_' _ ___ S I6

POLICY RESEARCH WORKING PAPER 1676

The Evolution of Payments Lessons from the evolution of Europe, Japan, and the payment systemsin Europe, in Europe, Japan, and the Japan, and the UnitedStates United States providea usefulguide for emergingmarket economies in improving their own

Lessons for Emerging Market payment arrangements to Economies foster economic grovth.

David B. Humphrey Setsuya Sato Masayoshi Tsurumi Jukka M. Vesala

The World Bank FinancialSector DevelopmentDepartment October 1996 IPOLICY RESEARCH WORKING PAPER 1676

Summary findings Some payment arrangements are more efficient than all types of transactions and the United States instead others in promoting economic growth in a market-based relies on checks. Finally, the fact that consumer payment economy. The payment experience of industrial needs were not met within the banking system led to the countries is diverse enough to identify those payment establishment of postal giros in Europe, while untimely arrangements that provide the infrastructure for business payments led to central bank involvement in sustained growth and the emergence of market-based payment processing in the United States. enterprise. Unmet user needs, inefficient payment arrangements, Based on the historical experiences of Europe, Japan, differences in payment instrument costs, and improper and the United States, a number of country attributes pricing of payment services will determine the future have led to the intensive use of different payment structure of payment systems in emerging market instruments and, in some cases, a different mix of private economies just as they have determined the evolution of and public ownership and participation in the payment payment systems in industrial countries. The authors system. Such attributes include country size, population discuss these issues and apply the lessons learned to density, banking structure, legal framework, safety, and payment arrangements in emerging market economies. payment instrument pricing. Although the evolution of payments has taken decades in These attributes explain why Japan relies heavily on industrial countries, emerging market economies hope to cash at the point of sale but uses electronic payments for complete the process in just a few years, and so will bill payments and business transactions. They also are the benefit by having a better "roadmap" for transforming reason Europe relies on credit-transfer giro payments for their payment systems.

This paper - a product of the Financial Sector Development Department - is part of a larger effort in the department to promote the development of financial sector infrastructure to support banking and capital market activities. Copies of the paper are available free from the World Bank, 1818 2H Street NW, Washington, DC 20433. Please contactTomoko Ishibe, room G8-136, telephone 202-473-8968, fax 202-522-3199, Internet address [email protected]. October 1996. (44 pages)

The Policy ResearchWorking PaperSeries disseminates the findings of work in progressto encouragethe exchangeof ideasabout development issues.An objectiveof the series is to get the findings out quickly, even if the presentations arc less than fully polished. The papers carry the names of the authors and should be used and cited accordingly. The findings, interpretations, and conclusions are the authors' own and should not be attributed to the World Bank, its Executive Board of Directors, or any of its member countries.

Produced by the Policy Research Dissemination Center The Evolutionof Paymentsin Europe,Japan, and the U.S.:

Lessonsfor EmergingMarket Economies

by

David B. Humphrey Florida StateUniversity, Tallahassee, U.S.A.

SetsuyaSato World Bank,Washington D.C., U.S.A.

MasayoshiTsurumi Hosei University,Tokyo, Japan

JukkaM. Vesala Bankof Finland,Helsinki, Finland

CONTENTS

I. Introduction 1 II. Economicactivity, types of transactions,and payment instruments 3 Sizeof economicand paymentactivity 3 Fourtypes of paymenttransactions 4 Matchingtransactions with paymentinstruments 4 IlI. The useof cashin developedcountries 6 Cashholdings as an indicatorof cashuse 6 Changesin the useof cashover time 7 Cashholdings and availabilityof P05 and ATM terminals 8 IV. Thecomposition and growth of non-cashpayments in developedcountries 11 Thecomposition of non-cashpayments 11 Paperversus electronic payments 11 Thetradeoff between cash holdings and non-cashpayments 12 Thegrowth in non-cashpayment instrument use 13 V. Largevalue payments in developedcountries and the roleof thecentral bank 16 Settlementfailure andpolicies to containsystemic risk and daylightoverdrafts 17 Therole of the centralbank 19 VI. Thehistorical evolution of paymentsin Europe,Japan, and the U.S. 20 Theevolution of paymentsin Europeand the establishmentof the giro 21 Giro paymentsmoving to electronics 23 Thesubstitution of cardpayments for cashin Europe 24 Reasonsfor giro dominancein Europe: Bankingconcentration, nationwide networks, and cooperation 24 Theevolution of paymentsin Japanand the benefitsof cashuse 25 Threestages in the evolutionof Japanesepayments 26 Theevolution of paymentsin the U.S.and the reasonfor checkuse 29 Stagesin the evolutionof U.S.payments 29 Thehistorical role of the centralbank in the paymentsystem 32 VIl. Lessonsfor emergingmarket economies 34 Countrycharacteristics and the adoptionof specificpayment arrangements 35 Point-of-salepayments 35 Bill payments 36 Disbursements 36 Financialmarket transactions 36 Evolvingpayment arrangements in emergingmarket economies 37 Therole of the centralbank in the paymentsystem 38 Bibliography 42

TheEvolution of Paymentsin Europe,Japan, and the U.S.: Lessonsfor EmergingMarket Economies

1. Introduction

Emergingmarket economies today face the need to upgradetheir payment systems and initiate new paymentarrangements in order better to providethe infrastructurenecessary for sustainedgrowth within a market-basedeconomy. This processrequires an integratedknowledge of enterpriseand consumerneeds, the necessarylegal framework,the applicationof existingas well as new technologies,and an analysisof themonetary policy and riskimplications associated with thevarious paymentsystem design options that are available. The developed countries of Europe,Japan, and the U.S.have had over 150years to transformtheir payment systems from onedominated by the useof preciousmetals (coin) and privatelyissued currency into one which relieson governmentissued currencyand specializedpaper payment instruments and finally into a modernsystem where card- basedconsumer payments and networksof electriccommunication systems have increasingly replacedboth currencyand paperpayment instruments. In transformingtheir paymentsystems, today'semerging market economies have the opportunityto learn from the diversepayment experiencesof developedcountries.

TheWorld Bankhas recently been involved in initiativesto modernizethe nationalpayment systemsin countriesof Asia(China, India, Indonesia, Viet Nam), of LatinAmerica (Brazil, El Salvador, Columbia),of Africa (Madagascar,Mozambique, Tanzania, Uganda, Kenya), of EasternEurope (Poland,the CzechRepublic, Slovenia, Albania), and countriesof the formerSoviet Union (Russia, Ukraine). In this paper,the paymentexperience of Japan,the U.S.,and a numberof developed countriesin Europeare outlinedin somedetail and the differingevolution of their paymentsystems are contrasted.The purposeis to identifyimportant country attributes that determinea nation's paymentstructure and illustratethe criticalareas that will affectemerging market economies as they seekto modernizetheir own paymentsystems. These conditions include:

(i) Thegeographical size of a countryand its populationdensity (making the communication of paymentinformation easy or difficult);

(ii) The concentrationof the bankingsystem and its interconnectedness(permitting greater movementof fundsinternally within a singleentity rather than externally between separate entities);

(iii) Thelegal structure concerning rights and liabilities of paymentparticipants (reducing risk for certainpayment instruments but not others)and antitrustlaws (affectingcooperation and competitionamong suppliers of paymentservices);

(iv) Theinfluence of culturalfactors such as crime rates (affecting the needfor cashsubstitutes); and 2

(v) The role of economicfactors that affectthe tradeoff betweenrisk and efficiency by type of transactionand paymentinstrument (reflected in relative paymentcosts, user convenience, paymenttimeliness, and the availabilityof paymentalternatives).

Theseconditions have combinedin differentways to stronglyinfluence the evolution of the paymentsystems of developedcountries and thus will also importantlyinfluence the restructuringof paymentsystems in emergingmarket economies. Overall, most countriesin Europehave evolved paymentarrangements that tend to rely on electroniccredit transfer giro payments,while Japan continuesto rely heavilyon currencytransactions and the U.S.concentrates on debit-transfercheck payments. Although all developed countries use some amount of all three of these payment instruments,along with credit and debit cards, the proportions used are quite different. The involvementof the central bank in the paymentsystem of thesecountries has also differed. While virtually all centralbanks provide settlementfor all payments,and most either operateor supervise domesticlarge value wire transfernetworks, only a few centralbanks compete with commercialbanks or postalgiros in providingprocessing for relativelylow valueconsumer and businesspayments.

The paymentsystem reforms adopted by emergingmarket economies, and the role that their centralbank will havein this process,will not follow exactlythe experienceof developedcountries. However, the conditionswhich have importantlyaffected the basic paymentstructure of Europe, Japan,and the U.S.will also affectthe evolutionof paymentarrangements in emergingeconomies. Importantly,knowledge of the paymentexperience and problemsencountered in developedcountries can serve to reduce problemsfor emergingeconomies as they restructuretheir paymentsystems. Emergingeconomies also benefit from having the opportunity of adopting modern, and proven, electronic paymentarrangements that can largelysidestep the need to first passthrough a stageof heavyreliance on paper-basedpayments. Fortunately, these countries have available to them different designoptions for modernizingtheir paymentsystems that were not availableto developedcountries. This designflexibility, however, is mitigatedby the need for emergingeconomies to more rapidly transformtheir paymentsystems, over years instead of decades.

As the paperunfolds, a numberof questionsare in effectposed and answered, such as:

* What is the generalrelationship between the value of a payment,the type of transactionit represents,and the particularpayment instrument used?

* How heavilydo developedeconomies rely on cashfor consumertransactions and what seem to be the maindeterminants of cashuse across countries?

* Why have somedeveloped countries come to rely on electronicpayments while othersstill rely on paper-basedpayment instruments?

* Which paymentinstruments are expanding their sharein total transactionsand why?

* How aredeveloped and emergingmarket economies tackling the problemof paymentrisk on largevalue paymentnetworks and which approachis the mostcost-effective in reducingrisk? 3

* How havepayment systems evolved in developedcountries and what lessonslearned here are applicableto emergingmarket economies?

* What has been the role of the centralbank in the workingsof the paymentsystem and how can it be mosteffective in improvingpayment system operations?

* How have country-specificcharacteristics shaped the structureof national paymentsystems and what implicationsdoes this have for restructuringpayment arrangements in emerging marketeconomies?

In what follows, economic activities in an economy by consumers,businesses, and governmentare reduceddown to a set of four primary types of paymentsand five basic types of payment instruments. This provides a usefulframework for outlining and contrastingthe current paymentstructures of Europe,Japan, and the U.S. To place thesedifferent paymentstructures and their developmentin a proper context, the historical evolution of paymentarrangements in these countriesis summarized. The purposeis to identify the major determinantsof why thesecountries seemto havefollowed a differentevolutionary path in the developmentof their paymentsystems and in the role played by their centralbank. In the end, we suggestthat had thesecountries been more similar in terms of their size, populationdensity, banking structure,legal framework,culture, and pricing, then the outcome-asfar as paymentinstrument use is concerned-wouldhave been pretty much the same. Having identifiedthe main determinantsof developedcountrys' paymentsystems, theseare then applied as lessonsfor emergingmarket economies in restructuringtheir own payment systems.Payment reforms in a numberof emergingmarket economies are surveyedand provide a look at how the experienceof developedcountries has been applied to date.

11.Economic activity, types of transactions,and payment instruments

Size of economicand paymentactivity. Gross national product (GNP) is an aggregate measureof economic activity. It representsthe sum of all paymentsmade to original factors of production (labor, physical capital, resources)and also equalsthe total value of all consumption, investment,and governmentexpenditures. A country'spayment activity, however, includes both final expenditure on goods and servicesas well as all of the associatedintermediate production and financialtransactions. In any given year,the total value of a country'spayments far exceedsthe value of its GNP.

This resultis seenin Figure1 which showsthe ratioof the value of paymentsin a countryto its GNP. The resultsfor a numberof Europeancountries as well asJapan and the U.S. indicatethat: (a) the value of paymentsis a multiple-sometimesa very large multiple-of GNP; (b) this multiple- reflectedin the paymentvalue/GNP ratio-has risen over the lastdecade; and (c) the level and growth of this multiple is due primarily to the level and growth of largevalue (typically financial payments) wire transfersrather than the more numerousand smallervalue retail (consumer)payments. Figure1 illustratesone way to gaugethe sizeand importanceof a country'spayment system and illustrateswhy issuesof paymentsystem risk havebecome increasingly important in developedcountries over the last decade. 4

Figure1. Ratioof annualvalue of fundstransferred to GNP

100 -- Retail payment transfers

90 - 0 Large value transfers

80RI

70

60 60~~~~~~~~0 50

10 '',20li11ld30

France Germany Italy Netherlands U.K Japan

Note: No retail transferdata are available for the first yearfor France,Germany, Italy andJapan Source: BIS: PaymentSystems in the C-10Countries (various issues)

Fourtypes of paymenttransactions. The manydifferent types of paymentsassociated with the consumption, investment,and governmentexpenditures that make up GNP can essentiallybe classifiedinto four categories: (1) Point-of-sale(POS) transactions by consumers; (2) Bill paymentsby consumers,business, and government; (3) Disbursementsby consumers,business, and government; and (4) Financialmarket transactions by businessand government.

The typesof transactionsinitiated at the point of sale(food, other retail) and for payingbills (housing, utilities, services)are well-known and do not differ markedlyacross countries. Disbursementsare defined to include person-to-persontransfers, business and government payroll payments,and governmenttransfer payments. Financial market transactionswill primarily reflect borrowings, repayments,asset purchases, and foreign exchange transactions (mostly for business).

Matchingtransactions with paymentinstruments. Generalizing, these four basic types of transactionsare associatedessentially with five differenttypes of paymentinstruments in developed countries: 5

Fourtypes of transactions Fivepayment instruments - Point-of-sale(POS) - Cash (coin and currency) - Bill payment - Cards (debit and credit) - Disbursement (payroll, transfers) - Check (debit transfer) - Financial - GIRO (credit transfer) - Wire transfer

Ranking the 4 types of transactions by their likely average value, POS would be the smallest, bill payments would be larger, disbursements would reflect an even higher average value, while financial transactions would clearly average the highest. The average value 2 per transaction for these five payment instruments for Europe (an average of 11 European countries), Japan, and the U.S. are shown below. As seen, cash transactions have the lowest average value, debit and credit cards are next highest, followed by check and giro payments which are markedly higher, with wire transfer payments in the millions of dollars per transaction.

Average value per transaction for five payment instruments (1993, in US$)

Card Wire Area Cash' Debit Credit Check Giro transfer2 (In millions) Europe 6- 14 52 91 3,405 14,423 4.9 Japan 25 165 163 79,754 3,820 93.1 U.S. 5 44 45 1,147 4,602 4.2

1 Boeschoten(1992), table 1-1,p.200. 2 Europeaverages CHAPS (U.K.), SAGITTAIRE and TBF(France), EAF (Germany), and ME (Italy);Japan is BOJ- NET;the U.S.averages CHIPS and Fedwire.

The match between transaction type and the payment instruments used for these transactions is related to the average value of the transaction and is illustrated below. Both check and giro payment methods are seen to be quite flexible in that they are often used for a wide range of payment values and transaction types. In contrast, cash and cards are typically tied to smaller value point-of-sale transactions while wire transfers are almost exclusively used for large value financial transactions.

A rankingis usedbecause only scattereddata are availableon the actualaverage value of eachof the 4 types of transactions.

The 11 countriesare: Belgium,Denmark, Finland, France, Germany, Italy, Netherlands,Norway, Sweden, Switzerland,and the U.K. 3 Japan'saverage value per wire transferis high becausemany of its transfersrepresent settlement payments for sumsof transactionsrather than individual transfers.Japan's average check value is high becauseit is almost exclusivelyused for a small numberof largevalue business/financial transactions. 6

The associationbetween transactiontype and paymentinstrument used reflects a tradeoff betweenthe efficiencyor costof usinga certaintype of paymentinstrument and the risk from lossor fraud associatedwith the transactionitself. Risk is higher when transactionvalues are large and/or crime rates are high. This obviously explainsthe use of (secure)wire transfersfor large value payments. Although the actual costof a wire transferis seeminglyquite high (on the orderof $10 to $20 or more per transfer),this costis extremelysmall relativeto the value of the transactionand is also lessthan the lossesthat could be realizedif a lower cost-but more risky and potentiallyless timely- paymentmethod were used. Thusthe choiceof paymentinstrument is influencedby the relativeprice or costfaced by the useras well asthe technicalfeasibility of actuallysubstituting one instrumentfor anotherto complete a particulartype of transaction.Cost, of course,is both explicit (externalprice plus internalexpense) as well as implicit (riskassessment, convenience, availability, and timeliness).

