university of michigan · commercial transactions.1 in form, it is nothing more than a letter from...

UNIVERSITY OF MICHIGAN

THE ROLE OF LETTERS OF CREDITIN PAYMENT TRANSACTIONS

Ronald J. Mann

UNIVERSITY OF MICHIGANLAW SCHOOL

PAPER #00-002

This paper can be downloaded without charge at:

The Social Science Research Network Electronic Paper Collection:http://papers.ssrn.com/paper.taf?abstract_id=214633

THE ROLE OF LETTERS OF CREDITIN PAYMENT TRANSACTIONS

Ronald J. Mann*

TABLE OF CONTENTS

I. The Basic Letter-of-Credit Transaction...................................8

II. Discrepancies in Letter-of-Credit Transactions ...................11A. The Rate of Discrepancies ................................................11B. The Types of Discrepancies .............................................21C. Waiving Discrepancies .....................................................26

III. Why Letters of Credit? ............................................................29A. Roads Not Taken: Rejected Explanations ......................30

1. Assuring a Source of Payment...................................302. Irrational Habit and Path Dependence.....................32

B. The Letter of Credit as a Verification Institution..........371. Verifying the Reputation of the Buyer.....................38

(a) Evaluating the Buyer’s Bank ................................39

* © 1999 by Ronald J. Mann. Professor of Law, The University ofMichigan Law School. I thank the individuals who took time from their busyschedules to allow me to interview them about letter-of-credit practices; I amparticularly grateful to the institutions that kindly allowed me to collect infor-mation from their letter-of-credit records. Because that information was collectedsubject to a variety of confidentiality requirements, I regret that I cannot thankeither of those groups by name. I also thank Jim Barnes for his gracious efforts tohelp me arrange my site visits and Atsushi Kinami for his similarly diligentefforts to arrange my interviews with Japanese bankers. I received usefulcomments from Jim Barnes, Dan Keating, Adam Pritchard, Bob Rasmussen, MarkWest, and Jay Westbrook. David Murrel provided assistance with graphics,Catherine Leggieri and Chris Killen with transcription of the interviews. Finally,I acknowledge with gratitude the continuously generous research support of theCook Fund at the University of Michigan Law School.

For convenience, all citations to provisions of the Uniform CommercialCode without indication of date refer to the current version (that is, takingaccount of revisions made in 1999). Similarly, I refer throughout toINTERNATIONAL CHAMBER OF COMMERCE, UNIFORM CUSTOMS AND PRACTICEFOR DOCUMENTARY CREDITS (ICC Publication No. 500) as the UCP. This paperwas prepared for the February 2000 Michigan Law Review symposium onEmpirical Research in Commercial Transactions.

March 30, 2000 Draft THE ROLE OF LETTERS OF CREDIT 2

(b) Vouching for the Buyer.........................................422. Verifying the Authenticity of the Transaction.........48

(a) Governmental Requirements ...............................49(b) Private Requirements............................................52

3. Verification Inside the Firm .......................................54

IV. Conclusion................................................................................55

Statistical Appendix ................................................................55

* * * * *

The letter of credit is one of the oldest and certainly thesimplest of the payment systems commonly used in moderncommercial transactions.1 In form, it is nothing more than aletter from a financial institution in which the institutioncommits that it will pay specified sums of money upon receiptof specified documents.2 Surprisingly, however, it is not at allclear why businesses choose to incur the transaction costsnecessary to cause banks to issue and process letters of credit.

The classic explanation is a simple one: the letter ofcredit provides an effective assurance of payment from afinancially responsible third party. In that story, the seller isreluctant to ship goods to an overseas buyer because the sellerfears that the buyer will refuse to pay once the goods havebeen shipped. The concern is focused on cross-bordertransactions, because the seller’s efforts to force the distant

1 For discussions of the history, see 2 W. HOLDSWORTH, A HISTORY OF

ENGLISH LAW 570-72 (1922) (discussing English history); Boris Kozolchyk, TheLegal Nature of the Irrevocable Commercial Letter of Credit, 14 AM. J. COMP. L. 395,395-400 (1965) (discussing history from the 12th through 19th centuries); RufusJames Trimble, The Law Merchant and The Letter of Credit, 61 HARV. L. REV. 981,982-86 (1948) (discussing usage in ancient Egypt, Greece, and Rome).

