a result, substitutes to cash payments are intensively used in developed countries for medium sized transaction values, say from $25 to over $5,000, depending on the specific payment instrument used. Ranked from lowest to highest average transaction value, the ordering of the use of these cash substitutes would be credit cards, checks, and finally ACH and GIRO payments. Finally, for very large value payments, those with values between tens of thousands to millions of dollars, a telex or wire transfer network is typically used. These networks offer the greatest safety due to the use of special security procedures, such as passwords, call-back requirements, restricted entry facilities, i/ National Commission on Electronic Fund Transfers, 1977, p, 333; Humphrey and Berger, 1990, Table 2-A1. - 31 - dedicated communication lines, and sometimes encryption. 3 In this chapter, we focus on small value payments. Large value payments are covered in Chapter 4. In what follows, various aspects of the many different types of small value payment instruments are discussed. In particular, we outline the primary legal and operational attributes of the two most important non-cash payment instruments--check and GIRO payments--and note the risks which exist on their associated networks. Also covered are cash, credit cards, direct debits, and various specialized checks such as certified checks, money orders, and travelers checks. Cash. As noted above, there is little information on cash usage. Of the information that does exist, the best comes from infrequent special surveys on cash and other small value payment instrument use by family units. One such survey for the U.S. illustrated the various ways family members obtain cash. These results are reproduced in Table 3-1. Over eighty percent of cash for families is obtained from: (1) cashing a check drawn on soi:t:1aona else's acca,,01 (44%) such as a payroll check from an employer; (2) cashing one's own check for cash at a financial institution (29%); or withdrawing cash from one's own account through an ATM (9%). Prior to the 1940s, most cash was obtained directly from one's employer since it was common for employers to pay wages in cash rather than by check. Because employers shifted to check payments over 1940 to 1960, only around 6% of cash needs are satisfied in this manner today (Table 3-1 ). ]/ Wire transfer systems in Europe handle a broad range of business payments which substitute for business use of checks the U.S. This leads to the lower average value of $37,500 per wire transfer in Europe versus the U.S. average value of $4,100,000 per transfer (Tables 2-1, 2-2). - 32 - Table 3-1 Source and Method of Obtaining Cash ( 1984, United States) ·.··· · · · ·· · •· · · · ·•· Percentirot;le .•.a .. . . ·.••.e.·.h .·• _t_·.·••.• ."Me ··· ·· ·• .<.·••. .s .····•. a .11 ···" ..·· _ ·e cl·< . . . . · i=amil,'iis VsiHA> .·•·. . •...•.•..••. o v .m _ · .·.. Y _••. J o.···. ·•.·.••t •M•· e ·· •.•.h·.· ..• o .·d .·..•.•.•.•.•.·· ··.··•.:.· • ..·.·.·...•..· .•.•.·- . •. ••-·•.•.•.·••. .· .·£. . r ···.• ····,··s .. •••························································ ·••>>• J \\ii<•• 1iJ/1X1.tll/.t€li!fi.Wvff}5p~ijfi{ftjfeA•• • •· · · · •· . .• •· · · · · · . •. · •· · If/ • Cash 36% 100% Credit card 6 52 Check 57 83 * Many families use multiple payment methods, so total can sum to more than 100%. Source: Avery, Elliehausen, Kennicke/1, and Spindt, 1986, Table 1, p.88. Costs and Benefits of Using Cash for Payments. The vast majority of all transactions in all countries--developed and developing--are made with cash. The actual value o currency an com per person available estimates of the proportion cash transactions are of the total volume of all transactions. Surprisingly, one reason why per person cash holdings appear to be so large in many European countries is that cash is hoarded and used as an important store of value outside of the banking system. Use of cash as a hoarded store of savings value is reportedly even greater in developing countries where the evolution of (governm1mt or private) financial institutions is still at a relatively early stage. Table 3-3 Value of Currency and Coin per Person (1992, 11 developed countries) Switzerland $2,748 Japan 2,739 Germany 1,534 86% Sweden 1,467 Netherlands 1,354 78% Belgium 1,239 France 828 Canada 652 U.S. 465* 83% U.K. 446 90% * ($1,167 total reported) x (estimated .40 inside the country} = $465 per person. Source: BIS, December 1993. - 34 - The primary direct cost of using cash for payments concerns printing, distribution, and reissuing expenses (when the currency is worn out and has to be replaced). 4 An indirect cost of cash is that it is an inefficient and insecure method for making large value payments typically associated with business transactions. In addition, development experts often point to hoarded cash savings in a country as a source of funds that, if deposited in a banking system, would promote economic growth by supplying loanable funds for investment. Consequently, many countries believe that an important priority for economic development is to try to shift people from using cash in transactions to using checks or GIRO payments provided by a banking system, in exchange for deposited funds which are then used to make loans. In this manner, savings .hoarded in idle cash balances can be utilized by a banking system to promote development at the same time that checks and/or GIRO payments are substituted for cash in making payments. This argument, while plausible on its surface, neglects the fact that when the government issues currency, it obtains seigniorage revenues. In effect, seigniorage represents a long-term, low cost "loan" to the government by users of cash. The "interest rate" on this "loan" is given by the ratio (cash printing, distribution, and other cash costs)/(value of cash issued). Importantly, the "loan" is not repaid as long as the currency which originally generated the seigniorage revenues remains in circulation (and is replaced when worn out). It matters not if the issued cash is used for transactions, held idle in the form of savings, or shipped overseas and used for transactions in another country. What does matter, however, is if this cash is subsequently deposited at a bank and check and/or GIRO payments are substituted for cash in transactions. When this occurs, the reduction in cash usage is equivalent to a redemption of the "loan" made earlier to the government. Government revenues will be smaller than what they otherwise would have been since the "loan" is repaid when cash is permanently withdrawn from circulation (or newly issued currency grows at a slower rate). The effects of seigniorage and payment substitution are illustrated in Figure 3-1 . The government issues cash and, in return, receives a long-term, low cost "loan". While seigniorage revenue is typically spent on government consumption or public assets, it could just as easily be earmarked for loans through government or privately-owned banks within a banking system. If the latter course were chosen, the results would be qualitatively identical to encouraging holders of idle cash balances to deposit some of these idle balances in the banking system, promoting loan growth, and shifting cash users to check and/or GIRO payments for all but very low value transactions. As Figure 3-2 makes clear, promoting check or GIRO payments for transactions will indeed succeed in mobilizing idle cash balances for the ~/ In the U.S., for example, a one dollar ($1) bill lasts only about 1 year before it has to be withdrawn from circulation and replaced with new currency. - 35 - banking system to make loans with (bottom half of the figure) but will also result in a loss of government revenues from seigniorage (top half of the figure). Importantly, the net effect on economic development depends on the productivity of what the government used these mobilized savings for--held as idle cash balances by the public--versus what the banking system uses the mobilized savings for--held as idle deposit balances by the public. Depending on the answer to this question, there need be little or no net increase in mobilized savings from reducing the use of cash in transactions. Either seigniorage or payment substitution can give the same result. The primary difference would be in the productivity of the mobilized savings used as government spending, government loans, or loans from privately owned banks. Figure 3-1 Seigniorage and Payment Substitution Government I USER 1 -·--ii GOVERNIENT 1.:p~ong .. , ~.'°811_·_,e_ve-nu-e .... ECONOMY ----- Issue cash Goods/assets _ _ _ _ ___, PAYMENT SUBSTITUTION Deposit cash I. · USER 1:~dM) • 1 BANKS --·-ti ~--Lo_an_s 1.... ECONOMY Check/GIRO ----- Bank assets Now that the savings mobilization argument is placed in its proper perspective, the focus of payment substitution should be on providing a low cost, timely, and safe method for making relatively large value payments for business or inter-enterprise transactions, not necessarily on reducing consumer use of cash in transactions. Consequently, it is important to shift from business cash payments to check or GIRO payments in developing countries. In developed countries today, a somewhat different shift is occurring--from consumer cash, check, or GIRO payments to POS payments and (in the future) from cash use to an electronic purse. While the availability of ATMs encourages increased use of cash, POS systems replace both cash and checks. All developed nations have shifted to A TMs over last 20 years but they are currently, especially in Europe, shifting to POS for retail payments. - 36 - Checks. Checks are used to some degree in almost all countries for retail and bill payments. Among developed countries, checks are used most heavily in the U.S., Canada, and France. As shown in Table 3-4, checks are used in 81 percent of U.S. noncash transactions. For Canada and France, checks account for 62 percent and 51 percent, respectively, of these countries' noncash payment volume. Among other countries shown in Table 3-5, checks are heavily used in India (99 percent), Korea (74 percent), Australia (60 percent), and South Africa (60 percent). Many countries in Western Europe--Switzerland, Sweden, Germany, and the Netherlands--do not intensively use checks but instead rely primarily on GIRO or credit transfer payments. Finally, the countries of Eastern Europe and the states of the former Soviet Union do not yet have well-developed noncash payment instruments and so, for now at least, rely on cash for almost all of their payment needs. Table 3-4 Noncash Payment Use in Developed Countries ( 1992, percent) } Use of.Credit > • Use of .Use of .chetks < fflahslets.< •••··.>.• ,•. •. •.v, ,.• < :::. :'::::.: .•. . . . . .. ,: . . ·.: u : :. ..: . . ·.· ..... : •::::::. ·'.::l '.·. · .•s e < .· •· ·· .... ':•,· ..·o ·'.·.· : f·... ,:,.•.o · . ,s\"' ... .•,·. • '· ·:f o .• .·.•.·, • ·c ··'. .,fe II "'•c .?·· •.•.:. k S>: · ·:. ,,. • ,;: se: : o, t 1eu1 >'< !t#riJr/Jt!s · :,> Us~/ofQaFdsU v; Debtts., •. , ._. India 99% 0% 1% 0% Korea 74 16 5 5 Australia 60 20 16 4 South Africa 60 6 19 15 Thailand 26 7 65 2 A check is a debit payment instrument and the information it contains is illustrated in Figure 3-2. Basically, a check is just a standardized communication to a bank to debit the account of the payor, giving funds to a payee; it also serves as legal evidence of payment. 5 The value to be transferred is written on the check twice, first as a number, and second in words. 6 The date appears on the check, as does the payer bank's routing/transit number (a special identifying number unique to each bank) and the payer's account number. These two numbers are on what is called the MICR line (for "magnetic ink character recognition") . Magnetic ink and specially designed printed numbers increase accuracy when the checks are read by high-speed check reader/sorter machines during processing, which is standard. 21 Recall that the entity to receive the funds is the payee while the entity whose account is being debited is the payor. !' If the two values do not agree, the value written in words takes precedence. However, since the value written in numbers is the value used to determine the amount to debit the payor' s account, nonagreement will only be discovered if there is some "problem" such as insufficient funds in the account or a complaint by the payor that the wrong amount was paid. - 38 - Figure 3-2 Illustration of Information on a Check Date Payee name Value in numbers Value in words Payor signature Payor bank number and payor account number The payor's signature on the check attests that the debit instruction is from the payor. While the paying bank has the legal right to verify the authenticity of the signature by comparing it against the payor's signature on file at the bank office before paying the item, this right is rarely exercised in practice. The signature is typically only examined when the value of the check is exceptionally large or when the bank is accused by the payor of cashing a forged check. When a check is deposited by a payee, the payee turns the check over and signs (endorses) it. This gives the payee's bank the right to collect the check on behalf of the payee and credit the payee's account with the funds collected. The deposited check is accompanied with a deposit ticket with the payee's account number and the payee bank's routing/transit number, both imprinted in magnetic ink. The first step in processing the deposited check is for a clerk or an optical character recognition machine to read the value of the check written in numbers (not words) and encode this amount in magnetic ink on the MICA line of the item. At this point, all the necessary information to process the check is in machine readable form: the amount to be debited, the bank and the account to be debited, and the bank and the account to be credited. What remains is a cumbersome process of physically transporting the check to the paying bank so the funds can be transferred between accounts either at different banks (a transit item) or between accounts at the same bank (an "on-us" item). When the banking system is concentrated, such as in Canada where the top 5 banks account for over 90 percent of deposits and nationwide branch banking exists, the above information can be transferred electronically. The physical item will follow later using low cost, non-time-critical transportation. Electronic transmission of payment information is possible in Canada because the banks have agreed among themselves to give availability to deposited checks on a same-day basis for their regular customers. In addition, since Canadian banks charge payors for the time it takes the payee's bank to collect funds from a payer's - 39 - bank, payors have a strong incentive to disburse checks near where the payee is located so that check float is minimized. Just the opposite incentive exists in the United States. There is no negotiated agreement among banks in the U.S. to pay deposited checks on a same-day basis. Instead, funds availability is given only after the payee's bank physically presents the check to the payor's bank, a holdover from the past when this was the only feasible method of collecting a check. As a result, payors attempt to lengthen the time it takes to collect a check by disbursing from points distant to where the payee is located (called remote or controlled disbursement). The more time it takes to collect a check, the greater is the float benefit to the payor. Consequently, electronic presentment is not allowed (except by special bilateral bank agreement) and expensive time-critical transportation procedures are always used to transport checks to paying banks. Because of a float-distorted incentive system--where it pays to maximize float in the U.S., rather than minimize it as in Canada--real resources are wasted as payor and payee banks fight over the division of float and payees wait 1 to 2 days (er FR ere) before the¥ cao abraio 11se of funds from deposited checks. 7 The perverse economic incentives associated with check float is perhaps the second greatest disadvantage of having a check-based payment system, a disadvantage shared by all but the Canadian check system. The greatest disadvantage of checks, however, lies in the fact that a check is a debit instrument. If the payor does not have sufficient funds in his account to pay the check after it has been presented to the payor bank, then it is returned unpaid to the payee who then has to recontact the payor and arrange for repayment. In this instance, the check collection process is reversed and the check becomes a return item. Return items are quite expensive to process and banks commonly charge $1 5 to $20 for each check returned, a charge that far exceeds the actual operating cost involved in the return item process. Return items can also lead to payee losses (as when the payor has the merchandise while the payee holds an uncollectible check). Return items account for around 1 percent of all checks. However, with 60 billion checks written in the U.S., return items occur 600 million times a year. If there is a lesson to be learned in the use of checks it is that, if checks are to be relied upon at all, the rules should be structured to permit electronic collection and payors-- z, Negotiation among banks to alter this arrangement would be difficult in the U.S. for three reasons: ( 1) payors would have to give up their present float benefits (although, since payors are also payees, many would gain back as payees what they lost as payors); (2) the banking system is highly disaggregated (it would take over 3,000 banks to account for 90 percent of deposits, where in Canada it takes only 5 banks) making negotiation cumbersome; and (3) nationwide banking does not exist, so the benefits of electronic collection of non-local checks would accrue to other banks rather than save costs internal to the bank (but since payee (collecting) banks are also payor (disbursing) banks, the reduction in collection costs would indirectly benefit all banks). - 40 - not payees--should bear the cost of check float, as in Canada. 8 Because of the perverse float incentives and the cost of return items, the U.S. check system is much more expensive than it need be to payment users. A more efficient alternative to checks is the GIRO, which is used extensively in Western Europe. Specialized Checks: Certified Checks, Money Orders, Travelers Checks. Specialized checks, such as certified checks, travelers checks, and money orders, all solve the return item problem inherent in the use of checks (as does use of a credit card). A certified check is where the bank certifies that funds are sufficient to cover the value of the check. The bank typically debits the payor's account for the full amount when the certified check is issued, transferring the funds to a special account internal to the bank so that the certified check is certain to be paid when presented. In addition to charging a small fee for this service, the bank earns interest on these funds--a float benefit--until the certified check is paid. Certified checks are used primarily in business transactions for large amounts where it is important to eliminate the possibility of a return item. Money orders are essentially just a lower value certified check that is available from more conveniently located nonbank suppliers (such as certain retail stores and the post office). Money orders are paid for when issued and thus also solve the check return item problem and earn a float benefit for the issuer. Money orders are most often used to pay bills by individuals who do not have a checking account, usually because they prefer the anonymity of cash transactions or because the minimum balance and other fees associated with a checking account make the establishment of such an account unattractive. Travelers checks are also similar to a certified check in that they too are paid for beforehand and eliminate the return item problem. However, travelers checks are issued in smaller, standardized amounts and the payee is written on the check as needed rather than at the time they are issued. As the payor's signature appears on the face of the travelers check, it is easy for the payee to determine if the payor's second, endorsing signature is the same. Because they eliminate the return item problem, travelers checks are more acceptable to merchants than is a standard personal check. This is especially true in situations where the payor's regular checking account is at a bank distant from the merchant. Since travelers checks are paid for when issued and are outstanding for an average of more than two months, issuers earn most of their revenue from float. Because travelers checks are used primarily as a substitute for cash while traveling, the existence of nationwide banking or nationwide A TM access will reduce the need for this instrument. Similarly, the existence of international credit !' Very large value items should probably still be presented physically in order to permit the payer's signature to be verified although, in the future, even this may not be necessary if an accurate electronic digital representation of the signature and endorsement can be sent instead along with the standard payment information. - 41 - card networks has also decreased the demand for travelers checks from what it would otherwise have been. GIRO. A check initiates a debit transfer while a GIRO payment is a credit transfer. With a credit transfer, the payor typically has to have funds in his account sufficient 9 to make the transfer, otherwise it is rejected. Credit transfer payment networks thus solve the biggest problem associated with check use--the return item problem. 10 And because a payor's account is typically debited one day prior to when the credit transfer is made to the payee, to cover the cost of making the transfer, the payor bears the float cost of the decision to use a GIRO payment. This solves the second most important problem associated with checks--the perverse incentive to delay funds availability to payees so that payors can capture the float benefit. For these two reasons, a credit transfer system, such as a GIRO, is clearly more efficient and more equitable than a debit transfer check-based system and i:; typically set up to be flexible and convenient to the user. GIROs are the dominant noncash payment method in Western Europe. As seen in Table 3-4, credit transfer or GIRO payments comprise 50 percent or more of all noncash payments in Belgium, the Netherlands, Germany, Sweden, and Switzerland. Among developed countries, the ranking of the intensity of GIRO use is the exact inverse of the ranking for these same countries by their use of checks. However, this inverse ranking between check and GIRO payments applies only weakly to the other countries shown in Table 3-5 and so does not appear to be universal. A national, centralized postal service and consortiums of banks run competing GIRO networks in many countries. The bank-run networks, however, were only recently developed. The postal GIROs handle the vast majority of consumer bill payments and employee or government payroll and retirement payments to consumers while many business- to-business payments are handled over the bank-run networks. As a result, the postal GIRO networks usually account for four times the transaction volume of the bank-run networks. GIROs are used for both personal and business bill payments and interpersonal and interfirm transactions, just as checks are used in the U.S. and Canada. The GIAOs in Western Europe are connected to one another and to SWIFT (a world-wide large value !' GIRO is a Greek word meaning to travel in a circle and reflects the flow of funds around a ring, or a network of payors and payees. Two useful sources of detailed information regarding GIROs are Thomson, 1964, and Hager, 1986. J.E.' In some cases, if funds in a GIRO account are insufficient, a bank check can be used to cover the deficiency. However, the credit transfer is not made until it is determined that there are sufficient funds in the checking account to cover the check. - 42 - payment message transfer network) so that international payments are also possible. The closest thing to a GIRO in the U.S. is an Automated Clearing House (ACH) electronic credit transfer. An example of an ACH credit transfer would be the pre authorized direct deposit by an employer of an employee's payroll amount into the employee's bank account or the direct deposit of retirement, Social Security, or corporate stock dividend payments to a payee's bank account. While ACH credit transfers exist, they account for 2 percent of all noncash payments (or about half of all ACH payments--Table 2-1 ). 