Associationbetween the averagetransaction value, transactiontype, and payment instrument used

Averagetransaction value Transactiontype Paymentinstruments used Smallvalue POS Cash,debit and credit card, check Intermediatevalue Bill payment Check,giro (and direct debit) Largervalue Disbursement Check,giro (and direct deposit) Largestvalue Financial Check,giro, wire transfer

The tradeoffbetween efficiency and risk is alsoan importantreason why the U.S.and some other countries have reducedtheir use of cash at the point-of-sale(high crime rate) while Japan continuesto rely heavilyon cashfor thesesame transactions (low crime rate). When risk is low, as it generallyis for smallervalue transactions,then differencesin relativeefficiency or cost will primarily determinethe type of paymentinstrument used for thesetransactions.

Ill. The useof cashin developedcountries

The limited datathat are available all suggestthat cashtransactions represent a largepercent of the number or volume of transactionsbut a very small percentof the value. Cashtransactions are estimatedto represent78°h of the volume of all transactionsin the Netherlands,83% in Finlandand the U.S.,86% in Germany,and 90% in the U.K. (Boeschoten,1992; Humphrey, 1984; Viren, 1993). In value terms,cash transactions would accountfor lessthan 5%, and in the U.S. lessthan 1%, of paymentexpenditures.

Cashholdings as an indicatorof cashuse. As dataon the volume or value of cashtransactions (a flow) are very sparse,the useof cashfor paymenttransactions is insteadinferred from the value of 7 cashholdings (a stock),either relative to the moneysupply (Ml) or relativeto GDP. Thele ratios,with the largestvalues underlined, are shownbelow for Europe,Japan, and the U.S.for 1993.

Indicatorsof cashuse, 1993

Cash/Mi Cash/GDP Europe .14 .04 Japan .27 .08 U.S. .11 .02

Cashholdings in Japanare effectivelytwice those in Europeor the U.S., suggestinga much greateruse of cashfor payments.But Europeis far from homogeneous.Within Europe,Switzerland, the Netherlands,and Belgiumall havea highercurrency/Mi ratiothan Japanwhile Sweden,Norway, Denmark,the U.K., and Finland havea lower ratio than the U.S. This result is evident from Table 1 which 0hows the cash/Mi and cash/GDPratios for 14 countriesduring 1993 as well as a decade earlier. Changesin the use of cash over time. Comparingthe level of theseratios between 1983 and 1993,the ratio of cashto Ml fell-sometimesdramatically so-for all but one Europeancountry (Italy), roseslightly in Japan,and wasconstant in the U.S. Similarresults were obtainedfor the ratio of cash to GDP over the sameperiod: this ratio fell for all but one Europeancountry (Germany)while both Japanand the U.S. experienceda slight rise. The clear implicationfrom changesin thesetwo ratiosis thatcash use is falling in Europe.

An alternativeway to illustratethe reduction in cash use is to comparethe value of cash holdingsper person(in constantU.S. dollars)with the numberof non-cashtransactions per peison. Acrosscountries, a high numberof non-cashtransactions per personis associatedwith a low stockof cash per person (comparingColumn 8 with 6 in Table 1). The estimatedrelationship across 14 developedcountries annually over 1987-93suggests that a 6.8% rise in non-cashtransactions is associatedwith a 100%reduction in cashholdings (Humphrey, Pulley, and Vesala, 1996).

4 Europeandata reflects the total U.S.dollar value of coinand currency (1993 exchange rate) for 11 countries dividedby the total dollar valueof Ml or GDP. It is equivalentto a weighted(not a simple)average of individualcountry ratios.

Sincecash balances can be held for prudentialor speculativepurposes (hoarding), in additionto pure transactionpurposes, observed levels of cashholdings only approximate the levelof cashuse. Cross-country evidencesuggests that hoarding is especiallyimportant in Switzerland,the Netherlands,and Germany but not importantin Denmark,Finland, France, Norway, or the U.K.(Boeschoten, 1991 and 1992).As seen in Table1, countrieswhere hoarding is thoughtto be importantalso have relatively large cash/MI and cash/GDP ratios distorting,somewhat, the implieddifferences in cashuse across countries.

6 Theresults for Germany and the U.S.must be taken with caution since it hasbeen estimated that perhaps 35% of thevalue of Germancurrency and 60% of U.S.currency is heldoutside the country (Boeschoten, 1992; Porter andJudson, 1995). Only if thesepercentages have remained stable over 1983-93is it possibleto accurately inferdomestic use of cashin eitherthe cash/M I or cash/GDPratios. 8

Table1. Cashholdings and annual non-cash transactions per person

Valueof cash holdings Cash' toM12 Cashto GDPratio perperson Non-cashtransactions ratio X (U.S.1993 dollars) perperson 1983 1993 1983 1993 19833 1993 1983 1993 Country (1) (2) (3) (4) (5) (6) (7) (8) Switzerland .35 .34 9.90 7.86 3,524 2,676 na 67 Nethrlands .33 .25 6.52 6.49 1,155 1,300 85 146 Belgium .41 .30 9.12 5.95 1,411 1,187 35 101 Italy .15 .16 6.37 5.75 610 1,191 13 29 Sweden .10 .09 6.21 4.63 940 1,210 na 93 Norway .33 .10 5.41 4.61 967 1,196 na 97 Germany4 .23 .20 3.53 3.89 742 995 86 139 France .17 .15 5.08 3.58 810 759 88 157 Denmark .12 .09 3.01 2.88 604 775 na 124 U.K. .26 .08 3.72 2.84 410 520 74 115 Finland .26 .07 2.07 1.97 277 438 83 120 Japan .25 .27 7.59 8.43 1,795 2,387 19 39 Canada .29 .20 3.16 3.43 529 716 88 128 U.S.4 .11 .11 1.79 2.06 328 509 184 292

Sources:IMF: InternationalFinancial Statistics, BIS: PaymentSystems in the G-10Countries (various issues) 1. Cashheld by the public (equalscurrency and coin in circulationless that held by the bankingsystem and the government). 2. M I equalsdemand deposits (transferable deposits) plus cash(defined above). 3. The1983 figures are deflated using changes in eachcountry's CPI. 4. It is estimatedthat 35% of Germancurrency and 60% of U.S.currency is heldoutside these countries, and the figures are adjustedaccordingly (Boeschoten, 1992; Porter and Judson, 1995).

Cashholdings and availability of POSandATM terminals.The decreased reliance on cashfor transactionsis associatedwith the increasedavailability of credit and debit cardterminals at the point of sale. This is seenin Figure2(a) which illustratesthe negativeand statisticallysignificant relationship betweenthe availability of electronicfunds transfer(EFT) POS terminals (terminals that acceptcredit and debit cards)across 14 developedcountries and the ratio of cashholdings to GDP for 1993 (R2 - .21). The increasedavailability of EFT-POSterminals leads directly to increaseduse, as evidenced from a very similar significantlynegative relationship in Figure2(b) betweenthe annual number of EFT-POStransactions per personand the samecash/GDP ratio (R2 - .25). Incomeand interestrates alsoaffect the useof cash. The positiveeffect of nationalincome is controlledfor by our measurement strategy(cash/GDP) while the opportunitycost of holding cashis included. The correlationcoefficient betweenthe averageof the monthlythree-month money market rate (1987-1993)and the value of cashoutstanding per person(1 993) is -0.64for the countriesshown in Table 1. 9

Figure2(a) and 2(b). EFT-POSterminals and transactions per capita with cashto GDP ratios,1993

Figure2 (a) Figure2 (b) 100 40

.35- 0o Finland c 80 Finland * France 30-

8 E 60 25 France

* Belgium DenmarkK S ~~~~~~~~~~~~~20 E NDenmark 2 . Norway .* Norway 40 U.K. 15 Belgium a U.K. Canada 10 3 20 Nether ds

Italy * SWitze nd Netherland Swteland Germany Canada U.S. Japan O U.S. Germany

0% 2% 4% 6% 8% 10% 0% 2% 4% 6% 8X 10% Cashto CDP ratio Cashto CDPratio

Source: BIS: PaymentSystems in the G-10 Countries, 1994

While debit and credit card transactionson POSterminals can directly substitutefor cash, payments,ATMs makeit easierto obtain cash:this can havetwo oppositeeffects on cashholdings. By making cash more accessible,ATMs would permit consumersto increasetheir use of cash for smallervalue transactionsand so may raisecash holdings. Alternatively,the greaterconvenience of ATMsas a way of obtainingcash may meanthat consumerswithdraw smalleramounts each time and makeup the differenceby visiting an ATM more frequently,and so may on balancereduce average cash holdings. The relationshipsshown betweencash holdings and the numberof ATM terminalsin Figure3(a) and the numberof ATM transactionsin Figure3(b) is in both casesinsignificantly negative or positive (neither R2 is greaterthan .02). Thus,in this crosscountry comparison,ATMs have no significant effect on cash holdings (in contrastto the EFT-POSresults where cash holdings are reduced).

7 Most ATM transactionsare cashwithdrawals. In the U.S., for example,86% of ATM transactionsare cash withdrawals,10% are deposits,3% are accounttransfers, and 1% are bill payments(Board of Governorsof the FederalReserve System, 1991). 10

Figure3(a) and 3(b): ATMsand ATM transactionsper capitawith cashto GDP ratios,1993

Figure3(a) Figure3 (b) 10 40 . Canada Japan. 9 . . Finland 35 Japan

80 8 Finland Netherlands D 30 13 7 - 8 25 U.S. * Sweden

6 Canada 5 t t U~~~~~~~~~~~~~~~~~~~~.K. 5a E20 -

eZ 4 U.S. Switzerland France 15 No o UK.. * Netherlands Fac Germany France Denmark ISwten 'y 10 Switzerland 2 Belgium 0

7 * Belgium 5 1 Germany Italy Denmark 0 l l l 0 I l I 0% 2% 4% 6% 8% 1091 0% 2% 4% 6% 8% 10% Cashto GDPratio Cashto GDPratio

Source:BIS: PaymentSystem in the C-10 Countries,1994

This analysissupports the conclusionthat in most major developedcountries exceptJapan, cash is being increasinglyreplaced by card-basedelectronic point-of-sale payment methods-such as credit and debit cards-broughtabout by the increasedavailability of EFT-POSterminals. Econometric analysissupports this contentionand also suggeststhat a heavyreliance on cashfor POStransactions, as in Japan,is associatedwith havinga low rate crime rate. As well, the substitutabilityof credit cards for cash is viewed as also being dependentupon pricing and a country's cultural attitude toward credit. Since per transactioncredit card fees are largely borne by retailersthat accept them for payment,credit cardsare Werceivedby consumersas a low cost, delayedpayment substitute for an immediate cash payment. Electroniccash loaded on chip cards or transmittedthrough open

Countriesthat have the highestincidence of credit card use per person(the U.S. and Canada)also have a culturein whichpersonal credit is quite acceptable.The cultural acceptability of credit cardsis lower in Europe in part becauseof a history of relying on savings(rather than credit) for many consumerexpenditures. This tradition is reinforcedby a historicalEuropean reliance on giros-a methodof immediateand final payment-for consumerbill payments. Giro transactionsrarely involve the extensionof credit while paymentsby check-a 11 computernetworks can develop into a strong substitutefor cash in the future. However, at the momentthe necessaryinfrastructure for the adoptionof e-cashis still by and largeunder development.

IV. The composition and growth of non-cashpayments in developed countries

Non-cashpayment instruments include checks, credit and debit cards,giro payments,and wire transfers.Although wire transfersaccount for the vast majority of the value of non-cashpayments irl countriesthat havesuch a network, they representa very small part of non-cashpayment volume. Thus our focus herewill be on the other non-cashpayment instruments, with wire transfersdeferred until laterwhen the issueof paymentrisk andthe role of the centralbank are addressed.

The compositionof non-cashpayments. The number and composition of non-cash transactionsper persondiffers considerably between Europe, Japan, and the U.S. As seenbelow, the annual numberof non-cashtransactions per personin Europein 1993 was 113. This was lessthan half that of the U.S. (at 292) while Japan'suse (at 39) was just over a third of that for Europe. The exceptionallylow annualuse of non-cashpayments in Japanis consistentwith its relianceon cashfor mostof its point-of-salepayments.

The compositionof non-cashtransactions clearly showsthat the U.S. is the heaviestuser of checksand credit cardswhile Europeis the heaviestuser of debit cardsand giro payments(underlined values).Japan falls somewherein betweenEurope and the U.S.,or lower, in its per personuse of each paymentinstrument (a resultof heavyuse of cash).

Annualnumber and composition of non-cashtransactions per person,1993

Total* Check Creditcard Debitcard Giro Europe 113 32 4 10 67 Japan 39 3 5 .1 31 U.S. 292 234 49 2 9

Datahave been rounded off and may not add to the total.

Paperversus electronic payments. An alternativeway to view the same information is in terms of paper-basedversus electronic payments. Checksand paper giro paymentscomprise paper- based paymentinstruments while credit cards,debit cards, and electronicgiro paymentscomprise electronic payments. As seen below, the U.S. is clearly characterizedas having a paper-based paymentsystem since the numberof paper-basedpayments per personis 5 times that of Europeand

deferredand provisionalpayment-provide for a short(1 to 2 day)extension of creditbefore a customer's accountis actuallydebited. In the U.S.,wire transfers account for 86%of non-cashpayment value but lessthan 0.2% of thevolume. 12

26 timesthat of Japan. Overall, only 20% of U.S.non-cash transactions are electronic. But, because the U.S.has so manymore non-cashtransactions than doeseither Europe or Japan,the actualnumber of U.S.electronic payments exceeds that of Japanand is closeto that of Europe. As seenabove, the U.S.initiates more card-basedPOS payments per person(51) than doesEurope (14) or Japan(5) but this higheruse of electronicpayments by the U.S.at the point of saleis offsetin Europeby greateruse of electronicgiro bill paymentsand disbursements-types of transactionsthat are dominatedby checks in the U.S.

Annualnumber of paperand electronic non-cash transactions per person,1993

Total Paper* Electronic** % electronic Europe 113 44 68 61 Japan 39 9 31 78 U.S. 292 234 59 20

* Checks,paper giro. * * Creditand debit cards, electronic giro.

The tradeoffbetween cash holdings and non-cashpayments. Figure 4 illustrateswhere developedcountries were in both 1987 and 1993 in termsof holding cashand usingelectronic versus paper-basedpayment instruments. The figure is divided into four quadrantsbased on the average percentof non-cashelectronic payments (61.4%) and the averagecash to GDP ratio (4.6%)over 14 developedcountries in 1993. An arrow is drawn from the point eachcountry was in 1987 to the point it movedto in 1993,showing how cashholdings and the useof electronicpayments has changed over this seven-yearperiod.

70 Figure4 updatesand addsto an earlierfigure developed by Yamaguchi(1993). 13

Figure4. Percentof electronicpayments and cash to GDP ratios,1987 and 1993

100% Electronics Switzerland *

I Belgium ~Netherlands

50% [ 80%-- Denmark * ~~~~~~Germany 0 Japan

Cash& 60% Finlandl ia Sweden Electronics

|;60% ------Average61.4% ------

U.K. 40X% * ~~~Franceim

Canada Nry

20% '

U.S. Checks Cash 0% 0% 2% 4% 6% 8% 10% Cashto GDPratio

At either point in time, most countriesfall into the lower left and upper right quadrants, indicatingthat mostcountries either usemostly paper-basednon-cash payments and little cash(U.S., Canada,France, and the U.K.) or use mostly electronic non-cashpayments and have high cash holdings (Sweden,Belgium, the Netherlands,Switzerland, and Japan). In contrast,3 countries use mostly electronicsand havelow cashholdings (Finland, Denmark, and Germany)while 2 countries appearto rely stronglyon paper-basednon-cash payments and also haverelatively high cashholdings (Italyand Norway). Exceptfor Japan,all countrieshave increased their relativeuse of electronicnon- cashpayments and all but the U.S., Canada,and Italy haveat the sametime reducedor essentially maintainedtheir cashholdings as a percentof GDP. Overall,it is clearthat the generaltrend in Figure 4 over 1987-93has been to shiftfrom holding cashto an increaseduse of electronicpayments.