2 For a typical form, see RONALD J. MANN, PAYMENT SYSTEMS ANDOTHER FINANCIAL TRANSACTIONS 218-20 (1999).


buyer to pay are likely to be hindered by the difficulties oflitigating in a distant forum.3 The seller solves that problemby obtaining a letter of credit from a reputable bank. Becausea reputable bank is most unlikely to default on its obligation topay the seller, and because the seller knows that it has anabsolute right to payment once it ships the goods –conditioned only on the seller’s presentation to the bank of thespecified documents – the seller that obtains a letter of creditcan rest assured that it will be able to obtain payment even ifthe buyer would not pay voluntarily.4

The payment-assurance explanation tells a logical andplausible story. The only problem is that it is largely untrue atone important and critical link: the seller’s possession of anabsolute right to payment. One of the first things that Ilearned about letters of credit when I spoke anecdotally tobankers and lawyers familiar with the industry – after hearingthe story summarized above – is that sellers ordinarily do notpresent documents that conform to the requirements of theletter of credit.5

3 Lisa Bernstein’s paper in this symposium also focuses on the special

difficulties of cross-border sales transactions. See Lisa Bernstein, PrivateCommercial Law in the Cotton Industry: Value Creation Through Rules, Norms, andInstitutions (Part V).

4 I myself have told that story. See MANN, supra note 2, at 215-24. Forsimilar accounts in the standard sources, see JOHN F. DOLAN, THE LAW OFLETTERS OF CREDIT: COMMERCIAL AND STANDBY CREDIT ¶ 1.01[3], at 1-7 (rev.ed. 1999); JAMES. J. WHITE & ROBERT S. SUMMERS, UNIFORM COMMERCIAL CODE§ 20-1, at 701-02 (4 th ed. 1995).

5 For published references to the discrepancy problem, see Vincent M.Maulella, Payment Pitfalls for the Unwary: How to Make Your Letter of Credit Work,WORLD TRADE, Apr. 1999, at 76, 76 (“US bankers report that 50% to 60% of allletter of credit document presentations are found discrepant on first exami-nation.”); Martin Shaw, Martin Shaw Claims There Are Better Ways to ReduceDiscrepancies and That ICC Should Take Advantage of Them, DOCUMENTARYCREDITS INSIGHT, Spring 1999, at 11 (reporting the views of “informed observers”that “at least 50% -- some say perhaps 60% or even 70%” of presentations do not


The anecdotal reports of high rates of discrepancies inpresentations against letters of credit puzzled me. Under thestandard payment-assurance account of letters of credit, thewhole transaction hinges on the seller’s having a reliable andassured right to payment by the bank that issues the letter ofcredit. But if the seller often does not bother to collect andsubmit documents that conform to the letter of credit, then theseller has no right to payment at all, but only a request for apayment that will be forthcoming only if the buyer waives thedefects in the seller’s presentation.6 And if the seller’s abilityto obtain payment rests on the buyer’s unconstrained choiceto waive defects in the seller’s presentation, then, I wondered,what is the point of buying the letter of credit instead of thesimpler (and presumably cheaper) course of shipping thegoods and just waiting for payment by the buyer.7 It is not atall surprising that parties to a sale transaction would ignoreformal documentation requirements, but it is a little puzzlingthat they systematically would pay banks to sell them a prod-uct that is conditioned on their compliance with thoserequirements.

Intrigued by that question, I undertook in the summerof 1999 to explore the topic in detail. I collected two sorts ofinformation. First, I made site visits to five separate banks tocollect data on letter-of-credit transactions at those banks.Although the banks are all located in the United States, I triedto visit institutions of sufficient variety to get a good pictureof the industry as a whole. The banks that I visited were (a) a

comply); see also UCP preface, at 4 (“Some surveys indicate that approximatelyfifty per cent of the documents presented under the Documentary Credit arerejected because of discrepancies or apparent discrepancies.”).

6 See UCP art. 14(c) (authorizing issuer of letter of credit to approachapplicant to seek a waiver of discrepancies in presentation).