11 Since a GIRO payment and a check payment, once they are completed, do the same thing, they both contain essentially the same set of information. Thus a GIRO payment contains the amount to be transferred from payor to payee, the account numbers to be debited and credited, the date the GIRO payment is to occur, and (if the transaction is initiated on paper--which is common for consumers) the signature of the payor giving the GIRO network the authority to debit the payor's account for the amount indicated. A legally acceptable receipt is also generated as proof of payment. One may obtain cash from one's own GIRO account and an overdraft facility may be arranged, just like for a checking account. As might be expected from a centralized system, there is intensive use of pre-printed GIRO payment request forms, optical character readers, and centralized electronic processing. While these automated systems for GIROs were introduced at about the same time (the early 1960s) as were high-speed check reader/sorters systems in other countries, the degree of GIRO automation far exceeds that in any check-based country. As a result, the real resource cost per transaction for a GIRO system is believed to be much smaller than that for a check system (Table 2-5). In addition to being a credit transfer rather than a debit transfer, there are two important functional differences between a GIRO and a check payment. First, it is possible to indicate the value date--the day the credit transfer is to take place--with a GIRO payment. In contrast, the "value date" with a check depends on how fast the check can be sorted, transported, and presented to the paying bank and can not be stipulated on the check by the payor. Second, funds in GIRO accounts pay interest, unlike funds in demand deposit checking accounts in the U.S. 12 As a result, business users of GIRO accounts do not have the same incentive to concentrate idle demand deposit balances in order to purchase interest earning overnight money market instruments. ll' A related development is the use of the ACH for corporate trade payments or EDI (for Electronic Data Interchange) . Here business-to-business debit-transfer payments are accompanied with invoice and remittance information, rather than running a dual system with electronic payments and a paper invoice flow. Although growing, EDI accounts for only 1 % of business-to-business payments and has yet to adopt standardized industry formats for the data being interchanged. ll' While demand deposits do not pay interest, there exist other types of consumer (not business) transaction accounts that do pay interest, such as NOWs (Negotiable Order of Withdrawal) and MMDAs (Money Market Deposit Account). - 43 - Credit Cards. Credit cards represent a way to obtain a preauthorized consumer installment loan from the issuer of the card. In terms of U.S. transaction volumes, bank issued credit cards account for around 45%, retail issued cards for 30%, and oil company issued cards 25%. The average transaction amount is intermediate to that for cash and checks--$45 in the U.S. but $200 in Europe (Tables 2-1, 2-2). Issuers of credit cards make their money in three different ways. First, each time a credit card is used, the seller of a service or merchandise pays a fee to the card issuer of from 1 % to 3 % of the value of the transaction. Second, the credit card holder often pays a small annual fixed fee to the issuer. This fee helps to cover the costs of extending credit to the card holder for from one to four weeks even if the holder pays his monthly bill in full. Third, and most importantly for the half of all card holders who do not pay their bill in full each month, there is an interest charge for the amount of credit extended. This interest charge is quite high--often over three times the normal money market interest rate--due in part to the very high losses incurred on defaulted credit card loans. The high loan losses are the result of poor credit quality controls by the card issuers, allowing cards to be issued to individuals with either a poor credit history or insufficient income to pay off the loans incurred. Merchants sometimes prefer a credit card to a check because their fraud losses are lower, due to the fact that each credit card transaction can be verified with the card issuing agent before the transaction is completed. 13 While card holders are billed for their transaction up to one month later, merchants receive funds from the transaction usually on a next-day basis. Thus float is given to the card holder. Because the annual fee is so small (around $10 to $20), the monthly billed amount would have to also be small (less than $200 to $400, if the market interest rate was 5%) for the card holder to fully pay for the float obtained. Countries that use checks intensively tend also to use credit cards more heavily as well. The data on card use in developed countries is presented in Table 3-4 but covers both credit and debit (POS) card use. Canada, at 29 percent, is the most intensive user of cards for noncash payments among developed countries while Thailand, at 65 percent, is the most intensive user of cards overall. For Thailand, the use of cards is dominated by debit card rather than credit card use (Datt and Shanmugham, 1994). Cards comprise 15 percent or more of noncash payments for Thailand, Canada, U.K., South Africa, U.S., Belgium, Australia, and France. Except for Thailand, cards are never the dominant noncash payment method: this distinction is reserved for either check or GIRO payments for the countries shown in Tables 3-4 and 3-5. 1ll Similar procedures for determining the credit-worthiness of individuals cashing checks exist but are not commonly used. - 44 - Direct Debits. A direct debit typically is a preauthorized instruction to pay recurring bills--such as insurance, electricity, mortgage, or credit card bills--when presented to a payor's bank. In effect, a direct debit is merely an electronic check since there may or may not be sufficient funds in the payor's account to cover the amount to be debited. As shown in Tables 3-4 and 3-5, Germany and the Netherlands are the most intensive users of direct debits, accounting for 39 percent and 24 percent (respectively) of these countries' noncash payments. Only these two countries plus the U.K. and South Africa use direct debits for 15 percent or more of their noncash payments. In sum, while the use of cards and direct debits are at times an important component of noncash payments they do not (Thailand excepted) dominate either check or GIRO payments. Risk on Small Value Payment Networks. There is very little risk on small value payment networks. This is because the average payment amounts are themselves small and the total values being transferred are only 1 5 to 22 percent of the value being transferred each day on large value networks (Tables 2-1, 2-2). The small risk that does exist on small value networks has been adequately dealt with by clearly spelling out the rights and responsibilities of the various parties to a payment transaction--the payor, payee, intermediate (processing or clearing) parties to the transaction and their agents (such as the payor's and payee's bank or payment agent). Most of these rights and rules are codified in legislation or government rule-making and have been well-tested in court. When the parties to a transaction clearly understand their rights, responsibilities, and liabilities, they are able to take proper action to protect themselves and preserve the integrity of the transaction. While risk is low on all types of mature, small value payment networks, some networks are inherently less risky than others by virtue of how they operate. This is the case for debit transfer versus credit transfer networks. Because a credit transfer system will reject a transfer if the funds in the account to be debited are insufficient to cover the requested payment, there are no return items as there are with a check-based debit transfer system. In effect, a credit transfer provides good and final funds for the payment 100 percent of the time while a debit transfer (statistically speaking) provides good and final funds for payments only 99 percent of the time (as return items are 1 percent of all checks written). While most return items are due to payors' believing (incorrectly) that they have sufficient funds in their account when they tender a check to satisfy a payment liability, some return items are the result of fraud on the part of the check writer. Since payment instruments that provide good and final funds contain less risk for payees, it can be concluded that a GIRO network is less risky than a check-based network. Similarly, the use of cash, credit cards, and specialized checks (certified checks, money orders, travelers checks), which also eliminate the return item - 45 - problem, are generally less risky than checks. 14 The Future of the Check in Developed Countries. As noted earlier, while check use represents more than half of all noncash payments in the U.S., Canada, and France, it is much less heavily used in other developed countries (Table 3-4). Check use, particularly in the U.S., has been associated with a highly decentralized banking and payment structure due to the large geographical size of the country and the particular political considerations which have led the U.S., until quite recently, to actively prevent any real consolidation of the banking system as has occurred in other countries. With a smaller geographic size and, importantly, a centralized structure of postal savings accou:its, most European countries have tended to concentrate on GIRO payments rather than checks. From this perspective, the existing structure of payments in developed countries is seen to have been determined historically by geographical and political considerations. As electronic substitutes to paper-based payments are increasingly adopted and made available to users, checks will become less important. This trend is already evident in most developed countries. Table 3-6 shows the number of checks written per person per year In 11 developed countries over 1988-1992. Two things are observed from this table. First, we can see more clearly how atypical check use is in the U.S. compared to other developed countries. Check use per person in the U.S. is more than~ as high as any other country. The second observation to be drawn from Table 3-6 is that in all but the U.S. and Germany (which already has a low use of checks) the per person use of checks has reached a peak--underlined in the table--and is falling. The fact that per person use of checks is increasing at a decreasing percentage rate in the U.S. suggests that it too may reach a peak soon and begin to fall. This possibility is illustrated in Figure 3-3. A longer series of estimated per filtu11 person use of checks in shown over 1980-1992 for the U.S. 15 A forecast suggests that the per person use of checks in the U.S. will reach a peak in 1995-1996 and fall thereafter. If this occurs, then we will have reached in stage in payment system development where the check will be in decline in developed countries, although in the U.S. this decline will likely be much less rapid than elsewhere because of the atypical disaggregated nature of the banking and payment system. ll' Of course, use of cash leads to the risk of loss or theft so that, depending on the safety of the country involved, cash can be a poor substitute for checks when relatively large payments are made. lll Adult (age 18 or older) use of checks is a more accurate indicator of the realized demand for checks. It abstracts from changes in that segment of the population which typically does not have a bank account and does not write checks. - 46 - Table 3-6 Number of Checks Written per Person per Year .1988 . .1989 U.S. 208 214 222 227 229 Canada 83 82 82 80 78 France 81 82 86 84 85 U.K. 54 55 56 54 52 Italy 11 12 13 12 11 Belgium Z§ 23 21 19 17 Netherlands .19 18 17 16 15 Sweden 19 20 14 9 8 Germany 10 11 10 1, 11 Switzerland 3.8 ~ 3 .6 3.2 2.8 Japan ~ 3.1 3.1 2.9 2.8 (Numbers are rounded to emphasize trends) Source: Computed from BIS, December 1993. Lessons Learned: How to Improve Small Value Payments Efficiency. In the long run, small value payments efficiency can be improved by developing or using more intensively a centralized, electronic credit transfer system in place of checks or cash. This eliminates the return item problem (reducing risk), improves efficiency (since GIROs are estimated to have the lowest per transaction cost of almost all other payment instruments except cash--Table 2-5), and properly places the full cost of using a payment instrument on the payor (who is the decision maker and should face prices which reflect the real resource costs of using each payment instrument in order to make the most efficient decision). - 47 - Figure 3-3 Actual and Predicted Numbers of Checks Written Per (Adult) Person Per Year in the U.S. Check PP 320 300 280 260 240 220 200 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 In the short run, small value payments efficiency can be improved by providing for the electronic delivery of payment information within an existing check-based debit transfer system. This would not only reduce transportation costs, but would also substantially eliminate the benefit of check payment float created through inefficient, real resource using remote and controlled disbursement by business payors. 16 And in the medium term, small value payments efficiency can be improved by rule-making or interbank agreement to provide access to deposited funds on a same-day basis while charging payors for any float created in the check collection process (as is now done in Canada). As seen, the primary way to improve the efficiency is to follow Europe's lead and develop a credit transfer GIRO network for small value payments. Failing that, efficiency can be improved by revamping an existing debit transfer check network along the lines developed in Canada . .ll' In the U.S., for example, business payments account for 90% of all check float because these checks are (al written for the largest average amounts and (b) are often remotely disbursed to make timely collection more difficult. Chapter 4: PAYMENT PRACTICES: LARGE VALUE PAYMENTS Large value payment networks are quite important for the efficient handling of interbank and business transactions in a private market. And there are a wide variety of these networks which serve this purpose in different countries. They range from centralized, virtually instantaneous, electronic, final payment, wire transfer networks in countries with well-developed markets (the U.S., Japan, and Switzerland) to decentralized, time consuming, paper-based, provisional payment, payment order arrangements in many newly emerging socialist countries with rudimentary markets (such as Russia and Ukraine). Intermediate to these two extremes are large value payment networks using centralized or decentralized, same-day, telex (telephone and telegraph), provisional or final payment procedures in developing countries with partially developed markets (such as India, Brazil, and some countries in Africa). In this chapter, we first distinguish between gross and net settlement large value networks, which may or may not extend intraday credit to participants. Second, we outline the general operation of three types of large value payment networks: entirely electronic, partially electronic (telex), and wholly paper-based. The purpose is to illustrate the conceptual similarities as well as the major technical and institutional differences that can be found among large value payment networks in different countries. Third, the remaining portion of the chapter is devoted to the risks that the operation of these networks pose to their participants, the central bank, and financial markets, along with their monetary policy implications. Since these risks are greatest for large value networks currently operating in international money centers, the focus will be the operation of two large value networks in the U.S.--the wire transfer operations of CHIPS and Fedwire. Although the focus is on these two networks, the potential problems associated with large value payments are generic and exist on all types of large value payment networks in all countries today. Importantly, if these problems are not adequately addressed, they will continue to exist on many developed country networks and will arise on the new networks now being planned in the newly emerging socialist countries as well. Before these problems are discussed, it is necessary to describe in some detail the different types of large value payment networks which exist and outline their operation. Gross And Net Settlement Networks. A gross settlement network is one where each payment sent is separately settled at the time it is sent. Fedwire, the network owned and operated by the U.S. central bank (the Federal Reserve), is a gross settlement network. Each payment is settled by debiting and crediting idle bank reserve balances and all payments are final at the time they are made. In the event that a reserve balance may be insufficient to cover a requested transfer on Fedwire, daylight credit is often extended, within certain - 49 - limits, to the sending bank. Thus Fedwire is a gross final payment network with an intraday credit facility. The Swiss Interbank Clearing System (SIC) network in Switzerland is also a gross final payment system. However, SIC does not provide daylight (or overnight) credit if the reserve balance with the Swiss central bank is insufficient to cover the transfer request. Instead, the payment request is either rejected (returned to the sender) or is held (and queued) until sufficient incoming funds are credited to the sending institution's balance to raise it to a level where it will completely fund the requested transfer. At times, if the transfer is exceptionally large, it can be broken down into smaller pieces so that at least part of the transfer will likely be sent earlier. The central bank of Japan (Bank of Japan) also provides a payment service similar to that of SIC where requested payments are sent only if covering funds are available. 17 For these gross payments, and those over SIC, no central bank daylight credit facility is routinely available. On a net settlement network, each payment sent and received among bank participants is cumulated and only the resulting net position between banks is settled at a particular time of the day. In contrast to a gross settlement network where settlement is continuous and there are as many settlements as there are payments, there are only one or two settlements each day on a net settlement network. While such a system reduces the size of the idle balance needed to be held for settlement purposes, it will also mean that the payments made are not final until settlement occurs. In effect, payments made over a net settlement network are provisional until settled. One important net settlement network is the Clearing House Interbank Payment System (CHIPS) which is owned and operated by a group of large U.S. banks. Net settlement on CHIPS occurs at the end of the business day. 18 A similar privately operated interbank, end of day net settlement network--Clearing House Automated Payment System (CHAPS)--exists in the United Kingdom. As well, the central bank of Japan operates a net settlement network (BOJ-NET) where net settlement occurs at two times in the afternoon of each business day. Because banks receiving payments over these net settlement networks typically treat them as usable funds prior to settlement, there is a de facto interbank extension of daylight credit among the network participants. On BOJ-NET banks pay a fee to one another when they l1/ This gross payment service is in addition to the provision of a net payment service, described next, which the Bank of Japan also provides. l!Y Previously, the CHIPS net settlement through reserve accounts at the Federal Reserve occurred on the day following the day the payment was made over CHIPS. This next-day settlement arrangement was moved up to the same-day in order to reduce the time between when the CHIPS payment was made and when it became final. This change reduced overnight risk on CHIPS. - 50 - borrow funds to make their afternoon net settlement. On CHIPS and CHAPS, which both net settle their positions at the end of the day, borrowing to make settlement occurs in the overnight funds market at the overnight rate. A Description Of Large Value Wire Transfer Networks. All large value networks essentially perform the same function: they transfer value from one enterprise or firm to another and in this sense are conceptually identical. Where they differ is in the speed, technology, security, finality of payment, and legal framework in which these transfers are made. Figure 4-1 illustrates a generic large value transfer: firm A initiates a payment to firm B by instructing its bank--bank 8--to pay firm B. If the same or different branches of bank B serves both firms, then the transfer is processed and settled entirely within bank B with the debiting of firm A's account and the crediting of firm B's account. This is an internal or intrabank transaction and the exact timing and specific nature of finality of payment is governed by contract, prior agreement, or bank policy between its customers and itself. In developed countries, this type of transfer can be completed on an intraday basis and is final when posted at the end of the day. The transaction may be initiated electronically from the firm to the bank with computers but use of the telephone, with recording, call-back, and password procedures for security, is common. In less developed countries, a transfer between firms A and B using different branches of the same bank can take a few hours to complete if telex procedures are used, or many days if the bank's branches are not connected electronically. In the latter case, each branch office effectively operates as a separate bank for payment purposes. Thus even in monobank (single bank) or highly concentrated banking systems, the determining factor for the speed by which payments can be made is the degree to which the branches of different banks and/or the branches of the same bank are connected by computer or telex. - 51 - Figure 4-1 Operation of a Large Value Payment Network Firm - Bank - Central - ' Bank " ... Firm? y A -- B -- Processor -- C -- D + Firm B + ' Settlement Perhaps a more common large value transfer would be between firms A and D in Figure 4-1 where each firm has its account at a different bank, banks B and C in this example. In this situation, firm A instructs bank B to pay bank C for the account of firm D. This represents an external or interbank transaction. Interbank transfers are typically processed by an central intermediary, which can be a privately owned and operated clearing organization or the central bank. If processing operations take place wholly in the private sector, then settlement of the interbank funds transfer can take place by debiting bank B's clearinn balance held by a clearing organization (independent of the central bank) and crediting that of bank C. But since clearing balances typically are idle, non-interest earning balances, this alternative is typically more expensive than debiting and crediting zero balance accounts, with settlement through idle central bank reserve accounts at the end of the day. This is, in fact, the way net settlement occurs for CHIPS in the U.S., CHAPS in the U.K., and on BOJ- NET in Japan (with afternoon rather than end of day settlement). As noted above, on a gross settlement network the funds transfer--with simultaneous settlement is functioned using reserve account balances held with the central bank. This occurs on Fedwire in the U.S., SIC in Switzerland, and through a separate payment service available through BOJ-NET in Japan. Importantly, the central bank, because of its role as an interbank credit facility or "lender of last resort", is often in a position to provide emergency credit to a bank participant with insufficient settlement funds at the time of settlement. Thus, to increase the likelihood of settlement occurring for all of the banks, privately-run funds transfer networks usually settle through the central bank at the end of the processing day. - 52 - In most developed countries, an interbank funds transfer between banks B and C in Figure 4-1 is fully electronic, uses computers, and can be completed in a matter of minutes regardless of the geographical location of the banks within the country. The speed of the transfer is important to firms A and D but just as important is the provisionally or finality of the transfer itself. That is, at what point can the transfer not be reversed so the receiving party can be assured the transaction has been completed and can thus obtain unqualified use of the funds transferred? Different Types of Payment Finality. There are a number of points in the funds transfer process where a payment may or may not be considered to be final. First, on privately run networks, payments are typically provisional and subject to reversal under certain pre-stipulated conditions, the most common of which is the failure of the sending bank to settle the transfer at the end of the processing day. But even if the payment is provisional to the receiving bank C, and hence to the receiving customer firm D as well, payments on all large-value