The growth in non-cashpayment instrument use. The level and growth in non-cash transactionsfor 14 developedcountries are illustratedin Figure5. Eachbar in the figure showsthe total number of non-cashtransactions per personin eachyear over 1987-93. All but two European countries(Italy and Switzerland,both of which havehigh cashholdings) initiate a similar numberof non-cashtransactions. The numberof non-cashtransactions in Europeis lessthan half that of the U.S. while those of Japanare only one-thirdof those of Europe. The growth in non-cashtransactions is 14 illustratedby comparingthe heightof the barsover time. Comparing1987 with 1993, the numberof non-cashtransactions, although varying, has risen for all countries.

The shadedportion at the bottom of eachbar indicatesthe number of check paymentsper personin eachyear. Annual check use per personin 1993 is by far the largestin the U.S. (at 234 transactionsper year). France(with 85 transactions),Canada (with 76), and the U.K. (at 50) alsorely on checksbut their use is lessthan one-thirdthat of the U.S. Importantly,for 12 of the 14 countries, checkuse per personhas reached a peakand is falling.

The solid portion in the middle of each bar in Figure5 showsthe annual number of card (credit plus debit) paymentsper personwhile the open portion at the top showsthe number of giro payments. As seen,the numberof card transactionsper personhas been expandingin all countries over 1987-93. Although not shown separatelyin the figure, the annual number of credit card transactionsper personare highestin the U.S.(with 49), followed by Canada(37), and then the U.K. and Finland(who both have 13). Annualdebit card transactionsper personare highestin Finland(at 30), followed by Franceand Denmark(who both have27), and finally Norway (18). Giro payments per person(at the top of eachbar) are much greaterthan card transactions.The largestuser of giro paymentsis the Netherlands(at 128 per year),followed by Germany(124), with Finlandand Sweden tied for third (75).

Figure5. Annualnumber of checks,credit & debitcards and giro transactions per person (14 developedcountries)

300 - o Giro

r Credit anddebit cards 250 * Checks

~200 I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.b150

100......

50

0 Eachcountry: Yearlyfigures 1987 to 1993

Theexceptions are the U.S.and Germany.However, since only 11checks per personare written each yearin Germany,the fact that checkuse here has not fallen is of little consequence. 15

The number of paper and electronicnon-cash transactions for these developedcountries is illustratedin Figure6. Paper-basednon-cash payments (checks and papergiro) are shownin the open portion of eachbar while electronicpayments (credit and debit cards,electronic giro) are indicatedby the solid portion. The numberof electronicpayments per personhas risen in all countriesover 1987- 93. Indeed,electronic payments are the only sourceof growthof non-cashpayments in 11 countries and the primary sourceof growth in two of the remainingthree Japanand France,not the U.S.). Asa result,the prime moverbehind the risein non-cashtransactions for all countriesbut the U.S.has been the rise in electronicpayments. But eventhe U.S. is forecastedto follow this trend by the turn of the centurysince the annualnumber of checkpayments per personis forecastedto startfalling within the next five years. When this occurselectronic payments will be the main reasonbehind the growth of non-cashtransactions in the U.S.as well.

Figure6. Annualnumber of non-cash,paper and electronic transactions per person (14 developedcountries)

300 -

* Electronictransaction t.250 - 0 Paper-basedtransaction

200

150

A

.~50

0 Eachcountry: Yearlyfigures 1987to 1993

Without neglectingsignificant cross-country differences in paymentevolution, certain universal trendscan be identified. First,cash is losingground to non-cashpayments, mostly to electroniccard- basedpayments that are close substitutesfor cashat the point-of-sale.Prepaid cards and smartcards also play a role here but their useis still at an early stage. Second,within the categoryof non-cash payments,checks are being replacedboth by POS card-basedpayments and, to a lesserextent, electronic credit transfersthrough giro or ACH networks. And finally, all paper-basedpayment instrumentsare being replaced by lower cost electronictransactions. Thesetrends in payment instrumentsubstitution have been affectedby a varietyof country-specificinstitutional, economic, and demographicdifferences across countries, which is the reasonwhy somecountries are furtherahead in this processthan others. Thesedifferences are taken up in SectionVI, below, where the historical evolutionof paymentsin Europe,Japan, and the U.S.are discussed in somedetail. 16

V. Largevalue payments in developedcountries and the role of the centralbank

In developedcountries which have largedomestic and/or internationalmoney markets, time- sensitivehigh value paymentsare typically sent over separate,secure, immediate payment, electronic interbankwire transfernetworks. While the U.S. Fedwirenetwork evolvedfrom early telegraphic, final paymentarrangements for settlinginterbank interregional consumer and businesscheck payments usingcentral bank offices,other networks-suchas CHIPS(U.S.), CHAPS (U.K.), and BOJ-NETUapan)- evolved as a more secureand timely extensionof bank clearing house check (non-final)payment arrangements.

The averagevalue of a payment?ver a wire transfernetwork is quite high, averaging$4. million (U.S.dollars) in Europefor 1993, $93.1 million for Japan,and $4.2 million for the U.S. The volume, value,anc 4averagevalue per wire transferfor 9 networksacross 7 developedcountries are shown in Table 2. The value of wire transfersin thesecountries rangesfrom 11 times the country's GDP to 100 times GDP. With such large values being transferredover wire transfer networks, the disruption of economic activity and/or risk of loss associatedwith the failure of participantson thesenetworks is a majorconcern.

Table2. Selectedlarge value wire transfer systems, 1993 (annual data)

Tracions Ownerl Volume Value Averagevalue Valueto Country System manager Settlemnet2 (millions) (US$ billon,) (US$millions) GDP France SAGITTAIRE CB Net 3.9 19,204 4.9 15 TBF CB RTGS 12.0 52,412 4.4 42 Germany EAF CB Net 10.9 78,158 7.2 41 Italy ME CB Net 1.9 10,846 5.7 11 Switzerland SIC CB+ B RTGS 67.4 23,097 0.3 100 U.K. CHAPS B Net 11.0 35,353 3.2 37 Japan BOJ-NET CB Net & RTGS 3.8 353,818 93.1 84 U.S. Fedwire CB RTGS 69.7 207,630 3.0 33 CHIPS B Net 42.2 262,256 6.2 41

1. CB- CentralBank; B - Banks;2. Net- Multilateralnet settlement;RTGS - Realtime grosssettlement Source:BIS: PaymentSystems in the C;-1OCountries, 1994

12 Thisfigure represents an averageof paymentvalues over 5 wire transfernetworks: CHAPS (U.K.), SAGITTAIRE and TBF(France), EAF (Germany), and ME (Italy). The SIC network in Switzerlandtransfers both large and smaller valuepayments (with an averagevalue of only $0.3 million) and was excludedbecause data are not reportedseparately for largevalue payments.

13 Japanis representedby BQ)-NETwhile the U.S.figure is an averageof CHIPSand Fedwire. 14 Other largevalue wire transfernetworks exist and are illustratedin Bankfor IntemationalSettlements (1994), Table lob. 17

Settlementfailure and policies to contain systemicrisk and daylight overdrafts. The risks associatedwith a settlementfailure on a wire transfernetwork differ dependingon how paymentsare settled. On a net settlementnetwork, payments are sent and receivedduring the businessday and only the net positionof eachparticipant is settledat the end of that day (or the next day). On a real time grosssettlement (RTGS) network, each payment sent is settledwhen it is made. Of the 9 major wire transfernetworks outlined in Table2, 5 haveemployed net settlement(SAGITTAIRE, EAF, ME, CHAPS,CHIPS) while 3 are RTGS(TBF, SIC, Fedwire). One networkhandles both net settlementand RTGSpayments (BOJ-NET).

Until recently,the failure of a participanton a net settlementnetwork would typically haveled to the unwinding or reversalof all of that day's paymentsand receiptsby the failed participant. The remainingparticipants would then be faced with trying to settle a recalculatednet debit or credit position-recalculatedafter excludingthe paymentsand receiptsof the failed participant. This would not be a problem if receiversof fundsdid not usethem-by releasingthem to customers-untilafter the day'stransactions were settled. However,because of the float costof waiting until the next morningto usethe fundsreceived, receivers typically usefunds before the transactionsare finally settled.

If there is an unexpectedsettlement failure, then it may not be possiblefor other participants on the network to obtain sufficient funds to settle their recalculatednet debit or credit positions. Indeed,in simulationsperformed for the CHIPSnet settlementnetwork using actual paymentdata (Humphrey,1986), the increases(decreases) in other participants'net debit (net credit) positionsin many casesexceeded their capital levels. Facedwith either (a) settlingtheir recalculatedpayment positionby effectivelyselling or pledgingthe entirevalue of their capital or (b) not settlingand having their own paymentsand receiptsthat day unwound from the settlement,it is clear that the logical choice would be (b).

The possibility of a domino-like seriesof settlementfailures is termed systemicrisk. In simulationsusing CHIPS data, the failure of one randomlychosen participant led to the failureto settle of almostone-half of the over 100CHIPS (domestic and internationalbank) participants. It also led to the elimination of around one-third of that day's paymentsof $885 billion. However, similar simulationsfor an Italianwire transfernetwork led to a muchsmaller number of additionalfailures and a correspondinglysmaller reduction in paymentvalue (Angelini,Maresca, and Russo,1996). As the value of paymentsover the Italian network were much smaller, so were the participant net debit/capitalratios. Thus systemicrisk is smaller when a country does not have a large and well- developeddomestic money market (so interbanknet debits are not absolutelylarge) and when a country's bankingsystem is highly concentrated(so that bank capital levels are more likely to cover net debits).

One way to effectivelyeliminate systemic risk on a net settlementnetwork is to haveeach participant post collateral sufficient to cover their net debit position. Paymentflows would be monitored in real time and paymentsthat would push a participant'snet debit above its collateral value would be rejected. If therewere a failureto settle,the collateralwould be liquidatedto obtain a centralbank loan to makesettlement. CHAPS, established in 1984,plans to haveall net debitsfully collateralizedby 1997. Participantson CHIPS(established in 1971)have since 1990 postedcollateral sufficient to cover the single largestnet debit permittedto occur on that network. The CHIPS approacheliminates systemic risk if there is a failure to settleby one participantwhereas the CHAPS 18 approacheliminates systemic risk if thereare multiplefailures to settle. Both CHIPSand CHAPSwere initially set up as net settlementnetworks to clear largevalue checkselectronically and are owned by largebanks in New York and London,respectively.

Thereis no systemicrisk on a realtime grosssettlement wire transfernetwork. This is because eachpayment is finally settledwhen it is made,rather than net-settledat the end of the day. However, if a participantcan send paymentswithout having sufficientfunds in their account,then a daylight overdraftis created. The risk on a RTGSnetwork concernsthe possibilitythat a participantwith a daylight overdraftmay fail, creatingcredit risk for the operatorof the network who guaranteesthe finality of eachpayment as it is made.

RGTSnetworks are operatedby central banks who can face credit risks associatedwith daylightoverdrafts. Exceptfor SICin Switzerland,the networksin Table2 permit daylight overdrafts but control credit risk by requiring that collateralbe availableto cover these exposures. This also servesto reducemoral hazard-typebehavior by paymentparticipants since their collateralis at risk if they fail to settle. In Germanyand the Netherlands,daylight overdrafts with the centralbank mustbe fully collateralized. In France,overdrafts may be partiallyuncollateralized without a fee beingcharged and net debit capsestablish a ceiling on the overdraftsparticipants may incur. Collateralizationlimits centralbank risk to the differencebetween the facevalue andthe marketvalue of the collateral.

The SIC RTGSsystem in Switzerland(established in 1987) representsa different RTGS arrangementand avoidsdaylight credit risk by rejectingany paymentrequest which would lead to a daylight overdraft. If sufficientfunds are not available,payment orders are put into a queue until the paymentrequest can be covered,either by borrowing or waiting until sufficientfunds have been receivedduring the day. Queued paymentsare sent and settledon a first-in-first-outbasis (Vital, 1990). Sometimes,if a bank's balancewould not cover the entire value of a large payment,the payment is divided up and sent in pieces as sufficientcovering funds become available. This procedurereduces the possibility of a payments"gridlock", as could conceivably occur if eRough sendingparticipants chose not to hold an idle balancessufficient to covertheir paymentactivity.

Fedwire was establishedin 1918 as a telegraphicsystem and has always been an RTGS network. Beforethe 1980s,account monitoring had always been at the end of the day. As total paymentvalues were smaller and reserverequirements were higher than today, there was little likelihood of a daylight overdrafteven thoughbalances were not monitoredintraday. However, the exponential growth in paymentvalues associatedwith a dramatic expansionof domesticmoney marketactivity led to dayl)ghtoverdrafts and stepswere takento contain their growth and the credit risk to the central bank. Currently, Fedwire permits overdrafts(up to a given multiple of a

While thepositive opportunity cost of an idle balanceprovides an incentiveto hold fewerbalances, the lossof businessassociated with an inability to make a payment when ordered by a customerhas, in practice, apparentlybeen a strongerincentive and gridlockhas not beena problem.

16 Sincepayment values associated with the productionof goodsand servicesthat underliesGDP shouldonly grow in roughproportion to GDP,the expansionof purely financialtransactions associated with moneymarket activity is the proximatecause of the exponentialgrowth in wire transferpayment values. Although initial paymentsmay take anotherform, such as checkswritten to clearingcorporations or depositories,Fedwire is usedto ultimatelysettle almost all stock,bond, option,and futuresmarket trades in the U.S.and theseactivities 19 participant'scapital) and in 1994 assesseda fee for the provisionof this daylight liquidity. At present, the fee is low enoughso that the centralbank is the only direct supplierof daylight liquidity. If the central bank fee were sufficientlyhigher, however,then somethinglike an intra-py interbankfunds marketwould likely arise,similar to existingmarkets for overnightinterbank funds.

Japan'sBOJ-NET large value network,established in 1988,can accommodatepayments to be net settledas well as RTGSpayments. Like the SwissSIC system, the RTGSpart of BOJ-Ij`ETdoes not permitany daylightoverdrafts and transfers without sufficientcovering funds are rejected.

In Europe,the implementationof RTGSsystems started in the late 1980s. Somesystems were devotedonly to largevalue interbanktransfers, while otherswere designedto handleboth retail and largevalue payments. For example,the Netherlands'FA was established in 1985 and Sweden'sRIX in 1986. However, the biggest large value systems-CHAPSin the U.K., EAF in Germany, and SAGITTAIREin France-continuedto operateon a net settlementbasis and only recentlyhave started to implementRTGS procedures. A phasedimplementation of RTGSon CHAPSbegan at the end of 1995 and EAFis being modified(new EAF2) to offerearly and intradaysettlement finality as in a RTGS system. In France, paymentsexchanged over SAGITTAIREand three additional net settlement networksare being shiftedto the new TBF RTGSsystem (established in 1992). Overall, the trend in Europetoward RTGSarrangements has been stronglyinfluenced by the EuropeanCommission which declaredthat RTGSshould be adoptedon each country's largevalue paymentnetwork in order to eliminat1$systemic risk and thus facilitatethe harmonizationof large value paymentsystems across Europe.

The role of the central bank. Central banks in developedcountries all provide for the settlementof differenttypes of non-cashpayment instruments: checks, credit cards,debit cards,and wire transfers.Central bank settlement of giro transactionsis unnecessarysince funds are merely2 eing transferredbetween accounts held within a single organizationand so settlementis internal. Of experiencedrapid growth since the 1970s. Transfersof U.S. governmentsecurities, which is a book-entry delivery-against-paymentsystem, uses the securitytransfer portion of Fedwire.See, for example,Stehm (1996).