7 For a brief description of the principal alternate methods of payment,see infra pages 33-36.


large U.S. regional bank headquartered in the Midwest withsignificant letter-of-credit volume; (b) a mid-sized U.S.regional bank headquartered in the Northeast with significantletter-of-credit volume; (c) a major U.S. domestic bankheadquartered in the West with worldwide letter-of-creditoperations; (d) a major foreign bank, with more than one U.S.location, with worldwide letter-of-credit operations; and (e) amajor U.S. bank headquartered in the Northeast withworldwide letter-of-credit operations.8 At each bank Ipersonally collected information on 100 transactions (50“import” transactions, in which the bank acted on behalf ofthe buyer, and 50 “export” transactions, in which the bankacted on behalf of the seller).9 For each transaction, Icompleted a database record that included 25 data points.10Among other things, I determined whether the presentationconformed to the letter of credit and, if it did not, what thediscrepancies were and how the parties responded to them.11

8 In order to obtain access to the files of the banks I had to agree not todisclose the particular banks that I visited.

9 The transactions were selected to provide a random sample of recentlycompleted transactions. Because the object of my study is to examinediscrepancies in presentations, I excluded files in which the seller never pre-sented a demand for payment under the letter of credit. Also, to avoid over-sampling particular transactions, I only collected one profile from each file (evenif the file included numerous presentations on a single letter of credit). At eachbank I continued to select files until I had a total (not counting the excluded files)of 50 import and 50 export transactions.

10 The information – about 12,500 data points – is in a Microsoft Accessdatabase. Copies of that file are available on request.

11 I also collected a variety of background information: whether the letterof credit was confirmed, the countries in which the buyer and seller were located,the time when payment was due under the letter of credit, whether the letter ofcredit permitted multiple draws, the type of goods covered by the letter of credit,how the applicant paid the issuer for drafts on the letter of credit, whether thediscrepancies suggested a contractual default, whether the discrepanciesappeared to be curable, how the export-side bank responded to the documents,whether an export-side bank missed discrepancies that an import-side bank


As a matter of practicality, I relied entirely on the banks’internal documentation of those issues. It is, of course,possible that the banks erred on occasion in their assessmentof discrepancies, but given the point of my study –understanding how parties react to discrepancies – dataregarding their perception of discrepancies is directlyrelevant.12

To supplement the raw data, I also conducted a series often interviews with bankers who engage in letter-of-credittransactions.13 Five of the interviews were with the bank offi-cers that supervised the sites that I visited; the other five werewith officers at other banks with substantial letter-of-creditportfolios (two other large American banks, and three Tokyo-based Japanese banks).14 Those interviews explored the viewsof those individuals about the significance of discrepancies inletter-of-credit transactions.

Part I of the Article briefly describes the basic letter-of-credit transaction that is the object of my study. Part IIdescribes the discrepancies that appear in those transactions,providing detail from the data described above. The data gen-erally support the anecdotal information that led me toconduct the study: the documents presented in my 500 found, how waiver of discrepancies was sought, and how many days elapsedbefore the applicant waived the discrepancies.

12 In any event, it would not have been possible to reexamine thedocuments; in most cases the original documents (which often include transportdocuments that the buyer must use to obtain the merchandise in question) nolonger were in the file.

13 Transcripts of the interviews (redacted to satisfy confidentialityrequirements) are available on request.

14 The two American interviews were conducted by telephone; theJapanese interviews were conducted in person in Tokyo. The two interviewswith American banks were conducted on condition of anonymity. Two of theJapanese banks were Fuji and Sumitomo; the third interview was conducted oncondition of anonymity.


transactions conformed to the letter of credit only 27% of thetime. It also shows, however, that the discrepancies arealmost irrelevant to the payment transaction: in all but one ofmy transactions the discrepancies were waived, so that thebuyer provided full payment for the shipmentnotwithstanding the discrepant presentation.15

Part III considers a variety of reasons other thanassurance of payment that might motivate commercial enter-prises to use letters of credit. Building on material from theinterviews described above, I argue that the most significantstructural reasons for use of the letter of credit rely on theability of the bank that issues the letter of credit to verify thereliability of the purchaser on whose behalf the bank issuesthe credit. I describe two different ways in which the issuercould provide that service. In the first and principal scenario,the willingness of the bank to issue the letter of credit signalsto the potential seller that the purchaser is unlikely towithhold payment for an illegitimate reason. In the secondscenario – less generally applicable but still significant – thebank’s willingness to issue the letter of credit verifies to thegovernment (or another financial institution) the legitimacyof the transaction. Thus, the letter of credit indirectly assistsin the enforcement of currency controls and laws againstmoney laundering. Finally, in a twist on both of thosescenarios, the upper management of a large business can useletter-of-credit requirements to limit the risk of imprudenttransactions by individual employees. In that case, the letterof credit implicitly verifies the transaction and tradingpartners to upper management.