networks are typically final--nonreversible--to the sending bank B and the sending customer (firm A). This is called sender finality and is illustrated in Figure 4-2: once the sending bank initiates the transfer and notice is given to the receiving bank, the sender can not reverse the transaction. Sender finality protects the receiving bank since the transfer is not reversible even if bank B sent the wrong amount to bank C. To correct errors, bank C would have to initiate a transfer to bank B returning the amount sent in error--bank B can not cancel or reverse the transaction once the funds have been sent. In addition to sender finality, central bank run networks also have settlement finality since once the transfer is made the central bank stands behind it even if the bank whose account is being debited happens to fail and has insufficient funds to make the transfer. This further protects the receiving bank since, even if the sending bank fails prior to settlement, the receiving bank still has good and final funds. While privately run networks do not have settlement finality, they often have something close to it. For example, CHIPS now requires its bank participants to post collateral which could be liquidated to cover an unfunded net debit of one of its bank participants on its network, and thus may still achieve end-of-day settlement through the central bank. This type of collateral arrangement is akin to settlement finality as long as the collateral posted is adequate. The Bank for International Settlements (BIS) guidelines for privately owned and operated large value transfer systems states that private networks should follow the lead of CHIPS and hold sufficient liquid collateral to cover at least the maximum net debit position of any single bank participant which may fail to settle on its network. Further, the BIS has recommended that privately run payment networks move to impose limits on the maximum net debit positions that can be incurred by their participants on their networks (BIS, November 1990). - 53 - Figure4-2 Different Types of Payment Finality in the Payment System Sending Firm A - - Sending Bank B - - Central Processor - - Receiving BankC - - Receiving Firm D ;+ Sender finality Settlement ~ Settlement finality Receiver finality lastly, there is receiver finality--which no large value network now has. Receiver finality would mean that as soon as the receiving bank C received a funds transfer for its customer, firm 0, the bank would irreversibly credit firm D's account. This would protect the receiving customer (firm D) from loss in the event that the sending bank B failed prior to settlement on a private network or for some other reason the receiving bank C wished to withhold funds from the receiving customer. But receiving banks have not yet given up their right of recourse to the funds transferred to the receiving customer. This is because receiving banks may wish to cover customer loan defaults or preexisting customer account overdrafts (in this or other accounts) with the funds received for the account of customer C rather than pass on good and final funds to firm D and attempt to collect on firm D's defaulted loans or account overdrafts in some other manner (e.g., pressing a separate legal action to collect the funds owed and/or standing in line as creditor with others owed money from firm D in a bankruptcy proceeding). In general, funds transfers over net settlement private networks in developed countries are provisional until end-of-day (or later) settlement, are not reversible by the sender (sender finality), and may or may not have guarantees, a line of credit, or collateral in place to provide something like settlement finality should a bank participant fail to settle its outstanding net debit position. In contrast, central bank run gross settlement networks have both sender and settlement finality. No networks yet have receiver finality. Telex and Other Large Value Transfer Systems. In other countries, a large value transfer between firms A and D can take place either intraday or with a one-day lag using a partially electronic system comprised of dedicated telephone or telegraph (telex) connections between banks 8 and C. These procedures usually involve recording, call-back, and passwords to reduce the incidence of fraud and to maintain a verifiable audit trail of the - 54 - transaction. Private telex transfers have sender finality and central bank run systems have settlement finality as well. In situations where the volume of large value transfers is relatively small, a telex-based large value network is quite adequate and the development of a geographically dispersed, high volume, completely electronic, large value transfer network serving numerous banks would unlikely be cost-effective. The exception would be a geographically centralized electronic network serving a relatively small number of bank participants using high speed personal computers in a local area network with "off-the-shelf" existing transaction processing software. In still other countries, a large value transfer network may be entirely paper- based. While such a system provides an excellent audit trail and the use of signature verification minimizes fraud, a paper-based network can take inordinate amounts of time to complete a transfer. This has been the case in Russia and Ukraine where a completely paper- based network operates and the postal system is used to move the necessary documents between most of the points shown in Figure 4-1 . These networks have both sender and settlement finality (although settlement finality is currently in doubt when fraud is uncovered). In transferring funds between firms A and D, firm A would first fill out multiple copies of a paper payment order document, hand deliver it to bank B which is likely to be in the same city, sign it in the presence of a bank officer, and keep a copy for its own records with the remaining copies being retained by bank B (taking 1 day or less). Bank 8 signs the payment order, debits firm A's account (crediting itself), keeps one copy for its records, and courier delivers (if in the same city) or mails (if not) the remaining copies to a branch of the central bank (taking 1 to 4 days). 19 Upon receipt, the central bank's branch signs the payment order, keeps a copy for itself, debits bank B's account (crediting itself), and mails the remaining copies to a central office of the central bank for processing (taking 2 to 7 days, depending on geographical location) .20 The central office follows the same procedure--signs the payment order, debits the sending central bank branch (crediting itself), keeps a copy of the order for its records, and mails the remaining copies to another central bank branch located close to the receiving bank C (taking 2 to 7 days, again depending on geographical location). Upon receipt, the second central bank branch signs the payment order, debits the account of the central office (crediting itself), keeps a copy for its records, and courier delivers (if in the same city) or mails (if not) the payment order to bank C (taking 1 to 4 days). Bank C, upon receiving the payment order, signs it, keeps a record copy, debits its account with the second central bank branch (crediting the account of firm 0), and waits for the receiving customer to call the bank to find out if the funds have arrived and are available. Since firm D will by this time usually be calling his bank multiple times during the day to see if the .l!' Courier delivery need not be same-day and often is every other day. '1,9.I On occasion, the central bank branch can instead mail the payment order directly to another central bank branch, bypassing the central office and reducing the time lag involved. - 55 - expected funds have arrived, the firm will typically obtain use of these funds on a same day basis. Finally, a representative from firm D arrives at bank C, receives the payment order, signs it, keeps a copy for its records, and knows that it can now use the funds received to pay some other firm using the same payment order process. The timing of the complete payment order transaction is unpredictable and, depending on the geographical location of the parties to this transaction, can take as little as 2 days (if all the parties are in the same city) to as long as one month to complete (as when the parties are quite distant from one another). Of course, use of a dedicated overland courier (or even small planes on high volume routes) in place of the mails could substantially reduce the time needed to complete a paper-based large value transfer while still retaining the same processing procedures and audit trail. Such a speedup could substantially reduce the need by firms for costly payments working capital, although the float benefits accruing to the banking system and the central bank would in principle be correspondingly reduced. While float on a large value payment network can be an important cost in a country with relatively stable prices or moderate inflation, the rapid price inflation experienced by Russia and Ukraine--in hundreds of percent annually at the time of writing--adds considerably to this cost through the rapid reduction in the purchasing power of funds received on a delayed basis. In this case, float would incorporate both the inflation rate plus the real return on borrowed funds. For Russia and Ukraine, the inflation rate far exceeds the real return on borrowed funds and thus is by far the major cost of a slow moving large value funds transfer system. In countries where inflation is quite low, it is the real return on borrowed funds that one associates with float. Overdraft Risk on Large Value Payment Networks. As seen, large value payment networks transfer large sums of money among firms using their banks as the intermediary to the transaction. The average transfer over a U.S. large value payment network exceeds $4 million. Consequently, when a transfer is made, it would be nice if the funds being transferred from one bank account to another were actually in the account before the account is debited. Of course, the customer may have funds in his account at his bank but the bank may not have funds in its interbank payment account. When an interbank payment account is debited but no funds are there, an unintended credit extension occurs. As a result, banks will have created new money to finance more transactions and loans than existed before. This is the monetary policy consequence--an unintended expansion of the domestic money supply. If this credit extension occurs only during the day, it is called a daylight overdraft. If the credit extension is still not made up by the end of the day and is carried as a net debit overnight, then an overnight overdraft has occurred. - 56 - Overnight overdrafts carry a penalty, and thus are infrequent. Daylight overdrafts carried no penalty, and thus were common. Daylight overdrafts were priced in 1994 and this has decreased their levels. Still, daylight credit extensions occur each day over Fedwire, which is the Federal Reserve's wire transfer network. Fedwire funds transfer daylight credit extensions totaled $ 70 billion each day. Daylight credit extensions on the Federal Reserve's U.S. government security transfer network (which facilitates the funding of federal government debt) have been even higher, averaging around $100 billion each day. The problem is that a bank with a daylight overdraft in its reserve account with the central bank could fail before paying off its daylight overdraft by the end of the day. This poses a significant risk of loss to the Federal Reserve, the owner and operator of the network. An even greater problem arises on CHIPS, the privately owned large value payment network operated by a set of large New York banks. CHIPS has funds transfer daylight overdrafts averaging $52 billion a day. The reason why the risk on CHIPS is greater than that on Fedwire is that the Federal Reserve faces the risk of loss from one bank's failure to settle its overdraft position, while on CHIPS one bank's failure to settle can lead to the failure to settle by other banks as well, in a domino-like fashion. This is called systemic risk. Some years back, the Federal Reserve simulated what might happen if one bank failed to settle on CHIPS. The simulation used actual CHIPS payment data and found that the unexpected failure to settle by one large participant on CHIPS could lead to the failure to settle by almost one-half of the over 100 other CHIPS (domestic and international bank) participants that same day. It could also lead to the elimination of around one-third of that day's payments, or around one-third of $885 billion. Such an event clearly would have severely disrupted financial markets (Humphrey, 1986). While a failure to settle on either Fedwire or CHIPS has not yet occurred and is a low probability event, the expected loss--where the low probability of occurrence is multiplied by the very large loss which could occur--would still be high. Importantly, while the effects of a settlement failure are large, they can be significantly reduced at relatively low cost. These overdrafts and their associated credit risks are so large that it is useful to review them again. The Fedwire funds transfer daylight overdraft of $ 70 billion, when added to the CHIPS funds transfer daylight overdraft of $52 billion, gives a total of $122 billion in daily funds transfer overdrafts. This $122 billion plus the Fedwire securities transfer daylight overdraft of $100 billion brings the total overdraft to $220 billion each day. If overdrafts were not free, as they am today, payments participants would have to borrow $220 billion in extra working capital just to cover their payments exposures. - 57 - At a market interest rate of 6%, this would increase costs by $13 billion a year (or $53 million each working day) just to make large value payments. But, as discussed below, there are other, lower cost ways to reduce overdrafts and their associated risk on large value networks. Causes of the Daylight Oven/rah Problem. How did the U.S. daylight overdraft problem arise 7 First, corporations and other businesses responded to rising interest rates, no interest being paid on idle demand deposits, and the development of new financial instruments, by sending ever increasing values of payments over large value payment networks. As seen in Table 4-1, while the dollar value of one year's payments in 1960 was 7 times the value of that year's GNP, 25 years later (1985) a year's worth of payments was 66 times GNP. In 1990, the payments/GNP ratio was 106. Table 4-1 ments to GNP an Wire Transfer to Reserves (1960-1990, United States) 1960 7 0.6 1970 16 2 1980 48 17 1985 66 42 1990 106 80 This can be put into another perspective by noting that the "money supply" used to make large value payments is roughly equivalent to reserves held in reserve accounts with the Federal Reserve. In 1960 this "money supply" or payments working capital was larger than the value of all wire transfer payments. In fact, the value of wire transfers in 1960 was only 60% of the payments working capital balances held. By 1985, however, the value of CHIPS and Fedwire funds transfer payments were 42 times the value of reserve balances. By 1990 the wire transfer/reserves ratio was 80. In short, the growth of payments value far exceeded the growth in payments working capital and daylight overdrafts were the result. In addition to the exponential growth in payments volume, a second contributing cause of daylight overdrafts was the fact that the Federal Reserve automated its wire transfer operations without real-time balance controls. In the 1960s it would have been very expensive to provide real-time balance controls for all payments users--some 13,000 banks. - 58 - At that time, daylight overdrafts, either did not occur or were very rare. With hindsight, had real-time balance controls been put into place, then the daylight overdraft problem never would have arisen. The rapid growth in payments volume and the lack of real-time balance controls, along with the fact that until 1986 there were no restrictions on the levels of overdrafts banks could incur, all led to the adoption of institutional practices in financial markets which contributed to overdrafts. One important institutional practice which developed was a time gap between returning previously borrowed funds and the receipt of new funding for the next borrowing period. With no incentive to more closely match in time these gross payment flows during the day, it became customary to return previously borrowed funds--federal funds, Eurodollars, bank CDs, and corporate commercial paper--in the morning or at least prior to when the refunding process occurred that same day. Cut out this time gap, and much of the daylight overdraft problem goes away. A second institutional practice which developed was to write contracts for future delivery of foreign exchange (or FX) in terms of the gross amounts agreed upon. Even if the gross deliveries of FX between two parties at a given value date exactly offset each other, the legal liabilities and payments flows were still the gross amounts specified in the contracts. Although these gross payments flows exactly offset each other, they rarely would occur at the same time during the day. This can create an overdraft at the bank which sends payments first, giving daylight working capital to the receiving bank. In effect, the daylight overdraft creating time gap inherent in the current process can be eliminated if offsetting FX positions with the same counterparty are netted. This should be done both in terms of the payments flows and, importantly, in terms of the underlying contracts involved. In this way the legal liabilities are reduced along with the level of measured overdrafts. Factors Affecting the Distribution of Overdrafts Among Banks. There are other things which are often said to cause daylight overdrafts. Sometimes they can lead to a higher average level of overdrafts being measured but most of the time they merely alter the distribution of overdrafts among the various payments participants--increasing overdrafts at one bank and reducing them at another. Changes in the distribution of overdrafts among payments participants can occur with purposeful delays in sending payments, usually reserved for non-time-critical payments, along with computer outages and/or software problems at banks or the Federal Reserve. While these problems do occur somewhere in the payments system each day, they do not importantly contribute to the average level of overdrafts noted earlier. Current Risk Controls. The current system of risk controls includes monitoring and quantitative limits on daylight overdrafts. The Federal Reserve monitors the level of - 59 - interbank overdrafts and banks monitor this as well as the level of customer overdrafts. The quantitative overdraft limits apply to senders as well as receivers. Sender net debit caps limit the size of the daylight overdraft to be some multiple of bank equity capital. Higher multiples are obtained by having more comprehensive credit procedures and operational controls in place, as this can indicate a greater ability to prudently manage customer-related payments risk. Receivers of payments can control their risk as well through bilateral net credit limits. These limits are set on the basis of a formal credit judgement of how much daylight credit a receiving bank may wish to extend to any given sending institution. A tangible result of the current set of risk controls is that funds transfer overdrafts have been relatively stable even as the value of payments has grown by 30% a year. Because of this, the ratio of overdrafts to payments has fallen by 20%--from 10 cents worth of overdrafts for every $1 transferred down to around 8 cents per $1 today. Recent Feeer81 ReeeFve Ji&liey de~i,ioR& bave caorcibured ro rbis result, Fjrst there was a 25% reduction in the net debit cap on funds transfers. Second there was a $50 million limit placed on securities transfers. The $50 million limit on securities transfers directly affects the institutional practice of "position building". Assume that an insurance company has contracted to buy $200 million in 90 day U.S. government securities to be delivered today. In the past, the seller of securities would purchase and accumulate 90 day securities, purchased elsewhere or from his own inventory of holdings, until the entire $200 million could be delivered. At the time of delivery, the buyer would automatically transfer $200 million to the seller ("delivery against payment"). However, if only $190 million in securities are accumulated, nothing could be delivered to the buyer. This is called a securities fail. Even though the buyer had not yet paid for or received any securities, the seller would still have to transfer all the interest on $200 million in securities to the buyer because he had missed the contracted delivery date for the entire $200 million. To minimize the cost of a securities fail, securities sellers routinely purchased securities during the day until their largest contracts were filled first, and then delivered these securities against payment by buyers the same day. While position building reduced a security broker's fail costs, it also typically created a securities transfer overdraft for the large clearing banks that handle securities transfers in the U.S. The $50 million limit on securities transfers permitted earlier deliveries of part of the total securities contract, reducing position building, securities overdrafts, and the cost of a securities fail. - 60 - The quantitative limit on funds transfer net debits--the net debit caps--and the $50 million securities transfer limit could be reduced further in the future. However, the current policy direction is to price overdrafts. In effect, "tariffs" are to be used in place of a further reduction in "quotas". New Risk Controls. In an attempt to further reduce systemic risk on privately operated large value payment networks, CHIPS has recently set up a collateral arrangement. In the event of a failure to settle by one participant, the collateral would be liquidated and used to provide the needed settlement funds. In this case, one participant's failure to settle no longer leads to the failure to settle by other participants and systemic risk is virtually eliminated, unless two or more participants happen to fail to settle on the same day. Since all CHIPS participants have contributed to this collateral pool, they all have an interest in reducing risk on their network. This arrangement is not unusual. Indeed, position limits, collateralization, and indemnification agreements are common on U.S stock and commodity exchanges as methods used to reduce the risk of loss on these trading exchanges. Recently, the Bank for International Settlement has essentially adopted the methods used on CHIPS as recommended procedures to be implemented on privately operated large value networks operating in Europe. These concern overdraft monitoring, net debit caps, bilateral net credit limits, and collateralization as useful methods for reducing payments risk. As mentioned above, the pricing of daylight overdrafts was adopted and implemented in 1994 on Fedwire. Overdrafts clearly have value to banks and their customers as payments working capital. Pricing is more flexible than caps or credit limits since, if necessary, overdraft credit could be purchased. The initial Fedwire price will be at an annual rate of 25 BPs per dollar of overdraft, or about $425 million a year if total Fedwire funds transfer ( $ 70 billion) and securities transfer overdrafts ( $100 billion) remain at $170 billion a day. This rate is considerably below a market interest rate (6% or 600 Bps) but then we are dealing with daylight use of funds, not overnight use of funds. But even a price of 1 2 BPs could reduce overdrafts. According to participants in the federal funds market, it may take only an eighth of a percentage point--which is 12 basis points--to induce sellers of overnight federal funds and Eurodollars to adopt rollover or continuing contracts. These arrangements would replace existing overnight funds practices where overnight borrowings are repaid each day but create an overdraft because of the time gap between gross funds flows. Alternatively, the overdraft price may have to rise to 100 BPs. For example, the only existing intraday market regularly charges 100 BPs for an intraday loan to securities broker/dealers. This gives them daylight working capital to finance their purchase of securities prior to delivery to customers for final payment. - 61 - Another way to gauge the maximum likely intraday funds price would be to ask what is the insurance premium-equivalent rate for credit risk when banks incur daylight overdrafts. This rate is approximated by the long-run difference between uninsured 90 day bank CDs (which reflects bank-specific risk) and the 90 day U.S. treasury bill rate (which is "riskless"). Over 1978-1990, this spread averaged 84 BPs. One country--Switzerland--effectively charges an infinite price for daylight