17 Bank and financial market responsesto charginga fee for daylight overdraftsare discussedin Humphrey (1995),Richards (1995) and Hancockand Wilcox(1996).

TB On BOJ-NET,RTGS transactions represent only 3% of transactionvolume. Designatedtime transactions, which are equivalent to transfersover a net settlement network, generate 97% of the volume. These transactionsoccur during the day (ratherthan end-of-day)and are usedto settlecheck clearings, ACH (Zengin) clearings,foreign exchange clearings, and interbanktransfers.

19 In a follow-up documentpublished in late 1993, the Working Group on EC PaymentSystems to the Committeeof the ECCentral Bank Governors noted that ¶...everyMember State should have as soon as possible an RTGSinto which as many large valuepayments as possibleshould be channelled'. It wasacknowledged, however,that paymentnetting arrangements may continue to existparallel to RTGSsystems provided that (i) they settleon a same-daybasis, and (ii) fully meet the Lamfalussystandards published earlier for net settlement networks(modeled on the arrangementsoriginally developed for CHIPSin the U.S.). 20 Theexception would be a transactionbetween (say) a memberof a postalgiro and a memberof a bankgiro in the samecountry which would likely requirecentral bank settlement. 20 course, if a giro did not have some type of centralizedaccounting, then funds transfersbetween customersof separategiro officeswould haveto be settledexternally through the centralbank. The samewould be true for fundstransfers between different branch offices of a singlebank which did not centralizeits accountinformation, a not uncommonsituation in emergingmarket economies.

The major point of departureregarding the role of a central bank in a developedcountry's paymentsystem concerns the extentto which the centralbank provides payment processing services for other than largevalue wire transfers.Central banks in Europeand Japan do not provide payment processingfor checks,ACH, credit cards,debit cards,or giro paymentswhile the U.S. centralbank competeswith banks,bank 5Jearing houses,and non-bankpayment processors in the processingof checkand ACH transactions. The reasonfor the differentroles played by developedcountry central banks,and the reasonswhy Europe,Japan, and the U.S. havecome to rely on a different mix of paymentinstruments, is largely relatedto their different historicalexperiences. The main elements concerningthis differentialexperience are now outlined.

VI. The historicalevolution of paymentsin Europe,Japan, and the U.S.

The paymentsystems of all developedcountries have passed through a stagewhere barterwas replacedby coin (gold,silver, copper)and then was itselfalmost totally replacedby currency. Europe, Japan,and the U.S. haveall experienceda similarevolution up to this point. However,beyond this point the evolutionstarts to differ.

For POSpayments, Japan clearly remainsa heavy userof cashwhile Europehas reducedits cashuse by first usingsome checks at the point-of-saleand more recently has expanded its useof debit cards. The U.S. hasexperienced a more rapid and greaterreduction in cash useat the point-of-sale, relying heavilyon checksand, morerecently, an increaseduse of creditcards.

For consumerbill payments,payroll disbursements,and financialtransactions, Japan, Europe, and the U.S. all have a long history of initially using cashcombined with specializedpaper drafts. Subsequently,payment arrangements evolved-at different ratesfor different instruments-awayfrog cashand into a heavyreliance on paperdrafts (checks, bills of exchange,and papergiro payments).

21 ACH transactionsare the electronicequivalent of a papercheck. In the U.S.,ACH transactionsare mostly debittransfers but somecredit transfers also exist (and are thus equivalent to a giropayment in Europe).With respectto credit and debit card processing,banks own or supervisethe processingof bank-basedcard transactionswhile non-bank processors typically handle the processing of non-bank(e.g., retail store, gas station) creditcard payments.

22 Whilea checkand a bill of exchangeare similar, there is onecrucial difference. A checkis an unconditional orderin writingby a person(the payor) to thepayor's bank signedby thepayor, requiring the bankto paya certainsum on demandto a payeelisted on the checkor to the bearerof the check.A bill of exchangeis the samething except that the payor orders another person-rather than a bank-topay funds to thepayee at a given futuredate. Forexample, if in theprocess of tradefirm I owesmoney to firm 2, andfirm 2 owesmoney to firm 3, a bill of exchangecan be draftedand signed by firm 2, drawnon firm 1,for firm 1 to transfera givenamount ona givendate to firm 3. Insteadof transferringdeposits, firm 2 (thepayor) is transferringtrade credit from firm 1 (the'bank") to firm 3 (thepayee). 21

More recently,the evolution has been away from paper and into electronics(electronic giro, direct debits,direct deposit,and wire transfers).

Althoughthe evolutionof paymentsamong these countries has been qualitatively similar, there are largequantitative differences. Simply put, the key distinctionin paymentinstrument use between Europe,Japan, and the U.S. is that Europerelies on giro payments,Japan relies on cash,and the U.S. relieson checks. Thesedifferences are mostpronounced at the point-of-salefor cashand checksand for consumerbill paymentsfor giro and checks,but also differ to some degree for businessand financial payments. In what follows, the primary institutional,legal, and cultural reasonsfor these observeddifferences in paymentinstrument use acrosscountries are briefly outlined. Historical reasonsfor differencesin the role of the centralbank in thesecountries are alsonoted.

The evolution of paymentsin Europeand the establishmentof the giro. Sixtypercent of all non-cashpayments in Europetake the form of credit transfergiro paymentsby individuals,businesses, andcentral and local government.As seen in Table3, giro paymentsaccount for over 80O/aof all non- cash paymentsin Germany,the Netherlands,Sweden, and Switzerland;in 9 of the 11 European countriesshown, giro transactionsgenerate over half of non-cashpayments. The remainingpayments are divided between cards and checks and, in 6 of the 11 countries,the share of card-based transactionsexceeds that for checks. Overall, only in 3 countries(the U.K., France,and Italy) do checksaccount for more than one-fifthof non-cashpayments. Thus, overall, Europehas a giro-based non-cashpayment system.

The historicalreason for the dominanceof credittransfer payments in mostEuropean countries is the early establishmentof giro systems,to which postal savings organizationsimportantly contributed. This developmenthas its roots in the bankingstructure and activitiesof Europeanbanks. In the early 1800s, bankingin Europewas conductedby small bankinghouses (banker families) that did not acceptpublic depositsto a significantextent. Indeed,the developmentof modern banking organizationsthat collectedsavings from the generalpublic beganin Europearound the mid 1800s when large, limited liability commercialbanks were establishedin many countries. The resulting increasein deposit-takingactivity led to the gradualincrease of accounttransfer based payments at the expenseof currencyand coin. Other credit institutions,savings banks, credit cooperatives,as well as postalbanks were establishedduring this time and lateraround Europe. These institutions have since developedtheir activitiesto closely resemblethose of commercialbanks in continentalEurope and Nordic countries. In Europe,the postaladministration began to channelpayments as postalpayment orders during the 1800s. In addition, the postal systemstarted to offer a major non-cashsavings service(postal savings accounts) to the generalpublic. Postalsavings institutions of one form or anotherwere establishedin all elevenEuropean countries under study.

During the late 1800sand early 1900s,postal savings institutions, commercial and savings banks,and credit cooperativesset up their own separategiro-systems. In somecountries, postal giro systemswere developedfirst and only later did commercialbanks and other institutionsfollow, in order to capturea share of consumerdeposits by providing paymentservices. After a phase of competitionbetween these separate systems, cooperative agreements between postal and bank giros wereestablished in mostEuropean countries (Bank of Finland,1993). The establishmentof "wagesto banks"systems in the late 1960s and early 1970s, where employers pay wagesand salariesdirectly to employees'bank accounts,strongly increasedthe use of bank and postal account transfersas a Table3. Percentageshares of non-cashretail payment instruments: 1983.93 (all figuresare percentages)

Debitand credit card payments Giropayments giropayments Total Check POScard Othercard ... of whichcredit Paper-based Electronic debits' Other giropayments payments payments payments cardpayments' giropayments Direct 1987 1993 1983 1987 1993 1983 1987 1993 1983 1987 1993 Country 1983 1987 1993 1983 1987 1993 1983 1987 1993 1983 1987 1993 1983 40 1 0 5 7 9 7 53 58 52 61 67 Belgium 45 33 16 na 5 14 0 1 3 na 15 17 29 17 32 37 42 23 26 29 89 91 89 Germany 11 9 8 na 0 1 0 1 2 na 99 87 34 42 6 5 16 16 22 20 59 61 78 81 88 Netherlands 22 19 8 na 0 3 0 0 1 na na na 12 9 6 6 9 15 12 13 15 30 32 36 U.K. 61 58 43 na na 11 8 10 10 na 100 53 na 66 50 na 1 3 na 7 18 na 74 70 Norway na 19 9 na 6 18 na 1 2 na 17 8 41 46 31 na 1 '2 na 34 29 81 80 62 Finland 18 9 1 na 2 28 na 9 8 na 82 30 na na 20 na na 5 na na 57 na na 82 Sweden na na 6 na na 8 na na 4 na na 16 2 2 1 6 9 12 8 15 16 15 27 29 France 83 65 54 na 8 15 2 0 2 na 1 2 10 43 42 1 2 3 3 4 12 14 49 58 Italy 86 51 38 na 0 1 1 1 4 na 98 82 na 0 0 na 3 2 na 84 81 na 88 84 Switzerland na 9 3 na 0 6 na 3 7 na 90 65 na 0 0 na na 11 na na 44 na 60 55 Denmark na 39 18 na 1 16 na 0 10 na 28 3 .0 0 0 1 0 1 0 1 2 1 1 3 U.S. 92 86 80 na 0 1 7 13 17 na 99 95 0 0 0 na 2 4 na 1 6 2 3 10 Canada 91 73 59 na 0 2 7 24 29 na 100 93 na na na 56 52 35 na na na 68 79 80 Japan 19 11 7 na 0 0 10 10 13 na 100 99

(1995),Bank of Finlandand Finnish Bankers Association Source: BIS: PaymentSystems in theG-10 Countries, European Monetary Institute: Blue Book Addendum Notes1: Thepercent credit card payments are of all (debitand credit) card payments. 2: Directdebit figures for Norway, Switzerland, the U.S.and Canada are for 1992. 23 paymentmethod in Europe. The existing giro infrastructureoffered ready and safe solutions and capturedthe largestpart of this growingpayment volume.

Giro paymentsmoving to electronics. The establishmentof wagesto bankssystems (or direct deposit of payroll)was also an original impetusfor electronicpayment data transmission.Recurrent wage paymentswere originally initiatedby sendingmagnetic tape data mediato banks. Now, in most countries,the bulk of all paymentorders are transmittedvia telecommunicationlinkages. Businel paymentshave moved rapidly to electronics,primarily through useof direct terminal links to banks, while giro paymentorders from individualsare still often paper-based.This is slowly changingas homebanking terminals are being installedand ATMs can increasinglyhandle account transfers and depositsin additionto cashwithdrawals and balanceinquiries.

The clearingsystems for credit transfergiro'Rayments started shifting to electronicsin the early 1970s,in somecountries even in the late 1960s. More recently,the popularityof electronicbill payments(through preauthorized direct debits) has significantly increased in manyEuropean countries. The shareof direct debitsin total non-cashpayments (Table 3) hasrisen most in the mostgiro-oriented countries(Germany and the Netherlands)but alsoin traditionallycheck dominated countries (the U.K. and France). The direct debit systemoffers the payor an easy and safe way to effect recurrent (electricity,gas) utility, telephone,insurance, rent, and other paymentson a specifieddue date. The one drawbackof direct debitsfor consumersis that, becausethe payeeinitiates the payment(albeit at a specifiedtime eachmonth), the payorgives up control over when the bill is paid. But this lossof payorcontrol is a benefitfor the payeesince cash flows are morecertain with a direct debit than with a payor-initiatedpayment. Direct debitscan alsosignificantly reduce the amountof papercirculating in the paymentsystem. Sinceelectro 9 ic paymentsare usually lesscostly than are paper payments, resourceswill be moreefficiently used.

23 Companyterminals (computerized cash management systems) offer an interactiveservice via an on-linelink to a bank'spayment service system. For examplein Finlandthe first terminalswere introducedin the late 1970s. Currently,banks offer a wide rangeof sophisticatedpayment services on-line to theircorporate customers. 24 Homebanking is still in the very early stagesof development,but hasstrong potential to increasein the future as it offersconsiderable convenience and control benefitsfor consumerbill payments. Currenthome banking servicesuse either push-buttontelephones, videotext service (e.g., Teletalin France),or home computersto accessa bank'scomputer and initiate a fundstransfer.

25 In Italy this progresshas beenmore delayed. In 1983the membersof the CIPA(Italian Interbank Convention for AutomationProblems) started to progressivelydevelop automated payment management techniques in order to reducethe volumeof paperpayment media.

26 Despitethe historically large shareof check transactionsin Franceand the U.K., check truncation and electronicpresentment of paymentinformation is minor. Checktruncation is one way in which the consumer could still usea checkbut the paymentinformation it contains-theamount of the check, the payor and payee depositand bank accountnumbers-would be readand capturedat the bank wherethe check is first deposited and transmittedelectronically to the paying institution. In France,only about 6% of checksare handledthis way while in the U.K. truncationhas yet to be introduced. In Europeancountries with a historicalreliance on giro payments,checks have never played a major role as a domesticpayment instrumentand efforts to modernizepayment arrangements for ordinarychecks have not beenconsidered a worthwhile investment(CEPS, 1994). 24

Thesubstitution of cardpayments for cashin Europe.Electronic POS card-based transactions with real-timelinkages to accountdata representa new stagein the evolutionof paymentsin Europe and combine the finality advantageassociated with currencyuse (legal tender)with the efficiency advantagesof electronic accounttransfers. This is the casewit) on-line POS terminalsthat offer immediatesettlement of transactionsat the momentof exchange . ThesePOS payments are close substitutesto currency in the sensethat the party acceptingpayment takes no risk since funds are effectivelytransferred to the payeewhen the terminal is activated. On-line card-basedpayments thus enhanceefficiency by reducingtransaction costs, reduce the time interval beforea paymentorder is creditedto a payee'saccount, and limit the amountof paperin the paymentsystem (Padoa-Schioppa, 1986).

In Europe,both the number of POS terminalsaccepting debit and/or credit cards and the numberof transactionshandled are, on a per personbasis, greatest in Finlandand France. Denmark, Norway, Belgiumand the U.K. are also relativelyadvanced in this regard. Finlandand Francein particular demonstratethe power of cooperationbetween banksas a way to bring about a rapid diffusion of electronic payments. Cooperationin the establishmentof infrastructure,the wide acceptabilityand compatibilityof cards issuedby different institutions,and the seeminglyeffective competitionamong institutionsfor customershas permitted the rapid diffusion of on-line card-based POSpayments in France(Kearney, 1993) and Finlandduring the 1980s.

Reasonsfor giro dominancein Europe:banking concentration, nationwide networks, and cooperation.A concentratedbanking industryand the nationwide branch networksof banks and postalsavings institutions made it possibleto establishgiro paymentsystems in Europebased on credit transfers.Within the postalbank network, for example,it wasonly a relativelysmall step to bring in businessaccounts and offer first local, then regional,and finally nationwidebill paymentservices to savingsaccount holders (Thomson, 1964). With centralizedaccounting, on a local, regional,and then a nationalbasis, it was possibleto determinewhether a payorhad sufficientsavings in his giro account beforea paymentwas made. As paymentsare rejectedif funds in an accountare insufficient,credit risk to the payeedoes not exist. Centralizationof depositoraccount information made the shift from paperto electronicgiro transactionsin the 1960sand 1970san internalissue, similar to that which has occurredwithin-but not between-banksin the U.S.during the sameperiod. Comparedto European countries,the U.S. is geographicallymany times larger, has a much more unconcentratedand decentralizedbanking system, and has no nationwidepostal payment service (other than little used postalmoney orders). In this environmentchecks have a clearadvantage over credit transfers,at least beforethe recentdevelopment of low costelectronic communication technology.

Bankingconcentration has also played a major role in Europeby facilitating cooperation betweenbanks and other institutionsin the paymentbusiness. In a concentratedmarket, it is much easierto reachan agreementto establisha jointly owned giro systemthat, becauseof centralizeA clearing, is a lower cost alternativethan use of geographicallydispersed local clearing houses.