15 Even in that one case (Profile 457), the seller did not refuse payment

entirely, but authorized a discounted payment of 94% of the amount upon whichthe parties originally had agreed.


The thread that ties all of those explanations together isan obvious one: the problem of information asymmetry. In allof those situations, the parties to a transaction (or members ofa firm) design a transaction to include a letter of credit as adevice for responding to an imbalance in informationavailable to those entities at the time the transaction begins.Thus, to use my own terminology, I argue that the letter ofcredit generally serves as a verification institution to resolvethat information problem.16

I. THE BASIC LETTER-OF-CREDIT TRANSACTION

The transaction on which this study focuses is the basiccommercial17 letter-of-credit transaction. That transactionhas two sides: an import side and an export side. On theimport side is the buyer; the seller is on the export side. Eachof those parties ordinarily has a bank, which makes a total offour parties to the transaction. Because the customer of thebank on the import side of the transaction is the buyer, thebank on that side of the transaction normally issues a letter ofcredit, which obligates the bank to pay for the goods upon the

16 For a general discussion of verification institutions, see Ronald J.Mann, Verification Institutions in Financing Transactions, 87 GEORGETOWN L.J. 2225(1999).

17 This article generally examines commercial letters of credit, that is,letters of credit used to provide payment in the ordinary course of a contract forthe sale of goods. I do not study the other principal type of letter of credit, the“standby” letter of credit often used to provide a secondary means of paymentwhen an obligor defaults on some other obligation. See DOLAN, supra note 4, ¶1.04, at 1-20 to 1-24 (discussing the distinction between commercial and standbyletters of credit); MANN, supra note 2, at 372-73 (same). For the sake of com-parison, I did, however, collect information at two banks on discrepancies instandby letter-of-credit transactions. My data on that point, however, are muchmore limited (only 24 files) because it is quite rare for presentations to be madeagainst the kind of standby letters of credit described above. {Presentations arecommon against so-called “direct-pay” standbys, see DOLAN, supra note 4, ¶ 1.06,at 1-45 & n.170, but I excluded those transactions from the sample that Iexamined.} That data is included in the same database as the principal data.


receipt of specified documents.18 Letter-of-credit rulestypically describe the importer as the applicant and theapplicant’s bank as the issuing bank or the issuer of the letterof credit.19

Because the customer on the export side of thetransaction is the seller, the role of the bank on that side is adifferent one. As payment for the anticipated shipment, theseller hopes to receive the funds offered by the letter of credit(and thus is called the “beneficiary” of the letter of credit).20Because the beneficiary and applicant ordinarily are indifferent countries,21 it often is useful for the beneficiary tohave its own bank provide assistance in processing the letterof credit when it is issued by the applicant’s bank overseasand then by forwarding the documents that seek paymentfrom the issuer when the seller ships the goods.22 The

18 See UCC § 5-102(a)(10) (defining a letter of credit as an undertaking to

pay in response to a documentary presentation); see also UCC § 5-108(g)(authorizing the issuer to “disregard” any nondocumentary conditions in a letterof credit); UCP art. 13(c) (same).

19 See UCC § 5-102(a)(2) (defining “applicant” for purposes of letter-of-credit law); UCC § 5-102(a)(9) (defining “issuer” for purposes of letter-of-creditlaw); UCP art. 2 (defining “Applicant” and “Issuing Bank” for purposes of theUCP).

20 See UCC § 5-102(a)(3) (defining “beneficiary” for purposes of letter-of-credit law); UCP art. 2(i) (defining “Beneficiary” for purposes of the UCP).

21 The goods were shipped from one country to another in all but 51(10.2%) of the files. Even that figure may be distorted by an unusually high rateof same-country files at the bank that I refer to as the Major Northeast Bank (33 ofthe 51 same-country shipments). Excluding the Major Northeast Bank, the rate ofsame-country shipments was only 4.5%.