overdrafts since they are prohibited. Other nations--Japan, Sweden--only require that daylight overdrafts be zero at one or two settlement points during the day. Market Response to Overdraft Pricing. The market response to more expensive overdrafts would involve a search for lower cost ways to reduce overdrafts, rather than paying a price for them. One virtually zero cost method would be to identify and delay non- time-critical payments until covering funds arrive and these payments can be sent later in the day. Another relatively low cost way to avoid overdrafts would be to implement payments netting procedures for certain types of money market transactions. This reflects ways that the funding process can be restructured to reduce the need for daylight working capital, essentially by reducing or eliminating the time gaps associated with refunding operations. Finally, an intraday funds market could evolve where funds are purchased during the day to cover payments which would otherwise create overdrafts (as now occurs in Japan). This alternative, however, is likely to be somewhat more expensive than the first two. So, if an intraday funds market does develop, it will probably be a later development. Payments Netting. Daylight overdrafts are a common event today because there are only weak incentives to develop alternative payment arrangements which minimize them. Payments netting is one low cost alternative to daylight overdrafts but it involves a fundamental change in how certain payments are made. Today, payment transactions are based on gross payment legal obligations which, typically, lead to corresponding gross payment flows between two parties to a transaction. Netting would alter this institutional structure so that the underlying legal payment obligations are in net terms, leading to smaller net payment flows. Two results follow from the fact that payments netting can transform gross payments exposures into smaller net exposures. First, this can significantly lessen actual risk between banks since legal exposures are reduced. Second, since this directly translates into a reduced value of payment flows over wire transfer networks, payments netting should also - 62 - limit the size and growth of daylight overdrafts. This addresses the measured risk of interbank payments now monitored by the Federal Reserve and other bank regulatory agencies. While netting can reduce overdrafts associated with the transfer of funds or the transfer of U.S. government book-entry securities, our focus here is only on funds transfers. The application of payments netting to four types of funds transfers is shown in Table 4-2. The four funds transfer categories are: FX Spot and forward foreign exchange transactions; FF Overnight federal funds borrowings; Euro$ Overnight Eurodollar positions; and CP Commercial paper flows. According to a recent survey of large New York banks and some foreign banks in New York, these 4 categories of payments accounted for over 60% of the value of all funds transfers over CHIPS and Fedwire. It is possible to obtain some idea of the impact of payments netting by determining the value of net-able payments currently made over U.S. large value networks. While all payments are potentially net-able, some are more net-able than others, meaning that not all net-able payments can be netted at the same low cost. The four categories of net-able payments shown in Table 4-2 account for 83% of the total value of CHIPS payments and 62% of Fedwire payments. Most of these payments are pretty much network specific. That is, foreign exchange transactions virtually all go over CHIPS while federal funds and commercial paper transactions, when they involve a wire transfer, go over Fedwire. Eurodollar transactions, however, are made over both networks. - 63 - Table 4-2 How Important Are Net-able Payments? ( 1987, United States) ·. :-··. .:-:·-·:.:-:.-: -:.: Type of Paym~dt \ Foreign exchange (FX) 55% 0% Federal funds (FF) 0 42 Eurodollar (Euro $) 28 10 Commercial paper (CP) 0 10 Total 83 62 8 9 Source: Federal Reserve Bank of New York, 1987. The value of these 4 categories of net-able payments far exceeds the percent that overdrafts are of total payments. Overdrafts as a percent of payments are only between 8 and 9%,. as shown at the bottom of the table. Thus, if only a relatively small portion of total net-able payments were actually netted, one could reasonably expect that daylight overdrafts would be markedly reduced or eliminated. The most well-known netting arrangement is FX netting using netting by novation. The three salient features of FX netting are: ( 1) It is currently bilateral in nature; (2) The bilateral gross transactions flows are being continuously netted by legal agreement so that at any point in time the total exposure is only the net position; and (3) There is a single settlement payment for this net exposure on the value date. The bilateral nature of the transaction permits the counterparties full control over their credit exposure and so assists in limiting payment risk. - 64 - The ability to continuously net gross transactions effectively means that each new transaction is associated with a new contract for the new net amount due. Since the old contract is replaced each time a new gross transaction is initiated, this has been called "netting by novation". In FX netting by novation, one relatively small net payment replaces what would otherwise be a series of larger gross transfers over a large value network during the course of a day. This contributes to overdraft reduction since fewer payments would need to be made, and those that are made would be for a smaller dollar amount. A somewhat different arrangement would apply for netting of overnight federal funds, overnight Eurodollars, and commercial paper transactions. Here netting takes the form of reducing or eliminating the time gap between repayment of funds and their same-day refunding. For banks, these funding time gaps can be reduced or eliminated when federal funds or Eurodollar overnight borrowings are transformed into rollovers, continuing contracts, or term funds. For example, when federal funds are to be reborrowed from the same seller, it is really not necessary to create a daylight overdraft by returning the gross amount borrowed in the morning and waiting for the same value of reborrowed funds to arrive in the afternoon. If, instead, the funds were rolled over, the time gap between funds outflow and inflow could be in effect netted to zero. Similarly, when funds are to be reborrowed from the same seller but the gross reborrowings differ in amount between overnight periods, a continuing contract could be specified. Here only the net change in the funded position needs to be transferred, also reducing overdrafts. The net amount transferred could, of course, also include the previous day's interest earnings. The overnight rate paid could be renegotiated each morning. Corporations have many of the same opportunities as bankers, only their rollover or continuing contract arrangements would apply to commercial paper and perhaps other short-term borrowings. The opportunity to rollover commercial paper is especially favorable because, reportedly, some 90% of this short-term borrowing is re-lent to same corporation by the same lender every time the instrument reaches maturity. To summarize, payments netting arrangements likely comprise one of the lowest cost ways to reduce overdrafts and risk on large value payment networks. These involve ways to reduce or eliminate the time gap that now exists between repayment of a financial instrument at maturity and the refinancing of a new instrument on the same day. These procedures are complex and involve rollovers, continuing contracts, and netting by novation. Although different types of payments may be functioned over large value payment networks in other countries, they all (except in Switzerland) create the same sorts of overdraft problems - 65 - as exist in the U.S. Fortunately, the same sorts of solutions to overdrafts talked about here can also be applied in other countries as well. It is also useful, when setting up large value payment networks in other countries, to learn from mistakes made earlier. Importantly, steps need to be taken to address this possible overdraft risk problem in the initial stages of development of a large value payment network rather than try to address the problem when it has become entrenched, as is in the U.S. Some simple principles to follow in this regard are, first, to make sure, whatever the payment flows are, that they mirror the underlying legal agreements. For example, netting of payments is inappropriate if legal agreements are not also in net terms. Second, if one adopts netting, then bilateral netting is safest since this permits counterparties full control over credit exposure and the payment risks are clear to parities involved. If multilateral netting arrangements are used, then the resulting contracts, credit risks, and operational controls put io place shoyld afford a similar level of comfort as obtained with bilateral netting. Lessons Learned: How To Reduce Risk for Large Value Payment•. It is possible to derive some general principles from the payments risk problem on large value payment networks. The most important principle is that a free good is overused. Daylight overdrafts were free goods--they had a zero price and consequently were not restricted in their growth even though they posed a risk of loss for large value networks and to the financial markets which they served. Second, overdrafts clearly have value to payments participants--which are banks and their customers. In the U.S., overdrafts relieve banks and their customers of the need to borrow $220 billion in payments working capital on an intraday basis to make all their payments without incurring an overdraft. Third, in order to make proper payment decisions, users of large value payment networks need to balance the costs they pay with the benefits they receive. They can't do this if they continue to receive a subsidy from being given free overdraft credit which is not really "free" because of the risk of loss associated with providing overdraft credit in the first place. Hence, this subsidy should be removed and incentives provided to reduce or eliminate payments system risk. The goal is to have the direct beneficiaries of the payments network bear the full costs of running their system so that total payment costs are roughly equal to the total user benefits. Finally, there are relatively low cost solutions to the payments risk problem that involve a basic restructuring of the way certain types of payments are made. Network - 66 - participants will not make these changes without the proper monetary inducements and legal protection so these have to be put into place before payments risk can be reduced or eliminated. Of course, the easy solution would have been to have recognized this problem about 20 years ago in the U.S. and dealt with the need for payments restructuring before the volume of payments got out of hand. Thus proper design of a new large value network is crucial to prevent this problem from arising in newly developed networks. In this respect, the most direct solution to the large value payment risk problem on networks which currently have low volumes is to shift to a strictly credit-based arrangement when moving from a paper- based system to a telex or all electronic network, or when updating a telex system to a computer based system. In this regard, the credit-based SIC system in Switzerland, which does not permit any overdrafts, should be the goal. One should learn from the U.S. mistakes on Fedwire and CHIPS rather than repeat them. Chapter 5: SHORT-RUN IMPROVEMENTS TO A PAYMENTS SYSTEM Short-Run Improvements for Small Value Payments. Short-run improvements are those that can be done with relatively little effort and expense but yet can substantially improve payments efficiency. These improvements do not involve major changes in a country's reliance on an existing payments structure, only an attempt to improve ~he operation of the present structure. Cash. In a cash-based system, such as that which e,cists in many less developed nations and in the countries of the former Soviet Union, domestic inflation is a recurrent problem. While the optimal policy is to control inflation, often this has proved to be impossible and paper currency rapidly loses value as the denominations in circulation increasingly bear little resemblance to the prices being charged for goods and services. As a result, it takes a greater and greater amount of the existing currency to complete transactjons, The obvious solution is to replace the old, smaller denominations in circulation with new, larger denominations to make transactions easier. However, if inflation is extremely rapid, as it has been at times for certain countries, the expense of printing the new currency can exceed the real purchasing power of the currency's stated face value due to the lags involved in finally introducing the new currency into circulation. In this event, currency replacement will waste resources and is ultimately self-defeating as the benefits would be quite temporary. Extremely rapid inflation--hyper-inflation-··is most often associated with the printing of currency to finance large government deficits. Even though inflation results, printing money becomes an alternative, easier, and more practical and effective way to collect "taxes" to finance government expenditures. This occurs when the domestic tax system is ineffectively structured and/or is widely evaded (and illustrates an important area where other improvements should be made). In these situations domestic transactions typically shift away from the use of the inflated domestic currency and payments are increasingly completed through use of an alternative "currency" that maintains better its value. The English pound, the U.S. dollar, and precious metals, have all played the role of an alternative to domestic currency in times of rapid money supply growth. Although other national policies may be put at risk, such as the continued ability of the government to merely print domestic currency to finance expenditures, payments efficiency is increased if access to alternative currencies is made easier by removing exchange controls and other impediments to the physical importation of currencies of other nations with lower rates of domestic inflation and a recognized ability to hold their purchasing power. It is emphasized that such a substitution will occur anyway, through smuggling or other illegal activities, and all that is being suggested here is that the process be made legal, easier, and more effective. - 68 - Checks. In a payments system where checks play more than just a very minor role, one of the most cost-effective improvements that can be made is to improve the collection process. This often has two components. First, it is possible to lower risk and uncertainty in the use of checks by establishing local clearing houses (if they do not already exist) and by clarifying the rights and liabilities of the parties to a check transaction through revised, more enforceable, legal agreements and/or rule-making (which also guards better against fraud). As well, it often is helpful to establish realistic funds availability schedules so that payment delays can be reduced. Further, if the mails are relied upon to move large volumes of checks between payees and payers (and their clearing agents), as now exists in the former nations of the Soviet Union, the check collection process can be substantially improved by the judicious use of dedicated couriers (vans, trucks, commercial aircraft) between cities where the check volume is greatest. In countries with more mature check systems, electronic rather than physical delivery of check payment information can not only reduce transportation costs but can substantially reduce the perverse incentives created when payers gain from the float they create. In most countries, payers face incentives to delay the collection of checks while payees face incentives to waste real resources to physically present checks drawn on geographically remote points selected and controlled by payers. Check payment efficiency can also be improved by rule-making or interbank agreement to provide access to deposited funds on a same-day basis while charging payors for any float created in the check collection process. Note that these improvements, while making check payments more efficient, do not address the more fundamental problem of return items inherent in a debit-based check transfer system. Overall, the set of methods which could be most effectively used to improve a nation's check payment system is most easily determined by simply comparing the detailed operation of one's existing check system with that of Canada which has devised the most comprehensive, efficient, and equitable check system in the world today. While many aspects of the Canadian check system may not be transferrable to different countries in the short-run, many others--including those noted above--may be. Short-Run Improvements for Large Value Payments. The basic problems with large value payments are their accessibility and reliability, their safety, and the risks that can be created from daylight or overnight overdrafts of accounts of participants of such networks. In terms of accessibility, reliability, and security for countries with relatively small volumes of large value payments, often all that is needed is a switch to and/or an expansion of dedicated phone or telegraph lines and the addition of frequently changed passwords, use of call-back or message sequencing authentication procedures, and other security arrangements to improve telex-based large value payments. As well, if not already in place, access to facilities - 69 - which initiate and receive large value payments should themselves be made secure by making access to these facilities restricted. These changes can be very effective where current large value payments are made using a slow, cumbersome, and insecure paper-based system or there is a current reliance on insecure, common carrier, telex procedures. All of these improvements are possible, at relatively low cost, if a country's domestic large value system is reconfigured to correspond to and connect with SWIFT. This not only improves the safety and security of domestic large value payments through use of more secure machine encryption and authentication procedures, but also provides greater security and accessibility to international payments. Depending on the volume of transactions needed to be processed, the limited substitution of personal computer technology and established payment processing software can usefully be applied where large value transaction volumes are highest. This is most likely within and between money centers rather than in all places where large value transactions may occur. This type of improvement is currently being implemented in Russia and will likely secve as a gujde for similar improvements in other countries. The risk from daylight or overnight overdrafts on large value payment networks can be most easily addressed when payment volumes are relatively low. (When payment volumes are high, controlling this risk is much more difficult and it becomes a long-run problem.) With few large value payments, there is little disruption to economic activity if overdrafts were simply prohibited. Failing this, some sort of interbank or central bank credit facility can be made available to participants who otherwise would overdraw their account in making a large value payment. Use of a credit facility, however, should not be open-ended and an appropriate--or penalty--rate should be applied on all extensions of credit. If there is no charge for the credit extended, or if the rate applied is unrealistically low, overdrafts will be overused relative to their value to participants and greater risk will exist on the large value network than is necessary to have it function properly. Chapter 6: LONG-RUN IMPROVEMENTS TO A PAYMENTS SYSTEM The Importance of User Need When Considering Major Changes in Payments. Long-run improvements to a payment system are those that lead to major changes in how existing payment instruments are used or to the development of newer, more efficient instruments for making payments. Such changes can be expensive, time-consuming, and difficult to implement. There is also a substantial element of risk since, to be effective, significant changes to a nation's payment system need to be accompanied by strong incentives and protections for users in order to be adopted. While the benefits of changing from one payment instrument, say cash, to another, say GIROs, may be obvious to the "experts", these benefits need not motivate many users to change unless the instrument they now use is clearly deficient in some way. If only a small net benefit accrues to the user--or worse, the supplier of the new payment service prices it in such a manner to retain virtually all of the benefit for themselves--change will come about primarily through the turnover of the user population. Larger proportions of each successive generation, not being tied to the habit of relying on the "old" way of making payments and necessarily more open to experiment, can be expected to shift to the (marginally) improved payment method or instrument. While change will occur, it may be exceedingly slow. Thus it is quite important that planned long- term changes to a nation's payment system be structured so that they solve some existing problem with the current system and therefore have a clear benefit to the intended user. Too often, payment instruments that are technologically feasible are also assumed to be desired by the user. To be successful, long run changes to a payment system should emphasize user needs and make sure that incentives for change exist. Predictions of an imminent "cashless society" in developed countries, over the last 20 years, is but one example of confusing "what is possible" with "what is needed" and neglecting to properly structure incentives--pricing and legal changes--necessary to make this come about. Long-Run Improvements for Small Value Payments. There are at least three major long-run issues for small value payments. These are: (1) the substitution of other payment instruments for use of cash; (2) the use of electronic payment methods in place of noncash paper-based methods; and (3), the shift from debit transfers to credit transfers. The Replacement of Cash Payments. For most developing countries and for the nations of the former Soviet Union, virtually all small value payments are made with cash. However, for many bill payments and small value business-to-business transactions, payments efficiency may be substantially improved and the possibility of theft and loss reduced if a more convenient, safer, and timely alternative to cash payments can be provided. The intensive use of cash keeps large amounts of currency in circulation. While - 71 - this benefits the government through seigniorage (e.g., printing money to cover expenditures rather than borrowing), it also reduces an important source of domestic savings, lowering the availability of deposit balances for the banking system to intermediate into working capital and other loans which promote economic growth. As well, the intensive use of cash hinders the development of financial markets and a more efficient distribution of loanable funds outside of the banking system. Historically, the two most favored cash substitutes have been checks and GIROs. The choice between these two alternatives is essentially a choice between a debit transfer and a credit transfer system. In part, this choice is affected by the degree to which check or GIRO payments can be made electronic, rather than paper-based. These issues are covered next. Checks: Paper Versus Electronics. With a paper-based check system, the existing legal structure typically requires that a check be physically presented to the paying bel'IIE eefeFe pav~aot is made Ibe reasonjng is that the paying bank will need first to determine the authenticity of the payer's signature and then determine if there are sufficient funds in the payer's account to cover the debit. However, signatures are normally examined only when the value of the check is above some predetermined cutoff amount. 