27 SomePOS systems do not operateon-line and allow the useof creditor deferredpayment. 28 In Europe,restrictions on bankgeographical dispersion within countries have been scarce (Gual and Neven, 1992)and antitrustpolicy hasbeen more behavior-oriented, trying to detectcases where firms exploit their dominantmarket position", rather than limiting market concentration itself (which has been the U.S. approach). 25

Cooperationamong banks and other providersof paymentservices, actively coordinated by banking associations,first took place in Europe among giro organizations. Cooperationhas since been extendedto other paymentsystems and instruments,particularly ATM and card-basedPOS terminal systems.This hasbeen relativelyeasy since other electronicinstruments can take advantageof the network infrastructurealready in place for electronicgiro payments. This supply-sidecooperation is an importantfactor in the swifterdiffusion of electronicpayments in Europethan in the U.S.

There are two primary reasonswhy cooperationencourages the installmentof electronic paymenttechnology (Vesala,1993, Chapter 4.4). First, there are important scale economiesin processingelectronic payments. The fixed costof settingup the necessarycommunication networks, links, and terminal systemsrepresents the majorityof the operatingcost, leadingto falling marginal andaverage cost as transaction volumes rise. As well, the additionof participantsin electronictransfer systems(or, equivalently,clearinghouse arrangements) reduces both the averageand marginalunit costsof processingpayment transfers for the group as a whole. Thesenetwork economiesreduce transactioncosts because it i5iless expensive to movefunds between accounts within a singlenetwork than it is betweennetworks.

Second,consumers prefer payment services that offer the greatestacceptability among retailers or have the greatestcompatibility with the largest number of POS or ATM terminals. Wide acceptability and extensive compatibility provide greater user convenience and reduce user transactioncosts. Hence,there exist significantpositive network externalitiesin modern payment systems(Neven, 1993). By supportingjoint networks,banks can meetthis demandfor compatibility and, also, small bankscan continue to provide paymentservices for their depositors(provided that accessto the networkis fairly priced). Insteadof proprietarypayment systems, payment pricing and servicequality issuesform the parametersbanks' increasinglyuse in competingwith one another (Commissionof the EC, 1992;Yamaguchi, 1993). Reinforcingthis trend, the EurjpeanCommission hasestablished guidelines for cooperationand competition in the EuropeanUnion.

Theevolution of paymentsin Japanand the benefitsof cashuse. Japan, in contrastto Europe and especiallythe U.S.,continues to rely heavilyon cashat the point of sale. While somemay view this asan indicationof "insufficientevolution" of the Japanesepayment system, it is more instructiveto insteaddetermine why Japandoes not need to rely on non-cashpayments as much as occurs in Europeor the U.S. In this regard,it is usefulto review the advantagesof cashover non-cashpayment instruments,such as checks,bills of exchange,or promissorynotes.

29 Paymentcosts can be furtherreduced when customers are inducedto shiftfrom paper to electronicgiro payments.In someEuropean countries, both bankand postalgiro systemsare currently: (a) reducingthe proportionof theirpayment cost covered by float whileraising the portion recouped through direct transaction fees;and (b) reducing the cross-subsidization among different payment instruments by makingtransaction fees of eachinstrument more closely resemble each instrument's average variable cost. Thesechanges should shift demandtoward lower cost electronic payments.

30 Cooperationis toleratedprovided that is doesnot reducemarket competition among institutions. In addition, thereshould be no attemptto blockentry to sharedpayment systems or to protectparticipants from outside competition. Eitherevent would be regardedas a violationof antitrustrules of the EuropeanUnion (Commissionof the EC,1992). 26

Thereare at leastthree advantages to the useof cash. First,currency and coin representfinal paymentand thus are more readilyaccepted than manynon-cash instruments which do not havethis attribute. Non-cashinstruments are provisionalbecause they may not be paid when presented(due to insufficientfunds in a payor'saccount for a check,refusal to pay a bill of exchange,or defaultby the makerof a promissorynote). Thus the creditworthynessof the payor needsto be determinedwhen paymentis offeredusing a non-cashinstrument while this credit risk doesnot exist with cash. Second, cashcan be immediatelyreused while collectionof non-cashinstruments is often delayedwhen drawn on non-localpayor institutions. Third, cashis divisibleand can be usedto completeother transactions of differingamounts while checksor bills of exchange(when signed over to anotherreceiver) are only poorly adaptedfor transactionsof a differingamount.

Under certain circumstances,the use of cash can be a disadvantage. First, cash is a disadvantagewhen the amountneeded for a paymentis very large. Second,cash is not a convenient paymentinstrument when paymentis requiredat a substantialdistance from the payor, which could be acrosstown or in a differentcity ratherthan nearby. Third, and perhapsmost important, holding a large idle balancefor cashpayments can, in countrieswVtp1 a relativelyhigh crime rate, increasethe risk of lossand/or personal safety to unacceptablelevels. Finally,when interestrates are high, the opportunitycost of cashcompared to depositmoney is high in termsof interestforegone on idle cash holdings.

Consideringthese advantages and disadvantages, cash payments would be preferredover non- cashinstruments for e5sentially local transactionswhere the valuesinvolved are not overly large and the crime rate is low. Indeed,if therewas little risk of theft or loss,there would be little needto use non-cashinstruments at the point-of-salebecause the disadvantagesof cashuse-large values, distance, and loss-would be minimal while the advantages-paymentfinality, timeliness,and divisibility-would dominate. Conversely,when the disadvantagesto cashuse loom large,non-cash instruments will be favored. In Japan,this occurstoday for bill payments(inconvenient distance) as well as business disbursementsand financial payments(large values and inconvenientdistance). In other countries, particularlythe U.S., a higher crime rte is an incentive to replace all but the lowest value cash transactionswith non-cashinstruments.

Threestages in the evolution of Japanesepayments. The evolution of the Japanesepayment systemhas undergone at leastthree major events. First,with the Meiji Restorationin 1868,there was a strong,coordinated effort by businessand governmentto establisha more modernbanking system.

In addition, cashdoes not leavea paper trail, makingit more difficult to keepaccurate, verifiable records. Althoughthis is a cost for some,particularly for legalbusiness transactions, it is a benefitfor otherswho wish privacy. While the use of cashhas an opportunitycost which is typicallya functionof the interestrate, this cost needsto be comparedwith the pricesof non-cashinstruments faced by usersto determinewhat the relative price incentivesmay be regardingpayment instrument use.

32 Japanis a relativelysafe country and this permitscash to be heavilyused in transactionsat the point of sale (Federationof BankersAssociations of Japan,1 994a, p.95 and 1994b,p. 1).

33 Statisticalanalysis indicates that a higherincidence of violent crime is positivelyand significantlyassociated with the increaseduse of non-cashpayment instruments across the 14developed countries covered in this paper (Humphrey,Pulley, and Vesala,1996). 27

The numberof banksand clearinghouses was expanded,a central bank was established,and new financial marketswere developed. Over 150 nationalbanks, dispersed nationwide, were established by 1878. Initially, thesebanks were able to issueprivate bank notesor currencywhich beganto replacecoin in transactions.As it turnedout, the expansionof privatebank noteswas too rapid and led to inflation. This problem was addressedby the establishmentof a centralbank which, by law, becamethe sole issuerof currencyso the growth of the money supply could be better controlled. With this development,private nationalbanks were forced to changefrom note-issuinginstitutions into depositbanks, leading to an expandeduse of paperdrafts (checks) for businesstransactions. This wasfacilitated by the growingcustom of payingchecks, bills of exchange,and other promissorypaper instrumentswithin a due date, rather than-aswith inter-enterprisebook credit arrangementswithin business(keiretsu) groups-leaving the paymentdate vague or unspecified.

Althoughlocal clearinghouses existed for paperdrafts (checks and bills of exchange)in major tradingcenters prior to the establishmentof the Bankof Japan(BOJ), their expansionwas facilitated by the openizggof regionaloffices of the BOJand the ability to settleclearing house net debits usingBOJ accounts. This permittedtrade to expand,both absolutelyand geographically.A nationalpostal systemwas establishedin 1881 and, in addition to providingfor the fastermovement of paper drafts between cities and trading centers, it also offered a postal money order service for consumer payments. However,the governmentlimited postalmoney orders to low valuesso as to encouragea bankcheck/draft system. As a result,postal money orders were usedprimarily for local bill payments while checksand bills of exchangewere usedfor largervalue businesspayments both locally and betweentrading centers. More flexibility in terms of paymentvalues was obtained later with the establishmentof a postalgiro in 1906. At present,the postalservice-which is exemptfrom corporate income tax-operatesa nationwidegiro systemwith 24,000 offices,versus 15,000 for banks. The postalgiro providesthe samepayment services for consumersas do banksand includesaccess to cash through offices and ATMs, paper-basedand electronic bill payments(direct debits)and electronic depositof payroll (directdeposit).

A secondmajor event in the evolutionof the Japanesepayment system occurred in 1943 with the establishmentof the NationalCentralized Domestic Exchange Settlement System (NCDE) operated (at least initially) by the central bank. Previously,reciprocal holdings of interbankcorrespondent accountshad been usedto clear non-cashpayments between banks and among clearinghouses in different cities and trading centers. When the number of banks is very small, clearing through correspondentbalances is workableand reasonablyefficient (Tsurumi,1994b). As more banksare added,however, this processbecomes quite cumbersome and inefficienq 3sincethe numberof balances to be held and maintainedgrows much faster than the numberof banks. Eventhough the numberof commercialbanks in Japanfell from 1,800in 1922 (prior to the GreatDepression) to only 78 in 1944, a centralizedclearing arrangementwas still preferredto one based on reciprocalcorrespondent

34 TheBank of Japan,during the 1890s,offered both clearingand settlement services in an attemptto gain more control over the financialsystem. This was successfulin a few largecities but, at the turn of the century,the BOJ reversedcourse. It withdrew from clearingpaper drafts in competition with clearing housesbut still providedsettlement. For moredetail, see Tsurumi (1991 and 1994a).

35 If all correspondentbanks held reciprocalclearing balances with eachother, then for n banksthere would be n(n-1)balances (which risesfaster than the changein n). 28 balances. Pressureto adoptcentralized clearing resulted from a desireto conserveincreasingly scarcelabor and materials resources in a wartimeeconomy. This pressure also forced the centralbank to operatethe NCDEsystem but, once the war was over, operationwas shiftedto the banks themselveswith the central bank only providingsettlement.

Asa clearingentity, the NCDE was a compromisebetween a creditand a debittransfer system. Theclearing rule adoptedwas that a payor'sbank had to havedeposited the amount to be paidin its centralbank settlementaccount-but not necessarilyto havetransferred these funds to the payee bank'ssettlement account-before the payee's bank would be ableto providefunds to the payee.In a credit transferarrangement, the transferbetween bank settlementaccounts would have occurred beforefunds could be releasedto the payee(eliminating systemic risk). In a debittransfer system, the payeecould receive funds from the payee bank before the payor's bank had even funded its settlement account,much lesstransferred funds to the payeebank's settlement account (creating systemic risk). TheNCDE clearing rule (whenfollowed) does not createsystemic risk as longas thefunds deposited in thepayor's settlement account can not be removed(or attached by someother creditor in theevent the payorfails) prior to settlementof the transaction.If theseconditions are met,then the NCDE clearingarrangement is similar to a fully collateralizednet debit and would notcreate systemic risk.

The third elementin the evolutionof Japanesepayments was the developmentof a nationwide,privately operated, electronic payment system (Zengin) in 1973. The Zenginsystem currentlyconnects 3,778 financial institutions (commercial banks, specialized banks, credit unions, and mutualfunds) and their 45,000branches. The purpose was to replacethe paper-basedNCDE systemwith a moreefficient electronic system. This was made possible in partby thefact that Japan is a smallcountry so the communication costs (laying cable, etc.) were relatively inexpensive. The bank- operatedZengin system processes only about26% morepayments than does the postalgiro. Bot networksprimarily process consumer bill paymentsand payroll,pension, and djyidenddeposits. TheZengin market share of thesepayments is 56%with 44%for the postal service.

Electronicpayments were expanded further in 1988with BOJ-NET,a nationwide large value wire transfernetwork for 386 depositoryinstitutions and securityfirms owned and operatedby the Bankof Japan.More recently, the useof checksand bills of exchange-traditionallyused for business payments-hasdecreased due to a stamptax on bills of exchangeand the availabilityof a new bank electronicpayment arrangement-a package transfer-where an enterpriseautomatically pays all of its supplierson a fixed day eachmonth (Federation of BankersAssociations of Japan,1994b, p.3).

Of the 78 banksin 1944,5 werelarge 'city m bankswith a nationwidebranch network while 73 werelocal bankswith a limitedbranch network.

37 Around 70%of all recurringelectricity, gas, water, telephone, and a TVbills are madeby direct debit through the postalgiro or the Zenginsystem. Electronic payments of credit cardbills, taxes,and rent are alsocommon.

38 The Zengin systemis backed by pledgedcollateral determined as a fixed percent of a participant'sself- selectednet debit cap. A loss-sharingrule has beenadopted in the event this collateralis insufficientto cover a participant'sfailure (BIS, 1993, p. 277). 29

Finally,19 electronicdelivery-against-payment system for governmentbond trading startedoperation in 1994.

The evolutionof paymentsin the U.S.and the reasonfor checkuse. The U.S. is a check- intensivecountry. The evolutionfrom cash to checksis largely due to two historicalfacts, both of which are uniqueto the U.S. First,from its inception,the U.S. bankingsystem provided savings and safekeepingservices to the general public. Thus, unlike Europe where banks served only the merchantsand the wealthy,there was no savingsor transactionservice vacuum for the postalservice to fill. As a result,a postalgiro wasnever established.

Second,the U.S. is a geographicallylarge country and the bankingsystem, although starting out as a national monopoly (e.g.,the Firstand SecondNational Banksof the U.S.), has been very unconcentratedfor the last 150 years. Until quite recently,banks were only permittedto operate within a single stateand, in manystates, to only havea singleoffice regardlessof whetherthey were stateor federally chartered. Only after the Great Depressionwere statelaws slowly liberalizedto allow a bankto branchwithin a state. Consequently,both consumersand businesseswere restricted (for legaland conveniencereasons) to hold their fundsat differentlocal or regionalbanks and account information was never centralized. When payors and payees have accountsat different and geographicallydispersed banks, the only cost-effectiveway to makea non-cashpayment-prior to the developmentof low-costelectronic communication-was to use a check or some other paper draft. Thus intensivecheck use was the logical and, at that time, cost-effectiveoutcome of limiting the concentrationof the bankingsystem in a geographicallylarge country.

At somepoint after 1997,when U.S.banks are allowed to branchnationwide and when POS terminalsbecome more availablethan they are at present,the U.S. will start to become more like Europe and Japan and increasinglyuse a final payment method to replace provisional check payments. However,for reasonsnoted below, the replacementof checksby electronicpayments will be slow.

Stagesin the evolutionof U.S.payments. With the expansionof domestictrade and the rise of populationdensity in trading centersduring the early 1800s,paper checks increasingly substituted for currencyin businesstransactions. This representedthe first stagein the evolutionfrom currencyto the useof non-cashpayment instruments. By 1865,bank deposits had exceeded the value of currency in circulation and checksplayed an increasinglyimportant role in the U.S. paymentsystem. The passageof the NationalBank Act expandedthe numberof nationalbanks and imposeda 10%tax on bank notes issuedby any institutionother than a nationalbank. The tax sooneliminated the useof bank notesissued by statechartered banks and reducedthe supply of currencyfor transactions.This vacuumwas filled by depositmoney in the form of increaseduse of checksfor tradeand exchange,in turn made more availableby the expansionof depositsin nationalbanks (Spahr, 1926, pp. 84 and 161).