21 And, when funds in a payer's account are insufficient, a prearranged and automatic account overdraft loan can be extended (for a fee) so the check need not be returned unpaid. In practice, then, it would be possible to capture the magnetic ink payment information from the face of the check, keep the paper check at the bank of first deposit--the payee's bank, and present the payment information electronically to the payer bank for all but the very largest value checks where the signature may still need to be verified before the item is paid. This type of partial substitution of electronics for paper is called check truncation and has been discussed for some time now. While small pilot operations of check truncation have been tested, industry agreement necessary for wide-scale implementation has not been forthcoming. This is due, in part, because many check writers prefer to receive each month their canceled checks (even though a microfilm copy of the check is now a legally acceptable alternative for proof of payment). Nonagreement by the industry also reflects the fear that by speeding up the check collection process, business payors--who create most check float--will lose their float benefit and thus will take their disbursement business to banks who do not agree to accept and pay items presented electronically. A more far-reaching way in which electronics can replace paper in a check- based system involves the use of a debit card. Current technology makes it possible to W Indeed, checks without signatures below the value cutoff are routinely paid since check reader/sorters only read the magnetic ink encoded value of the check and the payor/payee account and bank numbers, not the signature. I - 72 - determine in real time (a) if funds are available in a payer's transaction account and (b) to debit that account at the same time the transaction occurs. Since the transaction is not completed if funds at that time are insufficient to cover the debit transfer, a debit card arrangement effectively works like a GIRO. With a debit card, a debit transfer check-based system is transformed into a credit transfer GIRO type of system. Such a transition is slowly occurring in the U.S. It is slow because (again) payors have no incentive to give up their current float benefit. That is, payments by debit card are, considering float, more expensive to the user than payment by check, even though from a social standpoint the real resource costs can be lower (see Chapter 2 and Table 2-5). In sum, the increased use of electronics in an existing paper-based check system can represent a way to eliminate the two most serious problems associated with checks--the costly return item (also leading to uncertain finality of payment) and the float benefit to payors (which provides an incentive for payors to delay collection and for payees to pay to speed it up, wasting real resources in the process). While the substitution of electronics for paper within an existing check-based system can mitigate or even eliminate the inherent problems associated with paper checks, the legal structure governing check payments typically needs to be altered to accommodate this change and incentives structured to induce users to change as well. If this is not done, then the problems with checks will remain. Debit Transfer (Checks) Versus Credit Transfer (GIRO). Because the return item and payor float problems associated with a debit transfer check-based payment system are eliminated with a credit transfer GIRO-based system, the latter is clearly preferred to the former. In this respect, the primary way to improve long-term payments efficiency is to follow Europe's lead--and that of Japan's Zengin System--and develop a credit transfer GIRO network for small value payments, especially for bill payments and business-to-business small value transactions in an essentially cash-based system. A GIRO network (or Zengin System in Japan) provides greater finality of payment, since there are no return items, and has a lower (user plus supplier) cost per transaction. Transaction costs are lower because most collection and presentment is electronic, not physical, and because the elimination of payor float removes the incentive for costly remote disbursement and expedited collection. Although the development of the GIRO is associated with Austria's Post Office Savings Bank in the late 1800s and the development of other Western European postal GIROs in the first quarter of the 1900s, long after the use of checks had become widespread in some countries fifty years before, there is no historical determinism to payment system development. That is, there is no particular reason for a currently cash-based nation to have - 73 - to first pass through a stage of check use prior to developing a GIFIO system. 22 The apparent historical evolution of cash, to checks, to GIROs, if it is happenin!g at all, is occurring (slowly) in geographically large countries with a disaggregated banking system where a centralized postal service was not the primary supplier of retail savings accounts. In comparing countries with well-developed GIRO systems with other countries with well-developed check systems, it is evident that a key difference has been whether a single, centralized entity or a set independent, disaggregated entities initially was the main supplier retail savings services. If a single, centralized entity such as the postal service is the main supplier, it is but an easy step to provide GIRO payment services to account holders. While GIRO services are now offered by groups of independent banks, this has occurred only recently, and then only because banks needed to expand their base of low cost deposits to remain competitive in the loan market. This need was sufficiently strong to overcome banks' aversion to developing a joint venture with competitors that would cannibalize portions of their check business. If postal GIROs had not developed in these countries, it is very doubtful that banks would have worked together to provide GIRO services on their own; instead, payments would be check-based using the t>a1tklr1g s9ste111 as they 11000 ere i" eeH"t,iee whe,e oaRk& face little or no competition in providing payment services to consumers. Thus there is nothing immutable about the development of a nation's payment system. Checks need not be an important part of any nation's payment structure and, in terms of long-run improvements, the development or expansion of a GIRO network is clearly preferred over other small value payment methods. Long-Run Improvements for Large Value Payments. At least four issues arise when considering long-run improvements for large value payments. These issues determine the essential features of a large value payment system and concern: ( 1) using funds transfer versus message transfer systems; (2) using gross settlement versus net settlement; (3) providing intraday credit versus prohibiting overdrafts; and (4) determining the level of payments finality--sender, settlement, or receiver finality. Funds Transfer Versus Message Transfer Systems. A funds transfer system is an interbank payment network where a centralized account of one bank is debited and the account of another bank is credited. Fedwire and CHIPS in the U.S., BOJ-NET in Japan, CHAPS in England, and SIC in Switzerland are all funds transfer networks made up of bank participants with the central bank debiting and crediting the participant bank accounts. Each transaction involves a movement of funds from a customer account at one bank to a customer account at another bank. A message transfer system achieves the same end--namely funds W A number of countries which rely heavily on cash payments (Aumania, Hungary) are attempting to set-up GIRO systems and "skip" the significant use of checks altogether. - 74 - are transferred to a customer's account at a bank--but they come directly from another account at the same bank. SWIFT is a message transfer system which relies on reciprocal holdings of interbank correspondent accounts to transfer funds between accounts at the same bank. Of course, to establish and maintain these reciprocal correspondent balances, funds have to be transferred between banks, a process best handled over a funds transfer network. For very small volumes of large value transactions, secure telex message transfer systems are appropriate. For somewhat larger volumes, a message transfer network like SWIFT is more than adequate. But when there are large numbers of participants, and very large volumes of transactions, the reciprocal balance requirements of a message transfer system become much more expensive to maintain and monitor. At this point, a funds transfer system can usually lower transactions costs over a message transfer system. Thus the choice of one system over the other should be based on the long-term expected number of bank participants and their expected transaction volumes. Because transaction volumes among countries internationally are markedly lower than the volume of domestic transactions, and because an international central bank with an acceptable international unit of account does not yet exist, message transfer systems are used in cross-border transactions even when a funds transfer system is used domestically. Gross Versus Net Settlement. On a gross settlement network, each payment is separately settled at the time it is sent. Fedwire, SIC, and a separate payment service available through BOJ-NET are gross settlement networks and payments are final when made. On a net settlement network, each payment sent and received among bank participants is cumulated and only the resulting net position between banks is settled at a particular time of the day; only then are all the prior payments final. In contrast to a gross settlement network where settlement is continuous and there are as many settlements as there are payments, there are only one or two settlements each day on a net settlement network. CHIPS, CHAPS, and BOJ-NET are all net settlement networks (BIS, May 1990; Borio and Van den Bergh, 1993). If customers receiving payments over a net settlement network did not use them until after settlement occurred, then the risk involved in sending payments over gross and net settlement networks would be the same. But since net settlement often occurs at the end of the business day, when overnight money markets are closed, waiting to use funds until after settlement occurs means the loss of one day's float. Because the values involved on these networks are large, the opportunity cost of this lost float can be substantial. Thus it is common for customers to use funds received over net settlement networks prior to end- of-day settlement. If settlement does not occur for some reason, these funds would need to be returned--from the receiving customer to the receiving bank and, possibly, from the receiving bank to the sending bank. Because such a settlement failure can be quite disruptive, - 75 - procedures have been implemented to attempt to provide funds for end-of-day net settlement even if a participant is unable to settle his net debit position. The primary method of "guaranteeing" against a settlement failure on a net settlement network is for all participants to post collateral that as a total is sufficient to cover any possible failed net debit position (with controls in place to make sure that the net debit of any participant can not exceed the value of the posted collateral). As a result, the risks of settlement failure, and the corresponding economic and financial market disruption which could follow, have been pretty much eliminated. Thus, if not already in place, a substantial improvement in a country's net settlement network can be obtained by implementing the Bank for International Settlement's recommendation that collateral be posted and other net debit controls be adopted to reduce the systemic risk of settlement failure. But even if this is done, there remains the issue of daylight overdrafts and whether or not they should exist. lntraday Credit Versus No Overdrafts. Daylight overdrafts can not be prevented n a net settlement system. This is because a net settlement network debits and credits payment flows to zero balance accounts. While t e sum o ne e I s w, a net credits, at any point in time some participant will have a net debit and thus be in overdraft. This need not be the case on a gross settlement network, sincu balances held may be sufficient to cover debits to accounts or rules may be in place to reject or pend any transaction that would result in an overdraft. Which system is "best" depends on the cost of avoiding systemic risk and whether the expense of covering a settlement failure should be shared or borne by the participant who may fail to settle. The lowest cost method of avoiding systemic risk is to hold a zero balance and for all participants to contribute interest earning collateral so that the total contributed is sufficient to cover any one participant's largest end-of-day net debit. This shared risk of loss, which minimizes the total collateral tied up in this way, is the method CHIPS has chosen to deal with systemic risk. Overdrafts occur but they carry no charge. The next least costly way to avoid systemic risk is to hold a zero balance but have each participant post interest earning collateral sufficient to cover his own maximum net debit. Here there is no shared risk in losing collateral but the total collateral held can be quite large, perhaps hindering the liquidity needs of the participants separately or as a group. Thus the two lowest cost methods involve net settlement networks. The cost of avoiding systemic risk on a gross settlement network will be higher than that on a net settlement network when positive idle (non-interest earning) balances, not zero balances, are being held. On Fedwire, when these balances are insufficient, a daylight overdraft occurs and a fee of 25 basis points (annual rate) is assessed. Even more costly would be a gross settlement network like SIC where either higher idle balances would have - 76 - to be held--because no daylight overdrafts are permitted--or payments would be rejected. As seen, there is no sharing of the risks of settlement failure on either of these gross settlement networks. In sum, the cheapest solution to the systemic risk of daylight overdrafts is a net settlement network with shared collateral to cover a settlement failure, the solution adopted by CHIPS. If sharing the risk of loss of collateral is not desired, then a net settlement network where each participant posts his own collateral to cover his own net debit is the next cheapest alternative. But, if either of these net settlement arrangements are cheaper than current gross settlement systems, why do we still have gross settlement networks? The reason is simple. Because the use of collateral to address systemic risk on zero balance net settlement systems is a very recent development, it is too early to tell if one system will replace the other when both exist within the same country (as in the U.S.). In addition, since in many countries idle balances are required to be held with the central bank for purposes of monetary control, these same balances can be used to clear interbank payments on existing gross settlement networks. If there were no preexisting centralized reserve accounts for a gross settlement network to use in this manner, then the costs of a gross settlement system would rise considerably when such accounts were set up. But the very same problem confronts a net settlement system since centralized settlement balances are a virtual necessity in order to settle many transactions for large numbers of participants in a cost-effective manner. Thus, as long as reserve accounts exist for monetary control purposes, existing gross settlement networks are likely to continue in operation. 23 Ssndsr, Ssttlsmsnt, or Rscsivsr Finality. As noted in Chapter 4, all gross and net settlement networks provide for sender finality: payment instructions by the sender, once the payment is sent, can not be altered or reversed by the sender. In addition, gross settlement networks also have settlement finality, as do net settlement networks where posted collateral is sufficient to cover the net debit of a participant who may fail to settle. Settlement finality is an assurance that the payments made will be settled at settlement time, so there are no settlement "unwinds" or reversals. No network has adopted receiver finality, so payments received by a bank and credited to a customer's account could be reversed by the receiving bank. Banks receiving payments have not yet given up their right of recourse to the funds transferred to the receiving customer. This is because receiving banks may wish to cover possible customer loan def au Its or preexisting customer account overdrafts with the funds received, rather than pass on good and final funds to the customer and become a general creditor on the customer's defaulted loan or account overdraft. ll' The CHIPS net settlement network arose primarily because the Federal Reserve System, which runs the Fedwire gross settlement network, does not handle international transactions. Over time, CHIPS expanded into domestic funds transfers in competition with Fedwire. - 77 - At a minimum, new and existing gross and net settlement networks should include sender and settlement finality. While receiver finality would also be desirable, and reduce further the payment uncertainty of receiving customers, it has not proved necessary. Indeed, the adoption of receiver finality is resisted by receiving banks. As no networks now have receiver finality, there is no "model agreement" to follow that would indicate the tradeoffs necessary to adopt receiver finality on a large value payment system. Chapter 7: RAPID ECONOMIC CHANGE AND THE PAYMENT SVSTEM 24 This chapter outlines the serious difficulties that rapid economic change can create for the payment system. Our focus is on Russia, whose economy has undergone rapid and radical economic restructuring. While not all the experiences of Russia will be duplicated in other countries that are currently shifting from a command to a market economy, many of the more important payment system problems created during such a transition do reflect general problems and thus are generic to the process. The chapter is composed of two parts. The first part, written by Bruce Summers, gives a broad overview of the payment system problems faced in Russia, and indeed in any country shifting from a command to a market environment. In the second part, these effects on the payment system are shown to have unexpected implications for the money supply and the ability to accurately gauge the impact of monetary policy. The Russian Payment System and the Transition to a Market Economy. 25 The Russian economy is making one of the most difficult transitions imaginable, from a monopolized and politicized command system to a market economy. A necessary condition for this transition to be successful is the development of market-orientated banking and financial markets, including markets for money and credit. Such markets, of course, require an efficient payment system in order to operate properly. Our emphasis is thus on the development in Russia of the financial institutions and institutional relationships that underlie such a payment system, concentrating on domestic, ruble-denominated payments. Because the financial and payment systems of all the republics of the Commonwealth of Independent States (CIS) reflect the structure of banking in the former Soviet Union, most of our discussion of the Russian system will be useful in understanding conditions in many of the other CIS republics. The Role of the Payment System in Russia. It is useful to first outline the basic stages of payment system development that characterize a functioning market economy and to show where Russia currently is in this sequence. The four basic stages of payment system development shown below represent different ways to carry out a transaction or market exchange which, at any given time, might be supported by more than one of these methods: ll' Written by Bruce Summers, Senior Vice President, Federal Reserve Bank of Richmond, Richmond, VA. W An earlier version of Summers' contribution to this chapter was titled "Russian Payment Institutions and the Medium of Exchange Function of the Ruble" and presented at the Conference on Institutional Economics and the Transition to a Market Economy in Russia, held at the University of Wisconsin October 8-9, 1992, in conjunction with Moscow State University. - 79 - 1. Barter; 2. Cash (physical money); 3. Deposit money; and 4. Deposit money combined with a credit system. Russia is somewhere between stages two and three--between cash and deposit money. There are, however, clear signs of falling back to stage one--barter--as more sophisticated and efficient forms of payment become unreliable or unduly risky. The use of barter is especially prevalent for inter-enterprise transactions between firms that are separated by large distances and that have difficulty in obtaining working capital (credit) from the banking system. The structure of manufacturing in Russia is characterized by the geographic dispersion of plants, which is a result of politically motivated decisions to segment and distribute integrally related manufacturing facilities throughout the vast nation. In this regard, the highly dispersed structure of manutactunng in the Soviet Union Nas now left ma119 interdependent enterprises scattered among different CIS republics. This dispersion has resulted in complex inter-republic payment problems where different currencies as well as cumbersome payment arrangements exist simultaneously. The challenge in Russia is to move ahead into more advanced and efficient forms of payment to support the development of markets and, in particular, to help address the severe working capital shortages faced by enterprises. In a well developed money and credit market, the payment system serves as an important mechanism for delivering working capital to firms which draw upon liquidity from commercial banks to meet their trade obligations. This working capital is channeled to the firms by banks as part of the process of honoring payment orders when money is not immediately available in the firms' accounts with the banks to fund their payments. Without this credit mechanism an essential tool for developing the money and credit markets is absent. A simple example will illustrate the connection of payment and credit mechanisms to the real side of the Russian economy. Take the case of the consumer durable goods sector, which will be an early area of expansion in the Russian economy. As things now stand, with cash being the only widely accepted form of payment in the retail economy, a consumer wishing to buy durable goods from a retail outlet will find it difficult to pay for large value purchases in a timely, convenient, and safe way. An uneven and uncertain "cash flow" for the retail store will make it difficult for the store to manage its working capital position and to maintain inventory at efficient levels. In turn, manufacturers supplying the retail outlet will have difficulty managing their production lines, raw materials inventories, and work forces. - 80 - Conceivably, the retail store and manufacturer may depend on short-term bank credit for working capital. The banks themselves, however, will have difficulty managing funding in the interbank market if the timing and reliability of interbank settlement is questionable, which indeed it currently is. Thus, liquidity problems for banks, manufacturers, and distributors all cumulatively add to the cost and uncertainty of doing business in a market economy, lowering the rate of growth from what it otherwise could be. In most Western countries, public policy responsibility for the payment system usually rests with the central bank. Central banks are naturally interested in the payment system because it is a key component of the operation of financial markets and has important implications for the trading efficiency of the real economy. Moreover, the payment system is one of the first places where financial stress will manifest itself, as firms in financial difficulty may be unable to meet their payment obligations. Of particular concern to central banks is systemic payment system risk, which arises when financial stress makes it difficult or impossible for one or more market participants with large payment obligations to meet these obligations effectively. Systemic risk concerns are greatest in financial markets where trading results in large payment obligations and in the interbank markets, such as the stock, commodity, money, and foreign exchange markets. Thus, a safe and efficient payment system is a necessary condition for the development of an efficient market economy. In the case of Russia, the objective should be to put in place a payment system based on safe and reliable transfer of deposit money combined with a mechanism for allocating commercial bank credit. In terms of the model described above, this would entail moving to the third and fourth stages of payment system development. The Payment System Under State Socialism. The medium of exchange function of the ruble for inter-enterprise transactions was not important under state socialism. Production and the physical flow of goods between enterprises were governed by the state plan. The financial component of the physical flow was simply a record keeping procedure determined by the product of planned physical units times their respective administered prices. The former Soviet Union had a monobank system, where the public's savings were channeled to the Savings Bank, which was owned and controlled by the Gosbank. The Gosbank further channeled the public's savings to enterprises requiring credit under the state plan. Within the parameters of the physical plan, credit was automatic and payments were assured through the monobanking system. Because of this, enterprises did not need to have a speedy payment system, nor was it necessary to assess the creditworthiness of other enterprises to which goods were sent or from which goods were received. - 81 - Accordingly, the payment system in countries like Russia is underdeveloped, not only in technological terms but in terms of the institutions and institutional relationships that respond to the needs of market participants. In a market economy, the payment system must reflect the interconnectedness and time criticality of economic relationships, which was unnecessary under state socialism. Moreover, the providers of payment services, especially commercial banks, must be prepared to manage the credit risk which arises in connection with meeting the payment and credit needs of clients. Accordingly, Russia faces a large institution building challenge with respect to its payment system. The Russian payment system of today is based on three types of payment instruments: cash; payment orders; and checks. Another instrument, the payment demand order, was traditionally the most heavily used instrument for transferring funds among enterprises. Payment demand orders, however, were forbidden in July 1992 by the Central Bank of Russia (CBR). Cash, and to a much lesser extent checks, are used for retail transactions. Payment orders are used primarily for wholesale transactions involving Retail Payments: Cash and Checks. Almost all retail payments are made using cash. A rough estimate is that only perhaps a few percent of all retail payments are made by check. While cash has the virtue of being a final payment instrument, vulnerability to theft, unsuitability for making payments over long distances, and awkwardness for large-value transactions, are serious drawbacks. For the Russian retail economy these drawbacks have not been insurmountable because large-ruble payments are rare and consumer markets are essentially local (one does not shop by mail order in Russia). Nonetheless, reliance on cash payment has caused inconvenience for Russian consumers. For example, waiting in payroll lines to receive cash and then in innumerable additional lines to pay for goods and services is common. 