Governmentsecurities are traded in thismanner, using a book-entrysystem, over Fedwire in the U.S. A debitcard which activates a real-timepayor account balance inquiry is, in effect,equivalent to a giropoint- of-saletransaction, since both are electronic credit transfers. 30

The rise in the use of deposit moneywas also accompaniedby a seriesof bankingpanics, leadingto major financialcrises. Theserecurring panics typically startedwith the insolvencyof a few banksand quickly spreadto solventbut illiquid institutionsas well. The depositruns and bankfailures that followed led to contractionsof the money supply, reductionsin bank credit, and economic depressions.These panics also reducedthe acceptabilityof checksin trade due to the uncertaintyof being able to collect funds when a check was presentedfor payment. The occurrencesof bank illiquidity were, in the breech,met by the issuanceof clearinghouse loan certificateswhich were used to satisfyinterbank net debit positions,and technicallyrepresented the untaxed(and therefore illegal) issuanceof private,non-bank currency (Roberds, 1995). Theserecurring bank liquidity problemswere finally addressedwith the establishmentin 1913 of a centralbank that had the sole power to issue currency,expand the moneysupply, and thus createneeded liquidity throughthe discount window. While this mitigatedthe illiquidity problemsfrom bankdeposit runs, the establishmentof a systemof deposit insurancein 1934 finally providedthe necessaryconfidence to preventdeposit runsto begin with, ratherthan provide liquidity througha centralbank loanafter a run would occur.

The settlementof checks among banksboth locally and interregionallyoccurred with the exchangeof interbankcorrespondent balances among institutions. Correspondentbalances were establishedand maintainedthrough the shipment of gold, and later currency, between banks. Shippingcosts were basically proportional to the valueof the currencyor gold being shipped. Out of this processevolved a systemof check exchangecharges that were proportionalto the par or face value of the checksbeing exchanged.Unfortunately, the systemof checkexchange charges was not uniform and this led to circuitousrouting of checksfor collectionwhich delayedtheir collectionand hinderedcommerce (an issuediscussed in moredetail below).

The establishmentof the FederalReserve represented a secondstage in the evolution of the U.S. paymentsystem and alteredit in two importantrespects. First, FederalReserve member banks (typically,the largestbanks) could transferlocal and interregionalsettlement funds by wire using a singlereserve-account balance held with an office of the centralbank. This reducedthe need to hold and maintaina complex network of correspondentbalances-and the physicalshipment of gold or currency-tosettle check payments.The secondchange to the paymentsystem was the eliminationof non-par banking(discussed below). This greatlyimproved the efficiencyof checksin the payment systembut requiredthe direct interventionof the centralbank in the processingof theseinstruments in competitionwith correspondentbanks and clearinghouses, a competitionwhich still existstoday.

A final stagein the evolutionof the U.S.payment system, similar to that in Europeand Japan, is the movementfrom paper-basedto electronicpayments. This hasoccurred on two frontsduring the 1970s:credit cardpayments processed by the privatesector and a FederalReserve effort to developan electronic substitutefor the check. The switch from processingpaper credit card draftsto making thesepayments electronic has been markedlymore successful than the effortto switchfrom checksto automatedclearing house (ACH)electronic payments. Credit card paymentscurrently accountfor over 16%of total U.S.non-cash payment volume while the ACH (which handlessome giro-type credit transferpayments) accounts for lessthan 2%. Credit card transactionshave been more successful becausethey meetbetter consumer convenience needs.

Seriousbarriers exist for the rapid expansionof electronicpayments in the U.S. Theseconcern the existenceof substantialcheck float, the way paymentservices are priced,bank sunk costs in check 31 processingoperations, and the unconcentratednature of the bankingsystem along with strongantitrust laws. Unlike Canada,'%hich also is a heavyuser of checks,check usersin the U.S. have a float incentiveto use checks. Businessesenjoy about 90% of the float benefitsassociated with checks despitewriting only about40% of the total volume (Humphreyand Berger,1990). Overall, from the check user'spoint of view, a check is cheaperto usethan is an electronicpayment, even though in termsof socialor realresource cost a checkcan be moreexpensive than an electronicpayment.

A secondbarrier to the useof electronicpayments is the lack of cost-basedpricing incentives to usersof paymentinstruments. This stemsfrom a Depressionera law forbidding paymentof interest on demanddeposits. Prior to the 1980s, banks generally provided check services to depositorsat no charge in order to compensatedepositors for their inability to pay intereston transactionaccounts. This "barter" arrangementwas changed in the early 1980swith legislationpermitting the paymentof intereston new consumertransaction accounts, but not on demanddeposits. As consumersshifted to the new interest-earningchecking accounts, banks attempted to recoup the interestpaid on these depositsthrough minimum balancerequirements and monthly fees. As these user costswere not relatedto the per transactionuse of checks,consumers "see" a zero price for eachcheck they useand view checkfloat asa privatebenefit.

Another barrier to electronic paymentsis banks' sunk cost in existing check operations. Although new electronicpayment technology has a high proportionof fixed to variablecost, check processingand delivery also hasa fixed cost. Importantly,the fixed costassociated with checkshas largelybeen paid for. Thus,from the bank'spoint of view, it is not profitableto adopta new payment technolog2unless its expectedaverage total costper transactionis lessthan the averagevariable cost of checks. Sunkcosts make it hardto changepayment arrangements at the supplierlevel because any new technology not only must have a lower averagecost per transaction,but also a cost sufficientlylower to offsetthe sunkcost in the currentsystem. Even so, if banksdid try to switch users from checksto electronicpayments, they would merelycannibalize their check serviceand not realize any importantgain in paymentmarket share. As depositoryinstitutions have monopoly accessto centralbank settlementservices, the only real competitionthat existsis within the banking industry and the existenceof sunkcosts inhibits change.

Lastly,the bankingsystem is very unconcentratedand the U.S.has strong antitrust laws. With 10,800banks, 2,400 savings institutions, and over 12,000credit unions,it would be difficult to obtain agreementto providea nationwidepaper or electronicpayment service similar to the giro in Europeor the Zenginsystem in Japan.The problemwith establishingsuch a bank-ownedfacility is that it would require the participationand agreementof hundredsof large banks in order to operateeffectively nationwide. And, if such a paymentfacility proved successful,non-participating institutions could argue-underexisting antitrust law-that they too shouldbe allowed accessas the facility is "essential"

41 In Canada,business payors are chargedfor the checkfloat they create. Becausefloat is costly to payors,there is a stronginducement to disbursechecks close to wherepayees are locatedto minimizecheck collection times. By agreementamong banks, payees who depositchecks are given immediateavailability to their funds (rather than havingto wait until a checkis collected).

42 Averagevariable cost, rather than average total cost,applies for checksbecause check fixedcosts have already beenlargely recovered and the existinginvestment still hasconsiderable economic value. 32 for their business(McAndrews, 1995). Thus,the institutionsthat developedthe facility could not realizeall of the potentialexternalities of their investmentand marketshare improvement. For these reasons,U.S. banksare unlikely to jointly developa new nationwideelectronic payment network for consumerand businesspayments. A more likely outcomewould be that banksadapt the existing credit cardand debit cardnetworks for this purposeover time.

The historical role of the central bank in the payment system. The principal roles of developedcountry central banks in the paymentsystem today are to (1) providesettlement for the full range of paymenttransactions and (2) superviseand, in many cases,also operate a large value domesticwire transfernetwork. While therehave been instancesin the pastwhen centralbanks have, for short periods of time, provided processingservices for other types of payments,only the U.S. centralbank continuesto do so. Althoughaccurate in broad terms,this generalizationglosses over importantdifferences within Europeregarding central bank involvementin clearingand settlement arrangements.In countrieswhere the centralbank has full responsibilityfor monetarypolicy, there hasbeen a tendencyfor the centralbank to exercisesignificant control over the operationand design of domesticpayment systems. At one end of the spectrum,the centralbanks of Germanyand France facilitatethe transmissionof retail paymentsand the associatedpayment information between banks in additionto largevalue paymentswhich makestheir role in the paymentsystem quite central. Central bank involvement in payment systemshas also included, in many cases, the promotion of standardizationand cooperationamong payment participants in order to minimize the social cost of supplyingpayment services. At the other endof the spectrum,the Bankof Englanddoes not haveany statutorypowers over the U.K. paymentsystem and the systemis almost solely dependenton the privatesector; CHAPS and Town Clearing,for example,are run by a privateorganization (APACS, the Associationfor PaymentClearing Services) of which the Bankof Englandis a member.

Typically, the provisionof central bank paymentprocessing services has been related to a breakdownin the capacityof the privatesector to safelyor efficientlyprovide these processing services for itself. A casein point is centralbank provisionof a largevalue wire transferservice. In this case, the central bank is a neutral party to transactionsamong banks, has the ability to createmoney and grantcredit, and thus is the desiredentity to provide for the transferand simultaneoussettlement of largevalue fundsand securitytransactions. Often this is facilitatedby the ability to use idle balances held for purposesof monetarycontrol for the secondarytask of clearingpayments. This reducesthe need to hold paymentclearing balances elsewhere, such as with correspondentbanks. In developed economies,the huge numberof financialtransactions and largeinterbank obligations demand timely same-dayfinal settlementof paymentsto reduce intertemporalrisk and facilitatetrade. In such an environment,it is much more efficient to have a centralizedsettlement facility rather than settle through numerous interbank correspondent(or nostro) accounts,each of which would require sufficientbalances to meet settlementobligations (Blommenstein and Summers,1994). Thus a two- tiered interbankpayment system is, at a minimum, desired:tier 1 would consistof the centralbank's settlementfacility while tier 2 would be composedof commercialbanks. In countrieswhere there happento be largenumbers of smallfinancial institutions which alsoprovide payment services, a third tier often ariseswhere someof the largerbanks act as clearingagents for this setof institutions. Such arrangementsare in place in the U.K. and Germany(as well as the U.S.). While there are strong reasonsfor centralbank involvementin largevalue payments,the justificationfor also being involved in the processingof other typesof paymentsis lesscompelling. The U.S.experience, outlined below, illustratesthe bestknown casefor centralbank intervention in this regard. 33

The U.S. centralbank startedout providing free check processingservices to banksthat were membersof the FederalReserve when it was establishedin 1913. Memberbanks held idle reserve balanceswith the centralbank for the dual purposeof settlingpayment transactions and for monetary control (throughthe depositmultiplier). The main purposefor havingthe centralbank alsoprovide a check processingservice was to eliminatethe private bank and clearinghouse practiceof non-par checking. Non-parchecking was where checksdeposited by customerswere paid, not at face value or at par, but at a discountfrom par. This discountfrom par was how collectingbanks covered their costsof collectinga check. However,the discountvaried acrossbanks according to their favorable geographicalplacement in the collection chain (higherdiscount) as well as their desire to attract correspondentdeposits associated with clearingchecks (low or zero discount). In order to minimize collectioncharges, collecting banks would attemptto routechecks through those correspondent banks with the lowestcollection fees, even though this often wasnot the shortestdistance to the payingbank on which the checkwas drawn. This led to circuitousrouting of checks,increased payment float, and therebyinhibited commerce(Spahr, 1926).

The centralbank was given the legal authorityto collect checksfor its memberbanks at par and, sincethis minimized collectionand float costs,more and morechecks came to be processedand collected through the FederalReserve. To remain competitive in the check collection business, correspondentbanks also insisted on being paid at par, threateningto otherwisecollect itemsthrough the central bank. Finally, non-parbanking was eliminatedregardless of who-the central bank or a privatebank-was the collectionagent.

Prior to the pricing of FederalReserve payment services in the early 1980s,the centralbank's shareof the check processingmarket was around 33%. After pricing, the4Federal Reserve'sshare fell to around 25%. The remaining sharesalf comprised of on-us items (3 0 %) plus correspondent banks and local check clearing houses(45%). With the advent of interstatebank branching in 1997, U.S. bankswill become larger and their offices more geographicallydispersed. This will increasethe proportiop of on-us checks and further reduce the check processingmarket share of the Federal Reserve.

In the long-run,the FederalReserve will increasinglybecome the check processorof "last resort"; it will be usedby only small banksfor the purposeof offsettingthe market power of large correspondentbanks. As its marketshare dwindles, due to the growth of on-us checks,reciprocal

One widely quoted exampleof circuitousrouting at the time concerneda check depositedat a bank in Hoboken,New Jersey,and drawn on a bank only 100 miles away in Sag Harbor, New York. This check '...passedthrough elevenbanks, traveled about 1,500miles, and was in transit about elevendays...' (Spahr, 1926,p. 105). Almost every businessman,the primary userof checksat the time, had a similar horror story to tell about their ability to collectchecks in a timely fashion.

44 On-uschecks are thosethat are depositedat thesame bank theyare drawn on andso areprocessed internally.

The lastsurvey of U.S.checks was for 1979,before Federal Reserve payment services were priced. Thusthese marketshare figures are estimates.

46 With the full implementationof nationwide bankingand branching,the central bank's check processing marketshare may fall to 13% to 18% (Bergerand Humphrey,1988). 34 electronic presentmentamong large banks and check clearing house associations,plus the replacementof checksby credit card, debit card, and ACH payments,the centralbank will likely at somepoint exit the checkprocessing business. The issuethen will be whetherthe centralbank should alsoturn over its ACH processingbusiness to the bankingassociations which use it. Currently,the FederalReserve's market shareof ACH processingis around 90%. However, sincethe volume of ACH transactionsis only 2.5% of the over 60 billion checksprocessed each year, the effectof sucha transferwould be lessthan that of exitingthe check processingmarket.

As outlined, the role of the U.S. centralbank in check processingis seen to result from a historicalfailure of the privatesector to providean efficient, low float, processingenvironment for the smoothflow of commerce. This failure was largelydue to the highly unconcentratednature of the bankingsystem brought about, in turn, by restrictivelaws which limited branching-andthus bank size-in a countrythat wasgeographically quite large.

Had the U.S. not restrictedbank size and their geographicaldistribution, it is then likely that non-par checking would not have had the disruptiveeffects on commercethat it did. Insteadof thousandsof tiny, one-officebanks, there would havebeen only hundredsof much largersized banks with geographicallydispersed offices-similar to what occurredin Canada. As a result, many more checkswould have been clearedas on-us items (so only one remittancefee would be chargedto collect a check)and non-on-usor transit itemswould havehad to go through far fewer correspondent banksto be finally paid. Had theseconditions existed, then the discountfrom par to collect a check would havebeen lessand the incidenceof circuitousrouting-with its accompanyingfloat cost-would havebeen considerably reduced. If the disruptiveeffects of non-parchecking on commercehad been mitigatedin this manner,it is certainly lesslikely that therewould havebeen a need for the Federal Reserveto provide a check %ocessing service,and the political justificationfor such action would havebeen markedlyreduced.

Overall, if the private sector is unable or unwilling to provide an environment in which paymentscan be made in a timely fashion,then it may be necessaryfor the central bank to (a) interveneindirectly by supportingchanges in paymentslaws or other incentivesor (b) intervene directly by taking an active role in paymentprocessing activities themselves to alter the situation(as occurredin the U.S.).

VIl. Lessonsfor emergingmarket economies

Paymentsystems involve the point-to-pointtransfer, processing, and settlementof payment informationbetween payors and payees(and/or their agents).Our economicand historicalanalysis of paymentsystem evolution in developedcountries has focused on what is effectivelya constrainedcost

47 Likethe U.S.,Canada had a systemof checkexchange charges-leading to non-par checking-but these charges wereuniform (Spahr, 1926, p. 32). Whileexchange charges would tend to limit the growthof checksas a replacementfor cash,particularly for retail and some business transactions, their uniformity would not haveled to the circuitousrouting and float costexperienced in the U.S. In addition,Canadian banks were relatively larger(so more checks were on-us) and nationwidebranching was permitted (reducing the needto useother banksto collectchecks outside of the localarea). 35 minimizationproblem for transferringpayment information and we havenoted how this processhas been molded by country-specificgeographical, institutional, legal, communication,cultural, and economicconstraints.

The evolutionof the paymentsystems of Europe,Japan, and the U.S. provide somelessons for how emergingmarket economies may cost-effectively direct andfacilitate the developmentof their own paymentsystems. Similarly, the role playedby developedcountry central banks in this process suggestsa guidelinefor the properrole of othercentral banks. As illustratedabove, the evolutionof a country'spayment system is largelydetermined by the following countryattributes and conditions:

(i) Countrysize and populationdensity; (ii) Bankingsystem concentration and interconnectedness; (iii) Paymentinstrument legal structure and antitrustlaws; (iv) Crime ratesand othercultural factors; and (v) Risk and efficiency tradeoffsfor different transactionsand payment instruments(relative paymentcosts versus user requirements and needs). By assessingthe extentto which eachof thesefive conditionsmay apply to a particularcountry, it is possibleto obtain a good ideaof the likely cost-effectiveevolutionary path of a paymentsystem.

Countrycharacteristics and the adoptionof specificpayment arrangements. As suggestedin the beginning,a usefulway to view a country'spayment system is in terms of separateyet broadly similar types of transactionsassociated with paymentsat the point of sale, the paymentof bills and other recurringpayments away from the point of sale,disbursements for wageand payroll payments, and all types of financial markettransactions. In each of thesecategories, payors and payeesface different cost and convenienceincentives that are associatedwith the size of the country, the concentrationand centralizationof the bankingsystem, the underlyinglegal structure, crime rates,and variousrisk, transactioncost, and userconvenience considerations. From our analysis,the following conclusionsare drawn regardingcost-effective payment instrumentuse for an emerging market economy.

Point-of-salepayments. POSpayments are best handledby cash if a country's crime rate is low, as evidencedby the experienceof Japan. Cashpayments are final, are immediate,and can be reusedfor other transactionsof a different value. Cash also createsseigniorage revenues for the government(as its productioncost is lessthan its facevalue). However,if the crime rate is high, the banking system is unconcentrated,and electronic communicationwithin and among banks is expensive(due to largegeographical size of the country),then a checkwill probablybe the preferred alternative,as is the casein the U.S. If the crime rate is high but the bankingsystem is concentrated and an efficient electroniccommunication infrastructure exists (because the country is small and/or denselypopulated), then a credit card or debit card systemcan be a lower cost alternativethan cash and/orchecks would be over time. Thesegeneralizations, of course,will be affectedby the existence (or the lack of) a properand enforceablelegal foundation dealing with the rightsand liabilitiesof users of non-cashpayment instruments (whether paper-based or electronic). 36

Billpayments. Bill paymentsin cashare inconvenientbecause cash usually involves a face-to- face transaction.Even when the crime rateis low, non-cashalternatives are preferredto cash(as seen in Japan). If the crime rate is high and the postal serviceis reasonablyefficient, mailing a check becomesthe preferredalternative to cashif the bankingsystem is unconcentratedand no centralized paymentfacility exists(as in the U.S.). If the bankingsystem is concentratedand/or a centralized paymentfacility exists, paper-basedgiro paymentsare likely a lower cost and more convenient alternativethan cash or checksfor bill payments. And if the internal accountingsystem of the centralizedfacility can be madeelectronic, then electronicgiro paymentswill be cheaperthan paper- basedpayments when the countryis not large(which is the experiencein Europe).

Disbursements.When the crimerate is low, thendisbursements in cash(payroll payments, person-to-personpayments) have been commonly used. When the crime rate is highand the banking systemis unconcentrated,then disbursement by checkis saferand cost effective as well, althou'ghit may also meanthat workerswill take time off from work to deposit/cashtheir checkwhen paid. Whena centralpayment facility exists, disbursements can be madedirectly to the employee'saccount usingeither paper-based or electronic payments (depending on cost).

Financialmarket transactions.Financial market transactions are typically of very largevalue and requiresecure and timely transfers. The best alternative, due to its safety,speed, and connection to the bankingsystem, is a wire transfernetwork within and betweenmajor moneycenters in a country.All developedcountries have such networks and/or have access to them. In thisinstance, the realchoice is betweena low volumetelegraphic communication system, a high volumededicated wire network,or a seriesof specializedand centralized paper or electronicbook-entry systems for differentfinancial instruments (or some combination of thesedifferent approaches).

It is seenthat thereis not just one pathfor a paymentsystem to follow but many,and the choicedepends on the specificsituation in a countryregarding the five constrainingfactors listed above.Overall, the historicalevolution of paymentsin Europeand the U.S.is arguablymore relevant for emergingmarket economies that do notyet have a relativelydeveloped electronic communication infrastructureso paper-basedpayments will haveto be reliedupon for manybusiness and financial transactions.Economies with a developedcommunication infrastructure will likely find it cost- effectiveto movemore rapidly to electronicpayments and, while passingthrough a periodof paper instrumeni8 use, will not concentrateas heavilyon these instrumentsas have the developed countries. In termsof developmentpriority, problems with businessand financial market payments should be addressedbefore importantimprovements are consideredfor point-of-saleand bill payments.However, reforms should not stophere. Improvementsin the efficiencyand convenience of retailpoint-of-sale and bill paymentsin a mannerthat meets users' preferences can greatlyfacilitate tradeand exchange for the majorityof the population,evidence of whichexists from the evolutionof paymentsin developedeconomies. Toward this end, a rangeof differentpayment instruments should be madeavailable since no one instrumentis bestsuited for all kindsof transactionsnor meetsall consumerneeds. Such a multiplicityof paymentinstruments, of course,needs to be balancedwith reformsto restructurea payment system in a costefficient manner.

48 Speedyprocessing of paper-basedpayments requires the standardizationof paymentinformation and its placementon paymentinstruments so it can be readelectronically. Similarly, common formats are a preconditionfor efficient electronic payments. 37

Evolving payment arrangements in emerging market economies. At the moment, many developingcountries use cashfor most POSand bill paymenttransactions. Enterprises, in contrast, rely on check or other paper-basedpayments as their paymentsare typically for largervalues and are relativelyfew in number. Electronicpayments, when they are being developed,are initially (and properly)focused on providingfor moretimely largevalue inter-enterpriseand governmentpayments. This generaldescription fits China, India,Thailand, Indonesia, the Philippines,Columbia, Mexico, as well as other developingcountries. This descriptionalso fits the manycountries in transitionfrom a commandeconomy with a monobankfinancial structure to a more market-basedstructure. Formerly centrally plannedeconomies, however, do not havethe necessarylegal, accounting,and regulatory frameworkneeded to implementsuccessfully payment system reforms, and so face many morebarriers in their attemptto improvetheir paymentsystem (Sato and Humphrey,1995).

The countriesin transition,for exampleRussia and other East-Europeanex-socialist countries, have relatively developeddeposit money basedsystems but the bulk of consumerand even many businesspayments are still madewith cash(Commission of the EC,1995). The common problem is that the paymentsystem is unreliableand slow; banktransfers can take weeksin somecountries and thereforeit is often bypassedwith manytransactions carried out usingcash. The needfor intervention in paymentsystems through well-defined reform programs has thus been clear in mostcases.

In an attempt to shift inter-enterprisepayments away from cash, and to enforce monetary control as well as to supportthe ruble (insteadof westerncurrencies) in domestictransactions, the Russiangovernment has stipulated that certain important Russianenterprises denominate their paymentsin rubles and transferfunds using bank transfers. No cash paymentsare allowed and deviationsrequire the permissionof the centralbank (Salonen, 1995). What effectthis ruling will have in practiceremains to be seenbut it illustratesthe strongintention of the stateto influencepayment practices.

In the early 1990s,the Russianpayment system was in dire condition and it took weeksfor fundsto be finally transferredto payees.In the hyperinflationaryconditions that existed,this wasquite detrimentaland encouraged the outflow of capitalfrom Russia.At thattime, non-cashpayments could only be clearedthrough existing central bank offices which were not set up to processpayment orders in a timely manner. After direct account relationsin rubles between a few Russianbanks were permitted in 1992, a number of banks have set up a competitive correspondent-basedpayment clearing network (Salonen,1995). While this resultedin a substantialimprovement in the payment situation,it also demonstrateshow fundamentalchanges in legislationare often neededbefore any progresscan be achieved. However,no centralizedor multilateralclearing operations are yet in place and the currentsystem is basedentirely on bilateralrelations. A weekcan still be easily lost in Russia when transferringfunds. The great size of the country, the differing stagesof developmentin the various regions,and the useof heterogeneouspayment technologies make it difficult to implementa comprehensivepayment development plan in Russia. For thesereasons it is difficult to forecastthe shapeof the "final" paymentsystem in Russia.The bankingsystem has also developed very swiftly but in an unpredictablemanner. The numberof commercialbanks increased enormously during the 3 to 4 yearsafter Perestroikalargely because the thresholdfor establishinga bankwas deliberatelyset very low in orderto rapidlydevelop a privatebanking sector (Laurila, 1995). 38

In many respectsthe developmentof the financial sectorhas been even swifter in Eastern Europe,where fewerobstacles were facedin shiftingthe earliercommand economy toward a market- basedsystem (Commission of the EC, 1995). Market-basedreforms seem to be working reasonably well in thesecountries and are facilitatingbanking and financialsector development. Two different approachesto reform seemto have generallybeen pursued:either a gradualconversion of the old statebanking system (as in Hungary,Poland, and the CzechRepublic) or a more radicaltransformation which has createda new bankingand paymentsystem (as in Estoniaand Russia).The future will determinewhich of thesetwo approacheshas proved to be the moresuccessful. Still, a lot remainsto be accomplishedbefore payment system efficiency and reliability reachesinternational standards.

An issuewhich hasplagued all EasternEuropean transition economies, and hashad significant economic consequences,is the growth of paymentarrears by enterprises,banks, and even the government. Paymentarrears have, for example,increased credit risks in the economy and made emergingbanks even more fragile. Paymentarrears have not stabilizedin Russiaeven though they havestabilized elsewhere in EasternEurope (Hirvensalo, 1995; Bankof Finland,1995). At the outset, the arrearsproblem in Russia,Romania, the Czech Republic,and Polandwas associatedwith the malfunctioningof the paymentsystem. However,in Russia,the problemhas persistedas enterprises continueto expectfinancial aid from the governmentand government initiatives to imposebudgetary discipline havenot been credible. As a result,many unprofitable enterprises have kept on operating as effectivebankruptcy procedures are not in place and there is a lack of financial discipline in the system. Improvingthe paymentsystem could alleviatethe arrearsproblem to someextent, although the primary solutionlies reducingthe numberof unprofitableenterprises.

The problemsfacing other developingcountries in reformingtheir paymentsystems are less dauntingand primarily revolvearound (a) obtaining the funds necessaryto up-gradethe operationof their current paper-basedsystem by improvingintra-bank payments between branches of the same bankand (b) establishingan electroniclarge value paymentnetwork between major bankswithin and betweenthe primary trading centersof the country. This is one areawhere geographicsize playsa very importantrole: Thailandand Viet Nam,for example,are finding it much easierto implementan electronicreal time grosssettlement (RTGS) large value paymentnetwork than is the casefor China, the Philippines,and Indonesiawhich areconsiderably larger countries with moretrading centers and a moredispersed system of bankbranches (World Bank,1995).

The role of the central bank in the payment system. The overall relationshipbetween a county's central bank and economicand financialdevelopment is quite broad in practice. Central banksmust consider the linkagesbetween the efficientoperation of a country'spayment system, the level of reserveor clearingbalances needed to contain paymentrisks, and the liquidity needsof the systemunder differentsettlement arrangements. These payment system issues are, in turn, linked with bank soundnessregulations and safetynet arrangements-alender of last resortbank liquidity facility and possibledeposit insurance or guarantees.

More narrowly,the principal rolesof centralbanks in the paymentsystem focused on in this paper have been: (1) providing settlementfor the full range of consumer,business, and financial payment transactions;and (2) supervisingand/or operating a large value domestic wire transfer network. Centralbanks are explicitly in chargeof the systemicstability of the paymentsystem in many counries. In addition, central bankshave servedto promotepayment system reform and facilitate 39

49 cooperation among banks to improve paymejatarrangements, fostered policies where banking servicesare priced close to their actual costs, and worked to red4ce interestand exchangerate controlswhich tend to distortresource allocation and tradingstructure. Of equal importance,due to the fragility of a highly leveragedbanking system, is the need for centralbank oversight and/or direct responsibilityfor maintainingbank soundness. This is achievedthrough regulatorychannels and safetynet arrangementswhich protect both the bankingsystem and the operationof the payment system.Banking system stability enables payment system users to developand maintainconfidence in lower cost,bank-related payment instruments and so improvespayment and economicefficiency by loweringtransactions costs.

While therehave been instances in the pastwhen centralbanks have, for shortperiods of time, provided processingservices for other types of payments(e.g., Japanduring the 1940s),the U.S. central bankcontinues to play a major role here. However,there are instancesin variousemerging market economies(e.g., the Philippines)where the central bank providesfor the processingand clearingof inter-regionalpaper-based payments, while the privatesector handles local payments.This should be viewed as a forerunnerof a central bank operatedor supervisedlarge value payment networkrather than the centralbank replacing the privatesector in processinglow value payments(as occursin the U.S.).

Settlementthrough a centralbank is a lower costand saferprocedure than havingthe private sectorperform this task. Similarly,having the centralbank also providepayment processing services hastypically been relatedto a breakdownin the capacityof the privatesector to safelyor efficiently providethese processing services for itself. A casein point is centralbank provisionof a largevalue wire transferservice, which is directly relatedto the safetyof havinga neutralparty with the ability to createmoney (liquidity) alsoprovide for the transferand simultaneoussettlement of fundsand security 52 transfersof exceptionallyhigh value. Often this is facilitatedby the ability to useidle balances,held for purposesof monetarycontrol, for the secondarytask of clearingpayments. This reducesthe need

49 Cooperationamong banks enabled Europe and Japanto establisha cost-effectivenationwide electronic paymentnetwork for relativelylow valuebill paymentsand POStransactions. Cooperation to shiftto lower social cost electronicpayments is markedlymore difficult once checksbecome an establishedpayment instrument.The float benefitsaccruing to checkusers (as in the U.S.)or the banks(as in emergingmarket economies)provide a strongincentive working against change. so Cost-basedpricing of bankand central bank services allocates resources more efficiently than whenthese servicesare heavily subsidized or taxed.

51 Forexample, deposit interest rate ceilings provide banks with an incentiveto overexpandtheir branching network.Deposits will begathered until their extra (interest plus branching) cost equals the extrarevenue from makingadditional loans. As a depositceiling results in below-marketinterest costs, deposits are accumulated until branchingcosts rise to offset this advantage, leading to morebranches than would exist without the deposit rateceiling. A similarmisallocation of resourcesresults from excessive exchange rate controls.

52 Almostall largevalue payment networks that exist today effectively represent extensions of eithercheck-based or giro-typepayments. CHIPS, CHAPS, and BOJ-NET are net settlement networks and represent extensions of an initiallycheck-based (or non-finalpayment) system of inter-banktransfers. Systemic risk arisesbecause check- basednetworks implicitly involve the extensionof creditby payeesto payors,especially when funds received areused intra-day prior to finalsettlement. 40 to hold paymentclearing balances elsewhere, such as with correspondentbanks and thus is more efficient.

Althoughthe historicalevolution of largevalue paymentnetworks has been briefly discussed, this experienceis not of great consequenceto emergingmarket economiesbecause the payment technology,the risk perceptions,and the timeframefor largevalue payments,are so differenttoday comparedto the past. The choices presentlyfacing emergingmarket economiesare essentially betweennet settlementand RTGSsystems as they are currentlyconfigured, rather than how they may have worked in the past where settlementswere performedover a period of days and account balanceswere monitoredonly on an end-of-daybasis.

Within this context, an important issue concerns the costs and benefits of adequately controlling systemicrisk. In a net settlementframework, participants can post collateralcovering the single largestnet debit (as on CHIPS),post collateral covering all net debits (CHAPS),or have participantsin a net credit positionloan fundsto thosewith a net debt for a fee (BOJ-NET).In a real time grosssettlement framework, participants may hold adequateidle balances(as on SIC) or pay a price for any daylight overdraft incurred (as on Fedwire). The choice among thesearrangements dependsupon the risk toleranceof a country'scentral bank and the degreeof control exertedby the centralbank over its (privatelyowned) commercial banks. It also dependson the extent of banking systemconcentration and the volume and value of largevalue payments,and their expectedfuture growth, since this determinesthe net debit to bank equity capital ratio, and hence the likely importanceof systemicrisk. A net settlementsystem would be acceptableif a country's centralbank risk toleranceis high,the centralbank haslittle controlover commercialbanks, the bankingsystem is highly concentrated(more "on us" paymentsso havesmaller net debits),and the volume and value of largevalue paymentsis relativelysmall. Otherwise,a RTGSsystem may be the preferredoption.