26 Even though checks have the potential to offer significant conveniences to Russian consumers, they are not widely used. The low usage of checks is explained in part by the unattractive design of checking services and also by the inefficient check collection system . Historically, even if consumers would have preferred to pay with checks, businesses have been reluctant to accept them. This is due to the slowness of the check clearing process. Businesses might have to wait weeks to receive value (final payment) for a check submitted for collection. One form of check available to the consumer is analogous to a certified check ~, There is a form of recurring payment service available to individuals for payment of communal services such as rent and utilities through the Savings Bank of Russia, and this is fairly widely used. - 82 - in the United States. To use this type of check for a purchase, the exact amount of the purchase must be filled in on the check by the consumer's bank. Clearly, this is not a flexible or convenient instrument. A second type of check comes in a book that has a preauthorized ruble limit placed on each check and on the total number of all the checks in the book. Subscribers to this type of service, called the "Russia check," are required to set aside in advance money totaling the value of checks contained in the book. These checks cannot be used for transactions over a certain value; they are clearly designed to reduce the risk of overdrafts in demand accounts, a type of risk that is accentuated in Russia because of the long time it takes to collect a check. Initially, use of the Russia check increased rapidly because businesses were able to receive immediate credit on checks they deposited with collecting banks which, in turn, received immediate credit from the CBR. Serious problems with counterfeit instruments, however, have led to significant losses and the CBR has placed severe restrictions on the use of the new checks, limiting their use to local areas. Accordingly, even this new instrument has not succeeded in giving consumers and businesses an effactive new payment vehicle. Wholesale Payments: Debit and Credit-Based Payment Orders. Payment demand orders and payment orders have been used for inter-enterprise or wholesale payments. The payment demand order, which the CBR eliminated in July 1992, was a debit- based payment instrument. A payee (seller of goods) would initiate a payment demand order by sending it to the bank of the payor (buyer of goods). The payor's bank would initiate a transfer of funds to the payee if no objection was raised by the payor in response to receipt of the order to pay after a certain amount of time. The payment demand order is similar to a corporate debit transaction in the U.S. ACH system, except that the ACH is based on a preauthorization agreement between the payor and payee. The primary concern leading to the CBR's regulatory prohibition against continued use of the payment demand order was a lack of attention by businesses to inter-enterprise counterparty credit risk created when the guarantees for inter-enterprise trade debt, which existed under state socialism, were effectively removed. The CBA, consistent with the prevailing view in market economies, believes that debit instruments are inherently more risky than credit-based instruments and for this reason has forbidden their use. A payment order is a credit-based payment instrument which must be initiated by the payer. The process of effecting payment with a payment order is virtually identical to that of using a payment demand order, once the instrument is accepted, only that the funds move to the payee on the instruction of the payor (not the payee). The payment order thus functions like a wire transfer of funds or a GIRO. Settlement ofPayments. The CBR maintains a network of approximately 1,400 cash settlement centers (CSC) which perform three essential payment functions. First, the - 83 - CSCs maintain accounts, primarily for banks, through which settlements for payments are made. In this regard, and in keeping with most former monobank systems, every branch of every bank maintains an account at its local CSC. Needless to say, this system does not encourage consolidated cash management on the part of banking organizations and it almost certainly results in inefficient use of bank reserves. Second, the CSCs physically process payment documents and thus also serve as clearing centers for payment orders. Finally, CSCs also play a role in the distribution of currency. Electronic Payments. Payment by electronic methods is not accepted in Russia. All payments must be in the form of paper, because physical authentication by means of signature is the only legally acceptable method of authentication of a payment. The legal requirement of physical authentication, and the predilection for maintaining precise and repetitious control over document flows, has resulted in Russia's payment system becoming paper-bound. Current thinking about movement to payment by electronic methods still often involves maintaining an elaborate paper trail. Impediments To Payment System Development. There are at least five serious impediments to the development of an efficient payment system in Russia. Briefly, these are: 1. legal/cultural resistance to more efficient processing methods involving less paper and more electronics; 2. Insufficient recognition of the "time value of money;" 3. Cumbersome accounting relationships between the central bank and commercial banks; 4. Limited technical, personnel, and financial resources; and 5. Insensitivity to customer needs in designing changes to the payment system. The first impediment was just discussed above. The second impediment-- insufficient recognition of the importance of the time value of money--is evident in the continued use of the Russian mail system to move payment instruments between locations, leading to very long and variable transportation delays for funds transfers, unnecessarily raising float cost. Added to this is the third impediment concerning central bank and commercial bank payment accounting procedures. The CBR's current accounting policy for payments results in debiting the account of the paying bank when a payment order is made but delaying crediting of the account of the receiving bank until the payment order arrives at the receiving CSC through the public mail. These procedures undermine the reliability of the payment process and remove large amounts of liquidity from the banking system due to the - 84 - large values of transactions which remain constantly "in the process of collection." One result is the need on the part of participants in the payment system, including banks and enterprises, to maintain high levels of balances in accounts used to settle payments. In short, excessive amounts of "payments working capital" are now required to make inter-enterprise transactions and payment system float is a significant cost. In addition, Russia faces serious resource constraints to establishing an efficient payment system. There is less concern about the technical and financial constraints at this stage of payment system development since Russia has sophisticated communications technologies and has access to modern computing technologies. As such, the challenge is to redirect these technical resources to the banking system, a process which has already begun. In the area of human resource constraints, the limitation is more subtle: namely, the focus in making payment policy decisions is on the narrow technical aspects of payment system operation not on how to design and develop payment services that address the range of payment and credit needs of users. In a market economy, payment services offered by commercial banks are a profit center. Accordingly, banks have the incentive to supply payment services which meet the needs of their customers. As market incentives increase in Russia, there needs to be a concerted effort to promote effective market research to determine better customer preferences for different types of payment services before major payment policy decisions are made. The technocrats who are making these decisions today need to be supplemented by others who can determine and then effectively represent customer preferences in this process. Plans for Payment System Improvement. As a legacy of the past, the CBR plays a dominant role in operating the Russian payment system. It has a significant amount of power and influence over how payments are handled as a result of both its operational presence and its regulatory authority. Moreover, control over access to and use of central bank accounts for the settlement of interbank transactions gives the CBR (as it does every other central bank) a powerful means by which to influence private clearing arrangements. Accordingly, the CBR is the single most important force in the Russian payment system today, and it is likely to remain so for some time to come. The CBR's plans for improving the payment system are to: encourage the use of checks; improve payment orders; plan for electronic payments; and develop same-day settlement for large-value ruble transfers. Encourage Check Use. With respect to retail payments and as described earlier, the CBR, through its regulatory powers and operational control over the payment system, would like to encourage the use of checks. The new Russia check has been introduced and the CBR has printed, for distribution to commercial banks, tens of millions of these new - 85 - checks. Vulnerability to fraud will continue to plague this new instrument and its widespread use will be restricted until some basic improvements are made in the check system, however. One important improvement that will help prevent fraud is speeding up the check clearing process. Another is clarification of legal responsibility for fraud risk so that merchants become more careful in determining the authenticity of checks. Improve Payment Orders and Plan for Electronic Payments. As a short-term improvement, a new standardized payment order form based on the Eurocheck design has been developed. The benefit of this redesign is that it incorporates document and format standards that facilitate efficient handling and supports optical character recognition (OCR). In fact, experiments have been conducted, using optical scanning techniques, to convert this paper payment order information to electronic form for subsequent electronic handling. Changes in Russian law to permit electronic payments are also being readied. And, for the long-term, the CBR is developing a fully electronic processing system for funds transfers. This system would handle electronic input, drawn from the optical scanners that would read paper electronic payment orders in the future. Develop a Same-Day, Large-Value Transfer System for Rubles. A core component of most developed nations' banking and financial structure and its payment system is a large-value transfer system which provides same-day settlement with finality. Finality is achieved through an irrevocable and unconditional transfer of funds which discharges the obligation to make a payment. Final payments made by the irrevocable transfer of balances held in accounts at the central bank are said to be made in "central bank money." Because balances held with the central bank are free of credit risk, settlement using these balances is the surest form of final payment (other than using cash). A same-day settlement system with finality for ruble payments is important in Russia for at least four reasons. First, by allowing large blocks of money to be transferred reliably and safely on the same day in response to the demand for and supply of liquid assets, the very serious liquidity problems facing Russia's domestic financial system will be eased. Second, participants in the interbank funds and corporate deposit markets, which exist on organized exchanges and are especially active over-the-counter, will be able to shorten the time gap between their trades and settlement, providing a foundation for the development of new types of money market instruments which require speedier settlement, to reduce counterparty default risks. Third, and related to the second point, final payment will contribute to the development of the government securities market. Altogether, these developments will serve to increase the public's confidence in the banking system and reduce the need to maintain large balances of payments working capital. - 86 - The importance of a large-value ruble transfer system for easing serious liquidity problems in the Russian economy cannot be overemphasized. The liquidity problems arising in Russia are related in part to the operation of the payment system and specifically result from payment system float. Payment system float is the effect on the balance sheets of financial institutions of crediting {in the case of a debit-based instrument like a check) or debiting {in the case of a credit-based instrument like a payment order) the bank account of the entity originating the payment before the offsetting entry is made to the account of the counterparty in the transaction. In Russia, since most noncash payments are payment orders, where a debit is made to the account of the payor before the offsetting credit is made to the account of the payee, significant credit float has built up on the balance sheet of the banking system. When this credit float shows up on the balance sheet of the CBR, there is a corresponding decrease in reserves of the commercial banking system, and therefore a reduction in total liquidity. In addition to addressing some of the liquidity problems in the Russian financial system, a same-day settlement network with finality for ruble payments should make a significant contribution to the integrity of the payment and financial system. There has been a virtual explosion in the number of commercial banks doing business in Russia, from only a handful several years ago to several thousand today, as a result of extremely loose bank chartering requirements. As well, it is difficult to perform financial analyses to support counterparty credit evaluations because of the inscrutability of financial reporting standards in Russia. Consequently, evaluation of counterparty credit risk, even for banks, is very, very difficult. A final payment mechanism for ruble payments would therefore significantly increase the integrity of the interbank and other financial markets. Importantly, the CBR will take on a large burden in operating a same-day settlement system with finality. This is because it intends to stand behind all payments it has agreed to process on this system and as well could decide to provide some intraday credit to commercial banks (a common practice in many Western large-value transfer systems). Therefore, while a same-day settlement system with finality for ruble payments is essential, the CBR must be exceptionally careful to also adopt the credit and operational controls needed to allow it to manage its credit risk vis-a-vis the banks using the system. Improvement Efforts by Private Clearing Organizations. Plans for improvement in the payment system are not confined to the CBR. In Moscow and St. Petersburg, and probably in other urban centers as well, commercial banks have developed some very innovative ideas for the formation of local and regional clearing houses to support the more timely and reliable exchange of payments. Some of the Moscow-based financial exchanges are also fairly far advanced in the design and testing of "closed" payment systems, involving - 87 - only the mutual obligations among the members of the exchanges. These private initiatives hold much promise and are receiving at least some support at the central bank. At the same time, however, the CBR believes, correctly so, that it has an obligation to ensure the integrity of the payment process. In light of recent financial fraud in Russia, including fraud perpetrated through the payment system, the CBR is being cautious about authorizing private clearing arrangements. Settlement of obligations incurred as a result of trading on exchanges is a continuing problem in Russia. For example, in the Moscow International Foreign Currency Exchange, where there is trading of the ruble against foreign currencies (primarily the U. S. dollar), settlement takes three or four days. On the Moscow International Stock Exchange, where shares are traded and interbank funds are bought and sold, the minimum time lag between trading and settlement is about two weeks. Not only are these lengthy settlement delays inefficient, they contribute to increased market risk for the participants and thereby hinder, to a large degree, the continued development of this and other financial markets. In the United States, by contrast, the settlement of foreign exchange transactions and interbank funds transfers is almost always done on a same-day or even intra- day basis. In the U. S. stock market, the vast majority of equity stock market trades (for mutual funds and pension funds) takes two days while settlement of the remaining trades can take up to five days. 27 Both of the Russian exchanges described above are actively exploring, and even experimeriting with, their own closed settlement systems. They are also experimenting with multilateral netting as a means to reduce the liquidity burdens faced by the exchange participants. While it has many benefits, multilateral netting also poses complex risk management issues which sophisticated financial markets are just coming to appreciate. 28 In this regard, the CBR should be satisfied that proper risk controls are in place as a condition to providing access to accounts on its books to effect final settlement of netted trades or payments. Lessons Learned. A number of general conclusions can be drawn from this analysis. First, very simple and low cost enhancements to existing payment arrangements, such as substituting dedicated bank couriers for the public mail system, can generate highly W The U. S. Securities and Exchange Commission has announced plans to reduce these time lags to one and three days, respectively, primarily by reliance on electronic confirmation of trades plus electronic payments. W See Chapter 4. This issue is also covered in BIS, November 1990 and, in considerable detail in BIS, September 1993. - 88 - cost effective improvements in the Russian payment system. While, as a longer-run matter, more efficient electronic payment methods can be introduced in line with availability of the technologies and human resources needed to manage them, there should not be an automatic aversion to the continued use of paper-based payments. Even in a nation like the United States, the paper check still dominates retail transactions. This should not be seen as advocating paper payments. Rather, this reflects the view that some fairly simple and inexpensive technical tuning of the current Russian payment system holds the promise for immediate improvements and the creation of some "breathing room" to pursue more complex and longer-run technological innovation. Second, the greatest need in the Russian payment system at the current time is the implementation of a same-day settlement system with finality for large-value ruble payments. Risk in the Russian banking and financial system is likely to increase as a natural consequence of market development. There are also sure to be bank failures, with the possibility of large systemic risks for commercial banks and (potentially) the central bank. A large-value ruble payment system with same-day finality, appropriately structured, will provide an important buffer against financial shocks when they arise. Third, procedures should be adopted to encourage and facilitate cooperation and coordination between the central bank and commercial bankers on payment issues. For example, private clearing arrangements that seek to achieve final settlement on the books of the CBR will need to meet preconditions for the safety and soundness of the particular arrangement before access to the CBR's settlement services is granted. It may be that these preconditions are not yet well understood, much less articulated, and an extensive dialogue between the private and public sectors will be needed to determine the appropriate risk control scheme for Russia. As is the case in the West, institutional means are needed to ensure that both private and public views are aired fully. Central banks, as governmental entities, should follow a defined process for seeking public input to important policy and operational decisions, such as the definition of terms of access to interbank settlement services. Consultation with the commercial banks and their trade groups, as well as with the end users of payment services, is essential to enlightened regulation. Finally, more deliberate attempts should be made to design payment services around the needs of the ultimate beneficiaries in the market, namely, consumers and enterprises. Market research should be adopted as an integral component of payment system change. While commercial banks should be primarily responsible for defining the needs of their · customers, by the same token the central bank should be attempting to meet the legitimate needs of the commercial banks. Effects of th11 Payment Sy:rtt1m on th11 Money Supply in Russia. An inefficient - 89 - payment system, besides hindering enterprise production and trade and limiting the growth of money markets for savings mobilization and monetary control, also can absorb a large portion of the money supply. This absorption is needed to provide payments liquidity during the transition from a planned economy to a market system and occurred in Russia for three reasons. First, there was a dramatic rise in credit risk for inter-enterprise payments. Second, there was a large increase in the time it took to complete a payment transaction. These two consequences combined to generate large increases in payment system credit float which has absorbed a significant portion of the money supply, rather than going to the rest of the economy. Third, the balances needed to clear and settle payments has also absorbed liquidity that could have been used elsewhere. Improvements in payments processing procedures and in how payments are cleared and settled will release much of this liquidity back into the system even if the money supply statistics remain unchanged. Payment Order Credit Float and Reduced Liquidity. As noted above, prior to the breakup of the former Soviet Union, inter-enterprise payments were handled through Gosbank. rnal accountin entries transferring enterprise balances from one part of government to