The mostsignificant drawback to a RTGSsystem is its expenseto banksand its absorptionof bank liquidity. Due to the opportunitycost of (i) holding idle balancesto coverpayment requests, or (ii) borrowingdaylight liquidity from the centralbank (or other sources),or (iii) postingcollateral to cover daylight overdrafts,there exist incentivesto delay makingpayments until sufficientfunds have been receivedfrom other participants.This reducesliquidity in the bankingsystem and may, in the extremecase, lead to a paymentsgridlock in the eventthat a greaterthan expectednumber of large value paymentsare requestedby enterprisesand financial marketson a given day. Although alternativesare available,such as splitting up especiallylarge payments into smallercomponents that are sent in a piecemealfashion at differenttimes during the day, the liquidity problem is potentially very importantin the often upstablemonetary conditions in emergingeconomies when inflation and marketinterest rates are high.

Anotherdrawback to an RTGSsystem, although a small one, is that it reducesthe incentivefor participantsto monitor eachother's risk positions,as counterpartycredit risk is alleviatedwith RTGS. However, this problem is less importantthan if an RTGSsystem did not exist; while payment

Russia,for example,has recentlyexperienced severe distortions in its interbankfunds marketdue to exceptionallyhigh interest rates. This adverse impact on liquiditycould be alleviatedif the centralbank paid a marketinterest rate on reservesheld for clearing and payment settlement purposes, effectively removing what is in essencea 'tax' on largevalue payments. 41 participantsmay be able to effectively monitor the credit risk of a reasonablesubset of other participants,the credit exposuresthey would face would be insteaddetermined by the patternof payment traffic over a network and the two would not necessarybe the same from day-to-day (Schoenmaker,1994).

To date, mostemerging market economies have implementedor are planningto implement RTGSon their largevalue paymentnetworks. This holdsfor China,Viet Nam, Hong Kong,Thailand, Korea,Indonesia, Australia, and othercountries of EasternEurope and the former SovietUnion. While mostcountries plan to reduce systemicrisk on thesenetworks by requiring that net debits be fully collateralized,the countriesof the former SovietUnion haverequired that adequateidle balancesbe availableto minimize the incidenceof uncoverednet debitsto begin with. This raisesan important point: in emergingeconomies the availabilityof liquidity in interbankmoney marketsis usuallyquite limited and the banking systemis still fragile. Consequently,the very high mandatoryreserve requirementsoften seenin thesecountries can serveseveral purposes, such as eliminating systemic risk, avoiding the possibilityof a paymentsgridlock, and reducingthe need for the central bank to provide intraday (and overnight) liquidity. However, high reserverequirements are costly to the banking system,although they may be necessaryuntil paymentoperations and financial markets developsufficiently to reducecosts by substitutinginterest earning collateral for idle reservebalances.

In summary,the overall considerationregarding central bank involvementin the payment systemconcerns the inabilityor resoluteunwillingness of the privatesector to provide an environment in which paymentscan be madein a safeand timely fashion. When suchconditions exist, it may be necessaryfor the centralbank to eitherintervene indirectly by supportingchanges in paymentslaws or other incentivesor to intervenedirectly by taking an active role in paymentprocessing activities to alter the situation. To date, the casefor direct interventionis strongestfor settlementof all types of paymentsand in processinglarge value payments. Additional intervention,by providing for the processingof other non-cashpayments, should be consideredonly on a specialcase basis and when possiblebe only temporary. 42

BIBLIOGRAPHY

Angelini, P., G. Maresca,and R. Russo,"Systemic Risk in the Netting System",lournal of Bankingand Finance, (une 1996): p.853-68. Bank for InternationalSettlements, Payment Systems in Eleven DevelopedCountries. Third edition, Bank AdministrationInstitute, Rolling Meadows, IL (May 1989). Bank for InternationalSettlements, Payment Systems in the Group of Ten Countries. Basle,Switzerland (December1993). Bankfor InternationalSettlements, Payment Systems in Finland. Basle,Switzerland (uly 1994). Bank for InternationalSettlements, Statistics on PaymentSystems in the Group of Ten Countries. Basle, Switzerland(December 1994). Bankfor InternationalSettlements, Payment Systems in Norway. Basle,Switzerland (May 1995). Bankof Finland, Paymentand SettlementSystems in Finland. Bank of Finlandpublications, A:88, Helsinki (1993). Bank of Finland, "Review of Economiesin Transition, Unit for EasternEuropean Economies", Working Document,Helsinki (uly 1995). Berger,A., and D. Humphrey,"Interstate Banking and the PaymentSystem", lournal of Financial Services Research,1 Uanuary1988): 131-45. Blommenstein,H., and B. Summers,"Banking and the PaymentSystem", in The PaymentSystem: Design. Management.and Supervision,B. Summers(Editor), International Monetary Fund,Washington, D.C. (1994). Boeschoten,W., "NationalTrends in PaymentSystems and the Demandfor Currencyand BankNotes", Research Memorandum9108, De NederlandscheBank (March 1991). Boeschoten,W., CurrencyUse and PaymentPatterns. Financial and MonetaryPolicy Studies,Vol. 23, Kluwer AcademicPublishers, Norwell, MA (1992). Boardof Governorsof the FederalReserve System, Functional Cost Analysis. Washington, D.C. (1991). CEPS,"European Payment Systems and EMU",CEPS Working Party, Report 11 (April 1994). Commissionof the EC, "Making Paymentsin the Internal Market", DiscussionPaper COM 90 447 Final (September1992). Commissionof the EC,"Progress and StrategyPaper: Banking", lead authorC. Clarke,Directorate General for ExternalEconomic Relations (1995). Federationof BankersAssociations of Japan,The BankingSystem in lapan. Tokyo,Japan (1994a). Federationof BankersAssociations of Japan,Payment Systems in lapan. Tokyo,Japan (1 994b). Gual, J.,and D. Neven, "Deregulationof the EuropeanBanking Industry 1980-1991", CEPR Discussion Paper 703 (1992). 43

Hancock,D., andJ. Wilcox, 'Intraday Managementof BankReserves: The Effectsof Capsand Feeson Daylight Overdrafts",lournal of Money,Credit, and Banking(November 1996) Hirvensalo,I., "PaymentArrears and RussianReforms" in Russia'sFinancial Markets and the BankingSector in transition,J. Rautava(Editor), Bank of FinlandStudies, Helsinki (1995). Humphrey, D., The U.S.Payments System: Costs, Pricing, Competition and Risk.Monograph Series in Finance andEconomics, Nos. 1 & 2, New York University,New York,NY (1984). Humphrey,D., "PaymentsFinality and Risk of SettlementFailure", in A. Saundersand L. White (editors), Technologyand the Regulationof FinancialMarkets, Lexington Books, Lexington, MA (1986). Humphrey,D., PaymentSystems: Principles, Practice, and Improvements.World Bank TechnicalPaper No. 260, Washington,D.C. (1995). Humphrey,D., and A. Berger,"Market Failure and ResourceUse: Economic Incentives to Use DifferentPayment Instruments",in The U.S. PaymentsSystem: Efficiency, Risk. and the Roleof the FederalReserve D. Humphrey(Editor), Kluwer Academic Publishing, Norwell, MA (1990):45-86. Humphrey, D., L. Pulley, and J. Vesala,"Cash, Paper, and ElectronicPayments: A Cross-CountryAnalysis", lournalof Money.Credit. and Banking,(November 1996). InternationalMonetary Fund, International Financial Statistics. Washington, D.C. (variousissues). Kearney,K., "An HistoricalPerspective on Researchand Studieson PaymentSystems", in M. Yamaguchi(1993), seebelow. Laurila,J., "RussianBanking Legislation and Supervision"in Russia'sFinancial Markets and the BankingSector in transition,J. Rautava(Editor), Bank of FinlandStudies, Helsinki (1995). McAndrews,J., "Antitrust Issuesin PaymentSystems: Bottlenecks, Access, and EssentialFacilities", Federal ReserveBank of PhiladelphiaBusiness Review, (September/October 1995): 3-12. Neven, D., "StructuralAdjustment in EuropeanRetail Banking:Some Views from IndustrialOrganization", in EuropeanBanking in the 1990s,J. Dermine(Editor), Basil Blackwell, London (1993). Padoa-Schioppa,T., "The BankingSystem and the PaymentsSystem", Economic Bulletin, Bancadltalia, Italy (March1986). Porter,R., and R. Judson,"The Locationof U.S. Currency:How Much is Abroad?",Working Paper,Board of Governorsof the FederalReserve System, Washington, D.C. (July1995). Rautava,J., "FinancialFactors and EconomicDevelopment: Russian Dilemmas", in Russia'sFinancial Markets and the BankingSector in transition,J. Rautava(Editor), Bank of FinlandStudies, Helsinki (1995). Richards,H., 'Daylight OverdraftFees and the FederalReserve's Payment System Risk Policy', FederalReserve Bulletin(December 1995): 1065-77. Roberds,W., "Financial Crisesand the PaymentsSystem: Lessons from the National BankingEra", Federal ReserveBank of AtlantaEconomic Review, (September/October 1995): 15-31. Salonen,I., "RussianCommercial Banks: A Banker'sView", in Russia'sFinancial Markets and the BankingSector in transition,J. Rautava(Editor), Bank of FinlandStudies, Helsinki (1995). Sato, S., and D. Humphrey, "TransformingPayment Systems: Meeting the Needs of EmergingMarket Economies",World BankDiscussion Paper No. 291,Washington, D.C. (uly 1995). 44

Schoenmaker,D., "Externalitiesin PaymentSystems: Issues for Europe",CEPS Research Report, London School of Economics,London (April 1994). Spahr,W., The Clearingand Collection of Checks.The BankersPublishing Co., New York, NY (1926). Stehm,Jeff, 'Clearance and SettlementSystems for Securities: Critical DesignChoices in EmergingMarket Economies,'World BankDiscussion Paper No. 321,Washington, D.C. (April 1996). Thomsom,F., GIROCredit Transfer Systems, Macmillan Company: New York, NY (1964). Tsurumi,M., The Establishmentof the DepositBanking System in lapan, the Bank of lapan. and the Money Market,(in Japanese)Yuhikaku Publishing Co., Tokyo, Japan (1991). Tsurumi,M., "The Establishmentand Reformof the NationalCentralized Domestic Exchange Settlement System in Japan",in The LocalBanks in Post-Warlapan, (in Japanese)The Associationof LocalBanking History in Japan,Toyokeizaishinpou Publishing Co., Tokyo, Japan (1994a). Tsurumi,M., "Developmentof Channelsof IndustrialFinance in Japan,1914", lournalof InternationalEconomic Stdie, No. 8, Tokyo,Japan (1994b). Vesala,J., RetailBanking in EuropeanFinancial Integration. Bank of Finland,Helsinki, Finland (1993). Vesala,J., Testingfor Competitionin Banking:Behavioral Evidence from Finland. Bankof Finland,Helsinki, Finland(1995). Viren, M., Use of PaymentMedia and Demandfor Currencyin Finland. Bankof Finland,Helsinki, Finland (1993). Vital, C., "Market Responsesto an ElectronicInterbank Payment System", Payment Systems Worldwide, (Spring 1990). WorkingGroup on ECPayment Systems, Minimum Common Features for DomesticPayment Systems. Report to the Committeeof Governorsof the CentralBanks of the Member Statesof the EuropeanEconomic Community(September 1993). World Bank, Viet Nam: PaymentSystem and Bank ModernizationProject. Country OperationsDivision, Washington,D.C. (October1995). Yamaguchi,M., PaymentSystems: Strategic Choices for the Future. Hitachi ResearchInstitute, Tokyo, Japan (1993). Policy Research Working Paper Series

Contact Title Author Date for paper

WPS1651 Bank-Led Restructuring in Poland: Cheryl W. Gray September 1996 B. Moore Bankruptcy and Its Alternatives Arnold Holle 38526

WPS1652 Intra-Industry Trade, Foreign Direct Bernard Hoekman September 1996 F. Hatab Investment, and the Reorientation Simeon Djankov 35853 of Eastern European Exports

WPS1653 Grants and Debt Forgiveness in Leonardo Hernandez September 1996 R. Vo Africa: A Descriptive Analysis 31047

WPS1654 Indonesia's Palm Oil Subsector Donald F. Larson September 1996 P. Kokila 33716

WPS1655 Uncertainty and the Price for Crude Timothy J. Considine September 1996 P. Kokila Oil Reserves Donald F. Larson 33716

WPS1656 The Investment Decision: A Cherian Samuel September 1996 C. Samuel Re-Examination of Competing Theories 30802 Using Panel Data

WPS1657 Is There an Optimal Structure for Frannie Humplick September 1996 J. Williams Decentralized Provision of Roads? Azadeh Moini-Araghi 82557

WPS1658 Decentralizing Structures for Frannie Humplick September 1996 J. Williams Providing Roads: A Cross-Country Azadeh Moini-Araghi 82557 Comparison

WPS1659 Unemployment Insurance in Algeria. Elizabeth Ruppert September 1996 H. Osselyn Implications for a Labor Market in 36039 Transition

WPS1660 Mind Your P's and Q's: The Cost of Lant Pritchett October 1996 S. Fallon Public Investment is Not the 38009 Value of Public Capital

WPS1661 Determinants of Public Expenditure Susan Randolph October 1996 Z. Bogetic on Infrastructure: Transportation and Zeljko Bogetic 623-7292 Communication Dennis Heffley

WPS1662 From Learning to Partnership: Giorgio Barba Navaretti October 1996 M. Patena Multinational Research and Carlo Carraro 39515 Development Cooperation in Developing Countries

WPS1663 Internal Finance and Investment: Cherian Samuel October 1996 C. Samuel Another Look 30802

WPS1664 Pensions in Germany Monika Queisser October 1996 H. Arbi 34663 Policy Research Working Paper Series

Contact Title Author Date for paper

WPS1665 How Important Are Labor Markets Andrew D. Mason October 1996 D. Ballantyne to the Welfare of Indonesia's Poor? Jacqueline Baptist 87198

WPS1666 Is Growth in Bangladesh's Rice John Baffes October 1996 P. Kokila Production Sustainable? Madhur Gautam 33716

WPS1667 Dealing with Commodity Price Panos Varangis October 1996 J. Jacobson Uncertainty Don Larson 33710

WPS1668 Small is Beautiful: Preferential Trade Maurice Schiff October 1996 M. Patena Agreements and the Impact of 39515 Country Size, Market Share, Efficiency, and Trade Policy

WPS1669 International Capital Flows: Do Punam Chuhan October 1996 T. Nadora Short-Term Investment and Direct Gabriel Perez-Quiros 33925 Investment Differ? Helen Popper

WPS1670 Assessing the Welfare Impacts Dominique van de Walle October 1996 C. Bernardo of Public Spending 31148

WPS1671 Financial Constraints, Uses of Asli Demirguc-Kunt October 1996 P. Sintim-Aboagye Funds, and Firm Growth: An Vojislav Maksimovic 37644 International Comparison

WPS1672 Controlling Industrial Pollution: Shakeb Afsah October 1996 D. Wheeler A New Paradigm Bemoit Laplante 33401 David Wheeler

WPS1673 Indonesian Labor Legislation in a Reema Nayar October 1996 R. Nayar Comparative Perspective: A Study 33468 of Six APEC Countries

WPS1674 How Can China Provide Income Barry Friedman October 1996 S. Khan Security for Its Rapidly Aging Estelle James 33651 Population? Cheikh Kane Monika Queisser

WPS1675 Nations, Conglomerates, and Branko Milanovic October 1996 S. Khan Empires: The Tradeoff between 33651 Income and Sovereignty

WPS1676 The Evolution of Payments in David B. Humphrey October 1996 T. Ishibe Europe, Japan, and the United Setsuya Sato 38968 States: Lessons for Emerging Masayoshi Tsurumi Market Economies Jukka M. Vesala