another. Under this arrangement, receivers o payments did not have to assess the creditworthiness of the enterprise sending the payment or the solvency of the bank upon which the payment demand order--a commonly used debit transfer instrument like a check--was drawn. All this was changed with the fragmentation of Gosbank and, especially, the uncertainty of the solvency of both other enterprises and their banks. Payment demand orders were subsequently outlawed, because of credit risk and fraud problems, with only payment orders (or cash or barter) being used for inter-enterprise payments. Payment orders eliminate credit risk because they are like GIRO payments, which is a pay-in-advance type of payment. In isolation, the shift to payment orders in a market economy is beneficial. This is because three common problems associated with debit transfer instruments are eliminated. As discussed in Chapter 2, a credit transfer instrument, such as a payment order, removes credit risk, eliminates returns of orders due to insufficient funds, and (in mature payment systems) experiences lower unit payment costs. But in Russia, with the increase in time it takes to complete an inter-enterprise payment order, excessive credit float was created. Credit float occurs because, to avert credit risk, the account balance of a paying enterprise or bank is debited with a payment order before the account balance of the receiving institution is credited, and only then after the payment process is complete. In Western Europe this payment delay is only 1 day and occurs because the interest earned on the 1 day delay is sufficient to cover alt of the GIRO's operating expenses. - 90 - In Russia prior to the breakup, payments through Gosbank were handled with only a 3 to 7 day delay. With the fragmentation of Gosbank into over 1,400 regional CSCs payment delays have expanded greatly. Only payments occurring between banks served by the same CSC can be handled in a timely fashion. Because payment order processing is primarily manual and these paper-based payments are often required to be transported long distances by inefficient means, payment delays can be 3 to 4 weeks. These two events--a shift to payments orders to reduce credit risk and a lengthening of the time it takes to clear and settle inter-enterprise payments--have resulted in a significant absorption of liquidity just to run the payment system. As of January 1993, credit float was estimated to be almost 55% of the monetary base in Russia, meaning that a substantial portion of the increase in the money supply during this period has not been injected into the economy but rather has been absorbed as credit float in the payment system. The vast majority of the credit float is retained within the CSCs, rather than individual banks. Although credit float within the CSCs does not earn an interest return--it is "dead money"--it does represent a one-shot drain on the money supply. The introduction of telecommunications and computing technology to speed up the currently manual clearing and settlement of inter-enterprise payments will improve enterprise production and trade by reducing the time it takes to clear and settle payments. As well, payment order credit float will fall dramatically and result in an effective expansion of the money supply--a development that needs to be planned for in advance for proper monetary control. Lower Balance Requin,rnents and Expanded Liquidity from Real- Time Gross Settlement or Payments Netting. As payment mechanization proceeds, balances held at CSCs for clearing and settlement purposes can be reduced. This reduction will release liquidity into the economy, just as occurs when credit float is reduced. One way these balances can be reduced is to move toward real-time gross settlement for large value inter-enterprise payments. Currently, settlement of payments at CSCs occurs when the fil!!!!. of gross debits to a particular account is posted to that account, followed by the posting of the sum of gross credits. This sequence of first posting total gross debits followed by the posting of total gross credits likely was instituted to help protect the CSCs from credit risk when payment demand orders (debit transfer instruments) were still in use. While such an arrangement could not protect CSCs against crediting enterprise or bank accounts for payment demand orders ·subsequently found to be fraudulent, it could assure CSCs that debits to an account would be covered with the balances on hand. After payment demand orders were outlawed, only credit transfers--payment - 91 - orders--could be used. Since one account is debited prior to another account being credited, payment orders represent good funds and there is no longer a need to post the total debits before the total credits. This means that the large clearing and settlement balances held at CSCs can be reduced in either of two ways without generating any increased risk. First, CSCs can move toward a more real-time posting arrangement by posting accounts multiple times either at the end of the processing cycle or at different times during the day, even under current manual processing arrangements. With little effort and no loss of control, one-third of the debits (for example) could be posted first, followed by one-third of the credits, and so on until all of the debits and credits have been accounted for that day. In a real-time gross settlement arrangement (such as that on Fedwire in the U.S. or SIC in Switzerland) individual gross debits and credits are posted to accounts as they arrive at the institution holding the clearing and settlement accounts. The important result of this multiple debiting and crediting is that clearing and settlement balances at CSCs can be reduced without any change in risk or operational control. A second, alternative way to reduce the excessive size o ese be to replace gross settlement with net settlement, as is done on CHIPS in the U.S., CHAPS in the U.K., and BOJ-NET in Japan. Because only payment orders are being used, there is no change in risk or operational control if the sum of debits to an account were first subtracted from the sum of credits so that only the net credit or net debit were posted to the CSC clearing and settlement balance. Since the resulting net debit would often be considerably smaller than the sum of gross debits, the idle clearing and settlement balances held with the CSCs could be significantly reduced, lowering the need for payments working capital in the banking system. Net settlement arrangements in the U.S., U.K., Japan, and elsewhere are actually more risky than would be a net settlement system in Russia based on payment orders. First, net settlement systems in developed nations typically start with zero balance accounts so that the net debits that occur may or may not exceed the reserve balances at the central bank used for settlement. If the net debit exceeds the reserve balance, and no additional credit is forthcoming, a settlement failure occurs. Second, net settlement systems can include debit transfer instruments, as in check net settlements, so that a portion of the gross credits (around 1 %) are routinely returned later for insufficient funds, understating the true net debit at the t ime net settlement occurs. By relying on non-zero balances at CSCs and using only payment orders, Russia effectively circumvents these two problems. Chapter 8: HARMONIZATION OF LARGE VALUE PAYMENTS ACROSS COUNTRIES AND A BENEFIT/COST ANALYSIS OF SETTLEMENT FINALITY The movement toward economic and financial market integration within the 1 2 member countries of the European Economic Community (EEC) has prompted efforts to devise and impose common standards on the operation of large value payment networks. 29 The purpose is to eliminate important differences between large value payment systems that can both affect the smoothness of domestic and cross-border payments within the EEC as well as distort competitive relationships among networks. This effort foreshadows the harmonization of the world's major payment networks. Achieving Settlement Finality in the EEC. The central banks of the EEC agreed to develop minimum standards and criteria concerning risk management, access, legal structure, operational integrity, pricing policies, and business hours on their domestic large value payment networks. 30 The most important standard concerned achieving settlement finality through collateralization of all net debits on a net settlement payment network. Settlement finality is to be achieved in the EEC through a variety of means. First, bankruptcy laws are to be harmonized among countries so that payment unwinds will not occur and the net payment position at the time of firm or bank failure will be protected. Second, risk-reduction standards comprising real-time sender net debit caps, receiver determined bilateral net credit limits, and posted collateral sufficient to cover the fa ilure to settle the single largest possible net debit are to be imposed on all large value net settlement networks. 31 Third, in the future, net settlement networks will not be allowed to operate with significantly lower risk standards than those that exist on real-time gross settlement networks in each country. This will mean that efforts will be made to have net settlement networks fully collateralize all their net debits, both during the day and at end-of-day settlement time . The basic difference between the achievement of settlement finality in the EEC and the U.S. lies in the percent of net debits to be covered by collateral. The major large value net settlement network in the U.S. (CHIPS) already meets these risk-reduction standards whereas not all EEC networks do. However, if the EEC adopts full collateralization of all net debits to eliminate all residual risk of settlement failure, then the U.S. and other countries w ill ll' The EEC is composed of Germany, France, Belgium, Luxembourg, Netherlands, Denmark, Italy, Greece, Spain, Portugal, Ireland, and the U.K. EEC payment systems are described in Committee of Governors of tho Central Banks of the Member States of the EEC, September 1992. ~, Committee of Governors of the Central Banks of the Member States of the EEC, November 1993. W These risk-reduction standards have been called Lamfalussy standards (BIS, November 1990). - 93 - face pressure to do so as well. Indeed CHAPS in London has already committed to this goal (Bank of England and APACS, 1994). If adopted in the rest of Europe and in the U.S., then domestic and cross-border payments would be fully protected from settlement failures. Such a development would also likely set the standard for payment networks in other countries. Due to historical and institutional reasons associated with the evolution of the banking systems of the U.S. and Europe, different approaches had been developed to address payment risk. The U.S. has relied on checks--debit transfers that may be returned for insufficient funds and thus face settlement f ailure--for its non-cash payments. In contrast, Europe has relied more heavily on GIRO payments--credit transfers that are not made if funds are insufficient and so have no risk of settlement failure. In addition, the legal structure regarding bankruptcy and the protection of net payment positions in the event of failure have also been different between the U.S. and Europe, with the U.S. legal structure offering greater protections for payments. The expectation is that these differences will be eliminated in the future for large value payment networks. The risk of settlement a1 ure res s on e many participants on a large value network may fail on a given day. Although these probabilities are unknown, both are believed to be small. As well, probability theory tells us that the probability of multiple simultaneous failures should be far smaller than the probability of a single failure to settle. 32 Because these probabilities are unknown, it is not possible to accurately determine the benefit from either eliminating the risk from the failure to settle from one or multiple payment participants, although the cost can be approximated. A Benefit/Cost Analysis of Settlement Finality. It is not really possible to determine accurately the risk of settlement failure on a large value payment network for the simple reason that such failures (depending on how one treats the 1974 Herstatt problem) either have never or almost never occurred. If the frequency approach to calculating probability is used here, the implied probability of a settlement failure by a single payment participant would be determined by the ratio of the number of actual settlement failures to the number of times there could have been a failure to settle (which is, approximately, the number of business days a net settlement system has been in operation). To illustrate, if a net settlement network has been in operation for each of 255 business days over 30 years then the possible number of times there could have been a settlement failure would be 7,650 (from 255 business days per year times 30 years). If this network experienced one settlement f ailure--or would have had a settlement failure had there W How much smaller depends on how independent the failure of one participant is from the same- day failure of other participants. - 94 - not been an agreement to deal with the problem in another manner--then the frequency approach to probability would define the probability of settlement failure as once cwery thirty years or .000131 on a given business day (from 1/7,650). Of course, if no settlement failures have occurred, then the probability of a settlement failure would be zero (from 0/7,650). Unfortunately, from a statistical standpoint, both of these probability estimates likely contain a high degree of forecast error. Although these probabilities represent our best estimate of the likelihood of future events, their rareness or nonoccurrence means that they are not actuarially sound. This result is understood intuitively by both central banks and participants on large value net settlement networks. Central banks have argued that the fact that an important settlement failure by one or more participants may not have occurred on a particular network does not mean that it will never occur. And payment participants on net settlement networks have, indirectly, supported this view by stating in legal agreements that they will not be held liable for losses associated with a settlement failure if these losses exceed arrangements in place (e.g., collateral, balances, or limited loss-sharing agreements) to cover such an event. In short, if the losses from a settlement failure are large enough, a payments unwind may occur. Thus, neither side really believes--as demonstrated by their respectiv13 statements and actions--that the existing arrangements or even the Lamfalussy risk-reduction standards are sufficient to completely eliminate the risk of settlement failure on a net settlement network. But how small would this risk have to be so that the benefits from eliminating it would exceed the costs of preventing it? Benefit/cost analysis is used regularly in pricing insurance. Here the expected loss to the insurance company from a particular event (plus a mark-up for profit) determines its cost to the purchaser. The expected loss is computed by multiplying the probability of loss (Pr(loss)) by the value of the insurance payoff amount if the event occurred. From the perspective of the person buying the insurance, the benefit is the insurance coverage obtained (equal to the expected loss to the insurance company) and the cost is the premium paid or: Benefit = Cost Pr(loss) over a year x payoff amount = premium paid over a year. For example, if an individual (of a given age and risk category) paid an annual premium of $300 for a $500,000 term life insurance policy, the probability of loss for the insurance company (neglecting profits) would be Pr(loss) = $300/$500,000 = .0006 or .06% over the year. A similar analysis can be used to determine the implied Pr(loss) of a settlement failure that would equate the benefit of preventing a settlement failure with the cost of providing for settlement finality. - 95 - One way of providing for settlement finality would be to post collateral that could be used to cover losses associated with a settlement failure. If participants on a net settlement network paid 25 basis points (annual rate) to obtain $1 of collateral for a year, then the cost over the year would total $.0025 while the cost per business day would be $.0000098. 33 With a 25 basis point (annual rate) cost of collateral and a net debit exposure to be covered each business day of $3 billion, the annual cost is then $7.5 million or $29,411 per business day (from (.0025 x $3,000,000,0001/255 business days).. Following the insurance example above, the implied probability of a settlement failure--the Pr(loss)--that equates benefits with costs is determined from: Benefit per working day = Cost per working day Pr(loss) each day x net debit exposure = daily cost of collateral Pr(loss) each day x $3 billion = $29,411 Pr(loss) eac It was seen earlier that if there was one settlement failure on a net settlement network over a thirty year period, the probability of loss would be .000131. This was determined from (1 settlement failure)/[(255 business days per year) x (30 years)) so that the formula [Pr(loss) x 255)"1 will give the number of years that may elapse before another settlement failure is expected to occur (in this case 30 years). Applying this formula to the Pr(loss) of .0000098 that equates the benefit of preventing a settlement failure with the cost of the collateral to assure settlement finality gives: [Pr(loss) x 255)" 1 = years to next failure (.0000098 x 255]" 1 = 400 years. ll' In the U.S., government securities have been used as collateral to provide for settlement finality. Collateral can be obtained by borrowing in the overnight federal funds market, purchasing government securities, and posting these securities as collateral with the central bank (or some other party). Since the bank still owns the securities, the cost of collateral is the spread between the return on the securities and the average overnight federal funds rate over the time period the securities are held. This spread should be very close to the spread between the average overnight federal funds rate and the dealer repurchase agreement rate, which averaged 20-25 basis points over 1983-1990. ll' Changing the exposure amount ($3 billion) will not change the computed Pr(loss): any percentage change in the exposure wi!I lead to an identical percentage change in the daily cost, leaving the Pr(loss) computation unaffected. The ultimate risk of loss of the collateral is faced by those banks that lent the federal funds to purchase it. - 96 - If it is argued that a failure to settle will lead to a loss of $3 billion over a longer time period than once every 400 years--say once every 500 years, then the benefit is $23,400 and is less than the cost of $29,411 .35 In this case collateralization should not occur because the cost exceeds the benefit. Alternatively, if it is argued that such a failure will occur within a shorter time period--say once every 300 years, then the benefit is $39,000 and will be greater than the cost. 36 In this case collateralization should occur. To summarize, if the cost of collateral is truly 25 basis points (annual rate), then central banks only need to argue that they expect (subjectively) a major failure to settle more frequently than once every 400 years to conclude that the benefits·are greater than the costs. In contrast, payment participants who will be incurring the cost of posting the collateral need to argue that they expect (again subjectively) a major failure to settle less frequently than once every 400 years to conclude that the benefits are less than the costs. Although only illustrative, this exercise demonstrates that even with a very low probability of settlement failure the benefit may still cover the cost because the costs of providing collateral to ensure settlement finality are so low. To illustrate this point, suppose instead that idle balances were to be held to ensure settlement finality. With the same $3 billion in exposure and at an interest cost of .05 (annual rate), the annual expense would be $150 million, giving a cost per working day of $588,235. When this daily cost figure is used in the above equations, the Pr(loss) that equates benefit with cost is .000196 (from $588,235/$3 billion). This probability means that a loss of $3 billion is expected to occur once every 5,100 working days or once every 20 years (from [.000196 x 255)" 1 ). The daily benefit of holding balances to prevent a settlement failure under different assumed frequencies of a settlement failure is compared with the daily cost of using balances in Table 8-1. If a settlement failure was expected to occur once every 10 years and the exposure is $3 billion, then the benefit ($1,176,000) would exceed the cost ($588,000) and balances should be held to prevent a settlement failure. A loss of $3 billion once every 20 years gives a benefit per working day of $588,235 and equals the daily cost, so here there is no clear guide on whether to hold balances or not. If the historical probability of a settlement failure has been one failure every 30 years (as used illustratively above), then the benefit of ensuring settlement finality using idle balances would be $393,000 per working W A failure once every 500 years implies a failure once every 127,500 working days, giving a Pr(loss) of .0000078 (from 1/127 ,500). Thus the benefit is .0000078 x $3,000,000,000 = $23,400 per working day. l!I .~ failure once every 300 years implies a failure once every 76,500 working days, giving a Pr(loss) of .000013 (from 1/76,500). Thus the benefit is .000013 x $3,000,000,000 = $39,000 per working day. - 97 - day. With a settlement failure of once every 30 years, the cost exceeds the benefit and idle balances as a means of obtaining settlement finality should not be used. However, using the same benefit estimate ($393,000), benefits will exceed costs ($29,411) if collateral is used to ensure settlement finality. This illustrates that the lower are the costs in providing for settlement finality, the greater is the likelihood that the alternative being considered will pass a benefit/cost "test". It also illustrates what the frequency of settlement failure would have to be to either pass or fail a benefit/cost test. The less frequent a settlement failure is believed to be, the more likely it is that proposed arrangements of providing for settlement finality will not pass the test and should be rejected. Table 8-1 Benefit and Cost of Obtaining Settlement Finality Using Balances (Gross settlement, annual cost of balance is 5%) ·. .· , . .. ·. . . . . . . . . .. . Failure to Settle · · ·. DailyPr(lossT · · Daily Benefit · .· ·.. 1/10 yr .000392 $1,176,000 $588,000 1 /20 yr .000196 588,000 588.000 1 /30 yr .000131 393,000 588,000 1 /50 yr .000078 234,000 588,000 Conclude: If expect a settlement failure more than once every 20 years, then hold balances to prevent it; otherwise do not. Pr(lossJ = 1/[years x 255]; Benefit = Pr(loss) x Exposure; Cost = [.05 x exposure]/255 Ways to Reduce Collateral or Idle Balance Requirements. While there is no real consensus view on the expected frequency of a failure to settle on a large value net settlement network, there is agreement that payment participants could--at some cost--reduce the size of net debits on these networks. Reductions in net dt~bits both reduces the risk that a large failure to settle will occur and decreases the cost of ensuring for settlement finality. Net debits can be reduced using payments netting, partial payments, and maturity extensions in funding operations. Payments may be netted outside of a large value payment network with the underlying payment instrumef)t acting as collateral for the transaction. This has been the approach taken for the settlement of commercial paper in the U.S. Payments netting has also been applied using contract novation for future delivery of foreign exchange. Under foreign exchange netting by novation, only the net amount due at the settlement date would be - 98 - transferred rather than the sum of all the prior gross positions. The potential for net debit reduction is large on CHIPS, for example, since over half the value of CHIPS payments involve foreign exchange. Another possibility is the use of partial payments. When a single payment would be so large as to be rejected--due to real-time controls on net debits--it can be broken up into smaller pieces which can be sent at different times during the day as covering funds arrive. Partial payments will not change the net debit position at settlement time but can reduce the maximum net debit incurred during the business day. Partial payments are, in certain instances, already being used on both the Swiss SIC system and on Fedwire in the U.S. Partial payments save costs on SIC by reducing the level of idle balances needed to make payments. As well, partial payments reduce the probability of a payments "gridlock" on SIC associated with the holding of lower idle balances for payment clearing purposes. Such an arrangement requires some added expense of reconciling payment flows to determine when the entire payment has been received, however. But this expense is not large and is already incurred for government security transfers on Fedwire where the maximum limit on a securities transfer is $50 million. Partial delivery of securities reduces the cost of securities fails and can reduce the size of net debit payment exposure for security transfers. When banks fund themselves with overnight funds, there exist a number of ways the net debits associated with this funding can be reduced. Eurodollar payments account for one-fourth of the value of CHIPS payments. The value of these payments could be reduced if .these funds were subject to rollovers or continuing contracts, when the buyer and seller remains the same, or when these funds are borrowed for longer time periods. Such . a way to reduce net debits may well save more in collateral costs than it takes to restructure funding arrangements. This has been the conclusion reached for Fedwire, where the potential cost of maturity extensions, rollovers, and continuing contracts in the federal funds market (accounting for over 40% of the value of Fedwire payments) was deemed to be less than the 25 basis point fee imposed for daylight overdrafts on that network in 1994. A Method for Tr11nsfonnlng II Net Settlement Network Into the Equivalent of a 1'1111/- Time Gross Settlement Network. The central banks of the EEC have made it clear they would prefer large value payments to be made over a real-time gross settlement network rather than a net settlement network. If all net debits are fully collateralized on a net settlement network, there is a simple and low cost way that it can be transformed into the equivalent of a real-time gross settlement network. A real-time gross settlement network transfers and settles each payment at the - 99 - time it is made. If a payment participant fails during the day, all payments made are final and no uncovered net debits exist. This eliminates the risk of settlement failure. However, if all net debits on a net settlement network were fully collateralized, each payment made could be matched with an equal and simultaneous transfer of legal title to the posted collateral (using book-entry procedures) from payment sender to payment receiver. At the end of the day, all collateral that had been transferred to receiving participants would be automatically repurchased by sending participants through the movement of settlement balances at the central bank (as usually occurs on these networks today). All payments over such a net settlement network would be final when made since the receiver would own the collateral until the end of the day. This arrangement on a net settlement network would be similar to an intra-day repurchase agreement and would give the same result as that achieved on a real-time gross settlement network using expensive idle balances. Because lower cost collateral may be used in place of balances, the arrangement would also be less expensive. Importantly, in the event he collateral it sent, the bank receiving the collateral would own it and could liquidate it through a central bank Discount Window loan that day. 1s eature both averts a settlement failure, in the usual sense of the word, and eliminates any net debits or extension of credit because each gross payment would be simultaneously backed by securities equal to the value of the payment. Chapter 9: CONCLUSIONS, AND WHAT TO LOOK FOR IN RESTRUCTURING A PAYMENTS SYSTEM A payments system facilitates the exchange of goods and services in an economy. It also allows for the efficient mobilization of financial resources and lowers transaction cost, both of which promote economic growth. All payment systems involve clearing and settlement functions in order to transfer value. The clearing function--composed of processing and collection--can be paper-based or electronic. Payment instruments involve debit or credit transfers and settlement occurs at the time the transaction takes place or is delayed until later. A detailed knowledge of how mature payment systems in developed countries are structured and operate is essential in order to accurately identify areas where less mature payment systems may be improved in other countries. This book offers a basic introduction into the operation of payment systems and outlines areas where improvements may be made. A properly structured payment system should have a number of attributes. Perhaps the most essential aspects are acceptability, timeliness, and reliability. After this comes safety, efficiency, and ease of use. In restructuring a payments system, the goal is to look for effective ways to improve these attributes for both current and future users. In undertaking this task, it is necessary to obtain a full and complete understanding of how a country's payments are currently structured. This concerns both the existing institutional and legal structure surrounding small and large value payments as they are now made as well as obtaining basic data regarding payment volumes and values by different payment instruments/methods currently in use. Because of a lack of basic information, it may at times be necessary to initiate special surveys to generate some (or most) of the needed data on payment use and composition. In addition to collecting basic data on the types, volumes, and values of payment instruments in use, it is also necessary to determine how acceptable these payment media are to their primary consumer and business users and how the payment information is processed, delivered, and settled. This involves an assessment of the reliability, safety, and timeliness of mail, courier, and telecommunications facilities in the country, at least to the degree that these transportation systems are being used to move payment information. With this basic frame of reference, it is possible to determine the ability of the existing payments structure to meet the forecasted growth in use, for consumer and business payments separately, relative to current capacities. In this analysis, points where payment delays seem excessive can be identified and the value of float estimated. Payment delays are usually due to problems in the timeliness of physical transportation of payment documents/information, suggesting that alternative transportation modes be considered. This - 101 - can range from replacing physical delivery of paper documents using the mails with sending the same documents using dedicated ground or air couriers to even sending the necessary information electronically. It is probably safe to say that the majority of all payments in all countries are made in cash. While this is the main similarity among payment systems of different countries, there are at least three areas of major differences. First, payment systems differ in their intensity of use of non-cash payments for bill payments and business transactions, with more mature systems relying more heavily on a wider range of cash substitutes. Second, cash substitutes are predominately based on either debit transfer or credit transfer payment instruments. Debit transfer instruments (checks) are used primarily in countries where the banking system was historically the most important supplier of retail savings and transactions accounts while credit transfers (through a GIRO) arose in countries where some centralized entity (the postal service) provided these savings and transactions services to the general public. A third difference among payment systems concerns the conditions under which large value a ments are made. Specifically, large value payments may not be final when made For smaller value, non-cash payments, two of the most important issues are whether non-cash payments are primarily debit transfers or credit transfers and the degree to which either system is based on electronic processing and presentment of payment information. Checks, the most used debit transfer, are typically more costly to use, have uncertain finality (because of return items), and generate float benefits for payors (which leads to unneeded spending to delay disbursement by payors and speed up collection by payees). However, the full use of electronics within a check-based system can serve to eliminate these problems. The end result is that a debit transfer check-based payment system may, over time, be transformed into a credit transfer debit card system. A debit card payment system would be equivalent to a GIRO network which does not have the problems associated with checks as so is the preferred long-run non-cash payment method. Regardless of the type of payment system in operation, it is important that the prices charged should cover all the costs involved and the individual or firm who chooses to use one payment instrument over another faces these prices. A major problem with checks, for example, is not having payors pay the full operating and float costs of the decision to use a check, so this instrument is overused and real resources are wasted from the point of view of the nation as a whole. This difficulty has been solved in Canada, but not in the U.S. or France which are heavy users of checks for non-cash payments. GIRO systems are the dominant non-cash payment method in Western Europe and, unlike checks, have low costs, a certainty of finality (no return items), and generate no float benefits for payors. - 102 - Large value payments have their own set of problems. Two of the most important are whether gross or net settlement is used and whether or not to allow daylight overdrafts. Both of these problems are associated with the possibility of the failure to settle by a participant on a large value network. Gross settlement systems, where each payment is settled and therefore final when made, is the safest arrangement but also can be more expensive for participants. It is more expensive when the non-interest earning balances needed for settlement are larger than the reserves required for monetary control or, if not larger, when daylight overdrafts carry a fee or are quantitatively tightly controlled. Net settlement systems--where gross payments are cumulated over the business day and net positions settled at the end of the day--can be cheaper because payments are made against zero balance accounts. With zero balance accounts, no non-interest earning balances need to be held and daylight overdrafts--a necessary corollary to zero balance accounts--are free. If daylight overdrafts are allowed, risk can be reduced to acceptable levels when overdrafts are priced at some market rate or if interest earning collateral sufficient to cover a settlement failure is posted by participants. The current trend in Europe is to collateralize all net debits on large value net settlement payment networks. This is cheaper than holding idle balances but still ensures that payments made will be final and not reversed, thereby eliminating the risk of settlement failure. The benefits from risk reduction are shown to be likely greater than the costs. Such payment arrangements support economic and financial market integration both domestically and across borders and foreshadows how the world's major payment networks will likely operate in the future. As should be clear from the foregoing discussion, improvements to a country's payment system are most easily determined by contrasting the operation and structure of the payment system in question with the operation and structure of mature payment systems in developed countries. Payments technology can usually, with some modification, be adapted to local conditions and perform with similar effectiveness in most environments. In any event, change should be evolutionary rather than revolutionary; successful improvements to a payment system depends to a large degree on identifying important unmet user needs and working closely with users to find solutions they can accept. These conditions for success recommend proven payments procedures rather than the newest technology. - 103 - BIBLIOGRAPHY Avery, R., G. Elliehausen, A. Kennickell, and P. Spindt, "The Use of Cash and Transaction Accounts by American Families", Federal Reserve Bulletin. February 1986. eank for International Settlements, Large-Value Funds Transfer Systems in the Group of Ten Countries, easel, May 1990. eank for International Settlements, Report of the Committee on Interbank Netting Schemes of the Central ~anks of the Group of Ten Countries, Basel, November 1990 (Lamfalussy report). Bank for International Settlements, Committee on Payment and Settlement Systems of the central banks of the Group of Ten countries, Central Bank Payment and Settlement Services with Respect to Cross-Border and Multi-Currency Transactions, Basie, September 1993. Bank for International Settlements, Committee on Payment and Settlement Systems of the central banks of the Group of Ten countries, Payment Systems in the Group of Ten Countries, Basie, December, 1993. n land and APACS, "The Development of Real-Time Gross Settlement (RTGS) in the United Kingdom", information re ease, Borio, C.E.V. and P. Van den Bergh, "The Nature and Management of Payment System Risks: An International Perspective", Bank for International Settlement Economic Papers, No.36, Basie, February 1993. Boeschoten, W., Currency Use and Payment Patterns, Financial and Monetary Policy Studies, Vol. 23, Kluwer Academic Publishers, 1992. Committee of Governors of the Central Banks of the Member States of the EEC, Payment Systems in EC Member States, September 1992. Committee of Governors of the Central Banks of the Member States of the EEC, "Minimum Common Features for Domestic Payment Systems", Working Group on EC Payment Systems, November 1993. Datt, S. and A. Shanmugham, "New Data on Payment Systems in Developing Countries", Payment Systems Worldwide, Spring 1994. Federal Reserve Bank of New York, A Study of large-Dollar Payment Flows Through CHIPS and Fedwire, New York, December 1987. A summary of these results was published as: "Large-Dollar Payment Flows from New York", Quarterly Review, Winter 1987-88. Hager, K., 'The Swedish Postal GIRO and its Progress", Journal of Bank Research, Volume, Month 1986: 227-31. Humphrey, 0., "Payments Finality and Risk of Settlement Failure", in A. Saunders and L. White (editors), Technology and the Regulation of Financial Markets. Lexington Books, Lexington, MA, 1986. Humphrey, D .• and A. Berger, "Market Failure and Resource Use: Economic Incentives to Use Different Payment Instruments", in D.B. Humphrey (editor), The U.S. Payment System: Efficiency, Risk and the Role of the Federal Reserve, Kluwer Academic Publishers, 1990, 45-86. National Commission on Electronic Fund Transfers, EFT in the United States, Washington, D.C., 1977. Thomsom, F., GIRO Credit Transfer Systems, Macmillan Company: New York, 1964. - 104 - Biogr11phic11/ lnform11tion about the Author David B. Humphrey is Professor of Finance and Fannie Wilson Smith Eminent Scholar in Banking at Florida State University. From 1975 to 1991 he was with the Federal Reserve System. At the Board of Governors he served as Assistant Director of the Division of Research and Statistics and at the Federal Reserve Bank of Richmond he was Vice President and Payment System Adviser. Prior to that, he taught at Tulane and San Francisco State Universities. He is currently on the Board of Editors of the Journal of Monetary Economics, Journal of Banking and Finance, and Journal of Financial Services Research. He has published over 60 papers on banking and payment system topics (mostly by luck) and continues research in these areas. He received his Ph.D. from the University of California, Berkeley in 1969, and his M.A. and B.A. from San Diego State University in 1963 and 1962, respectively; all degrees are in Economics. Distributors of World Bank Publications ARGENTINA The Middle East Observer KOREA, REPUBLIC OF SOUTH AFRICA, BOTSWANA Carlos Hirsch, SRL 41, Sherif Street Pan Korea Book Corporation For single titles: Galeria Guemes Cairo P.O. Box 101, Kwangwhamun Oxford University Press Florida 165, 4th Floor-Ofc. 453/ 465 Seoul Southern Africa 1333 Buenos Aires FINLAND P.O. Box 1141 Akateeminen Kirjakauppa Korean Stock Book Centre Cape Town 6000 Oficina del Libro Internacional P.O. Box 128 P.O. Box34 Alberti 40 SF-00101 Helsinki 10 Yeoeido For subscriptior. orders: 1082 Buenos Aires Seoul International Subsaiption Service FRANCE P.O. Box 41095 AUSTRALIA, PAPUA NEW GUINEA, World Bank Publications MALAYSIA Craighall FIJI, SOLOMON ISLANDS, 66, avenue d'lena University of Malaya Cooperative JohaMesburg 2024 VANUATU, AND WESTERN SAMOA 75116 Paris Bookshop, Limited D.A. Information Services P.O. Box 1127, Jalan Pantai Baru SPAIN 648 Whitehorse Road GERMANY 59700 Kuala Lumpur Muncli-Prensa Lioros, S.A. Mitcham 3132 UNO-Verlag Castello37· Victoria Poppelsdorfer Allee 55 MEXICO 28001 Madrid 53115 BoM INFOTEC AUSTRIA Apartado Postal 22-860 Librer!a Internacional AEOOS Gerold and Co. GHANA 14060Tlalpan, Mexico D.F. Consell de Cent, 391 Graben31 Greenwich Mag. and Books 08009 Barcelona A-lOll Wien Rivera Beach Holle NETHERLANDS PO Box Oll98 De Lindeboom/lnOr-Publikaties SRI LANKA AND THE MALDIVES BANGLADESH Osu-Accra P.O. Box202 Lake House Bookshop Micro Industries Development 7480 AE Haaksbergen P.O. Box2-!4 Assistance Society (MIDAS) GREECE 100, Sir Chittampalam A. House 5, Road 16 Papasotiriou S.A. NEW ZEALAND Garcliner ~tawatha Dhanmondi R/ Area 35, Stournara Str. EBSCO NZ Ltd. Colombo2 Dhaka 1209 106 82 Athens Private Mail Bag 99914 New Market SWEDEN BELGIUM HONG KONG, MACAO Auckland Fritzes Fackboksforetaget Jean De UMOY Asia 2000 Ltd. Regeringsgatan 12, Box 16356 Av. du Roi 202 46-48 Wyndham Street NIGERIA 5-106 47 Stockholm 1060 Brussels WiMing Centre University Press Limited 7th Floor Three Crowns Builcling Jericho WeMergren-Williarns AB BRAZIL Central Hong Kong Private Mail Bag 5095 P. 0. Box 1305 Publicacoes Tecnicas Internacionais Lida. Ibadan 5-171 25 Solna Rua Peixoto Gomide, 209 HUNGARY NORWAY SWITZERLAND rmation Center Librairie Payot CANADA H-1117 Budapest Book Department Le Diffuseur P.O. Box 6125 Etterstad CH 1002 Lausanne 151A Bou!. de Mortagne !NOIA N-0602 Oslo 6 Boucherville, Quebec Allied Publishers Private Ltd. Van Dierman Eclitions Techniques - ADECO J4B5E6 751 Mount Road PAKISTAN P.O. Box465 Madras - 600 002 Mirza Book Agency CH 1211 Geneva 16 Renouf Publishing Co. 65, Shahrah-e-Quaid-e-Azam 1294 Algoma Road INDONESIA P.O. Box No. n9 TANZANIA Ottawa~ Ontario Pt. Indira Limited Lahore 54000 Oxford Universitv Press KlB3W8 Jalan Borobudur 20 Maktaba Street · P.O. Box 181 PERU P.O. Box 5299 CHINA Jakarta 10320 Editorial Oesarrollo SA Dares Salaam China Financial & Economic Apartado 3824 Publishing House IRAN Lima 1 THAILAND 8, Da FoSi DongJie Kowkab Publishers Central Department Store Beijing P.O. Box 19575-511 PHILIPPINES 306 Silom Road Tehran International Book Center Bangkok COLOMBIA Suite 1703, Cityland 10 lnfoenlace Lida. IRELAND Condominium Tower 1 TRINIDAD&: TOBAGO Apartado Aereo 34270 Government Supplies Agency Ayala Avenue, H.V. dela Systematics Stuclies Unit Bogota D.E. 4-5 Harcourt Road Costa Extension #9 Watts Street Dublin2 Makati, Metro Manila Curepe COTE D'IVOIRE Trirudad, West lnclies Centre d'Edition et de Diffusion ISRAEL POLAND Africaines (CEDA) Yozmot Literature Ltd. International Publishing Service UGANDA 04 B.P. 541 P.O. Box 56055 UI. Piekna 31/37 Gustro Ltd. Abidjan 04 Plateau Tel Aviv 61560 00-677 Warszawa !st Floor, Room 4, Geogiaclis Chambers P.O. Box 9997 CYPRUS RO.Y. International PORTUGAL Plot (69) Kampala Center of Applied Research P.O. B. 13056 Livraria Portugal Cyprus College Tel Aviv 61130 Rua Do Carmo 70-74 UNITED KINGDOM 6, Diogenes Street, Engomi 1200 Lisbon Microinfo Ltd. P.O. Box 2006 ITALY P.O. Box3 Nicosia Licosa Commissionaria Sansoni SPA SAUDI ARABIA,QATAR Alton, Hampshire GU34 2PG Via Duca Di Calabria, 1/1 Jarir Book Store England CZECH REPUBLIC Casella Postale 552 P.O. Box3196 National Information Center 50125 Firenze Riyadh 11471 ZAMBIA P.O. Box 668 University of Zambia Bookshop CS-ll357 Prague 1 JAMAICA SLOVAK REPUBLIC Great East Road Campus Ian Randle Publishers Ltd. Slovart G.T.G Ltd. P.O. Box 32379 DENMARK 206 Old Hope Road Krupinska4 Lusaka Sarnfundslitteratur Kingston6 P.O. Box 152 Rosenoerns Alie 11 852 99 Bratislava 5 ZIMBABWE DK-1970 Frederiksberg C JAPAN Longman Zimbabwe (Pvt.) Ltd. Eastern Book Service SINGAPORE, TAIWAN, Tourle Road, Ardbennie DOMINICAN REPUBLIC Hongo 3-Chome, Bunkyo-ku 113 MYANMAR,BRUNEI P.O. Box ST l:S Editora Taller, C. por A. Tokyo Gower Asia Pacific Pte Ltd. Southerton Restauraci6n e Isabel la Cat6lica 309 Golden Wheel Builcling Harare Apartado de Correos 2190 Z-1 KENYA 41, Kallang Pudcling, #04-03 Santo Domingo Africa Book Service (E.A.) Ltd. Singapore 1334 Quaran House, Mfangano Street EGYPT, ARAB REPUBLIC OF P.O. Box 45245 AlAhram Nairobi Al Galaa Street Cairo RECENT WORLD BANK TECHNICAL PAPERS (continued) No. 225 Dia, A Governance Approach to Civil Service Reform in Sub-Saharan Africa No. 226 Bindlish, Evenson, and Gbetibouo, Evaluation of T&V-Based Extension in Burkina Faso No. 227 Cook, editor, Involuntary Resettlement in Africa: Selected Papers from a Conference on Environment and Settlement Issues in Africa No. 228 Webster and Charap, The Emergence of Private Sector Manufacturing in St. Petersburg: A Survey of Firms No. 229 Webster, The Emergence of Private Sector Manufacturing in Hungary: A Survey of Firms No. 230 Webster and Swanson, The Emergence of Private Sector Manufacturing in the Former Czech and Slovak Federal Republic: A Survey of Firms No. 231 Eisa, Barghouti, Gillham, and Al-Saffy, Cotton Production Prospects for the Decade to 2005: A Global Overview No. 232 Creightney, Transport and Economic Performance: A Survey of Developing Countries No. 233 Frederiksen, Berkoff, and Barber, Principles and Practices for Dealing with Water Resources Issues No. 234 Archondo-Callao and Faiz, Estimating Vehicle Operating Costs No. 235 Claessens, Risk Management in Developing Countries No. 236 Bennett and Goldberg, Providing Enterprise Development and Financial Services to Women: A Decade of Bank Experience in Asia ·vate Sector Manufacturing in Poland: A Survey of Firms No. 238 Heath, Land Rights in Cote d'Ivoire: Survey and Prospects or ro;ec No. 239 Kirmani and Rangeley, International Inland Waters: Concepts for a More Active World Bank Role No. 240 Ahmed, Renewable Energy Technologies: A Review of the Status and Costs of Selected Technologies No. 241 Webster, Newly Privatized Russian Enterprises No. 242 Barnes, Openshaw, Smith, and van der Plas, What Makes People Cook with Improved Biomass Stoves?: A Comparative International Review of Stove Programs No. 243 Menke and Fazzari, Improving Electric Power Utility Efficiency: Issues and Recommendations No. 244 Liebenthal, Mathur, and Wade, Solar Energy: Lessons from the Pacific Island Experience No. 245 Klein, External Debt Management: An Introduction No. 246 Plusquellec, Burt, and Wolter, Modern Water Control in Irrigation: Concepts, Issues, and Applications No. 247 Ameur, Agricultural Extension: A Step beyond the Next Step No. 248 Malhotra, Koenig, and Sinsukprasert, A Survey of Asia's Energy Prices No. 249 Le Moigne, Easter, Ochs, and Giltner, Water Policy and Water Markets: Selected Papers and Proceedings from the World Bank's Annual Irrigation and Drainage Seminar, Annapolis, Maryland, December 8-10, 1992 No. 250 Rangeley, Thiam, Andersen, and Lyle, International River Basin Organizations in Sub-Saharan Africa No. 251 Sharma, Rietbergen, Heimo, and Patel, A Strategy for the Forest Sector in Sub-Saharan Africa No. 252 The World Bank/FAO/UNIDO/Industry Fertilizer Working Group, World and Regional Supply and Demand Balances for Nitrogen, Phosphate, and Potash, 1992/93-1998/99 No. 253 Jensen and Malter, A Global Review of Protected Agriculture No. 254 Frischtak, Governance Capacity and Economic Reform in Developing Countries No. 255 Mohan, editor, Bibliography of Publications: Technical Department, Africa Region, July 1987 to April 1994 No. 256 Campbell, Design and Operation of Smallholder Irrigation in South Asia No. 257 Malhotra, Sinsukprasert, and Eglington, The Performance of Asia's Energy Sector No. 258 Willy De Geyndt, Managing the Quality of Health Care in Developing Countries No. 259 Chaudry, Reid, Malik, editors, Civil Service Reform in Latin America and the Caribbean: Proceedings of a Confe The World Bank Headquarters 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. European Office 66, avenue d'Iena 75116 Paris, France Tokyo Office Kokusai Building 1-1, Marunouchi 3-chome m Chiyoda-ku, Tokyo 100, Japan Telephone: (202) 477-1234 Telephone: (1) 40.69.30.00 Facsimile: (202) 477-6391 Facsimile: (1) 40.69.30.66 Telephone: (3) 3214-5001 Telex: MCI64145 WORLDBANK Telex: 640651 Facsimile: (3) 3214-3657 MCI 248423 WORLDBANK Telex: 26838 Cable Address: INTBAFRAD WASHINGTONDC ()(.)/ ./·····.t J. ::::. '{ II Cover design by Kathy Rosen ISBN 0-8213-3111-6