22 The beneficiary presented documents directly to the issuer, withoutretaining its own intermediary financial institution, in only 15 (3%) of my files.Again, that rate may be distorted by an unusually high rate of direct presentationfiles at Bank Number 5 (9 of the 15 direct presentations). Excluding the MajorNortheast Bank, the rate of direct presentations was only 1.5%.


beneficiary’s bank ordinarily takes one of two forms.23 If itonly “advises” the beneficiary of the issuance of the letter ofcredit, its role is limited to processing; it has no direct liabilityon the letter of credit.24 Alternatively, it might “confirm” theletter of credit, in which case the beneficiary’s bank is directlyobligated on the letter of credit25; in that case the beneficiary’sbank pays the beneficiary directly and then forwards thedocuments to the issuer for reimbursement.26

23 The beneficiary’s bank is described in the statute as a “nominated

person,” that is, a person that the issuer permits to process documents from thebeneficiary and obtain payment from the issuer. UCC § 5-102(a)(11).

24 See UCC § 5-107(c); UCP art. 7(a).

25 See UCC § 5-107(a); UCP art. 9(b). In the 250 export-side files that Iexamined, the beneficiary’s bank confirmed the letter of credit in 55 (22%) of thefiles.

26 The UCC grants that right of reimbursement indirectly. UCC Section 5-108(i)(1) grants the issuer a right of reimbursement against the applicant; UCCSection 5-107(a) states that the confirmer has the same rights against the issuer asthe issuer has against the applicant. See MANN, supra note 2, at 230-31.


II. DISCREPANCIES IN LETTER-OF-CREDIT TRANSACTIONS

The principal purpose of my study was to collectinformation about discrepancies in presentations thatbeneficiaries make seeking payment on commercial letters ofcredit. Accordingly, I start by describing what my datasuggest about those discrepancies. My data, of course, canonly be suggestive, because I did not collect information froma valid sample of all letter-of-credit transactions. Although Iattempted to collect my data at banks of varying sizes andtypes, the broad variations from bank to bank within the datathat I collected indicate that a complete picture ofdiscrepancies would require a larger study based on astatistically valid sample of all transactions at all banks.

Nevertheless, the consistency of some things that Ifound provides considerable new information about thedynamics of commercial letter-of-credit transactions and howthey can function with such high rates of discrepancies. Threetopics are particularly probative: the rate of discrepancy invarious kinds of transactions; the types of discrepancies; andthe response of the applicant to the discrepancies.

A. The Rate of Discrepancies

The most fundamental feature of my data is a high rateof discrepancies, as expected based on the anecdotal percep-tions described above. As mentioned above, the presentationsconformed to the letters of credit in only 135 (27.0%) of the500 files that I examined. Although the rates did differ frombank to bank – with a high of 36% and a low of 17%27 – it isclear that conforming presentations are the exception to ageneral pattern of discrepancy. A closer examination of the

27 The conformity rates were 26% at the Midwest Bank, 17% at the Mid-

Sized Northeast Bank, 29% at the West-Coast Bank, 27% at the Foreign Bank, and36% at the Major Northeast Bank.


files, however, suggests that the relatively limited range ofthose numbers probably overstates the uniformity ofdiscrepancy rates in the industry. The underlying variety ofexperience is shown most clearly by dividing the files intoimport and export transactions.

As discussed above, each transaction has an import sideand an export side. Because I collected information abouttransactions from each of the two sides, I can explore thepossibility that discrepancy rates are related to the role thatthe bank is playing. The most obvious reason why the rate ofdiscrepancies might be related to the bank’s role is that a bankon the import side normally is reviewing documents that havebeen reviewed already by the beneficiary’s bank. Hence, allother things being equal, you would expect the rate ofdiscrepancies on the import side of the transaction to be lowerthan the rate of discrepancies on the export side of thetransaction: the beneficiary’s bank could be expected to weedout defective documents in some class of cases, so that themost obviously defective documents would not even beforwarded to the issuer. Similarly, the beneficiary’s bankcould help the borrower to correct simple discrepancies. Theresult of those processes would be that an issuer – acting onthe import side of the transaction – would receive a“cleansed” pool of documents to review, with a lower rate ofdiscrepancies than it would find in documents it reviewedfrom the export side of transactions.

At first glance, however, my data seem to contradictthat understanding, because the rate of discrepancies inimport-side files in my sample (78.4%, 196 out of 250 files)28

28 That rate does not change significantly even if I exclude the fifteen

direct presentations (as to which the “cleansing” hypothesis is not relevant). Allof the direct presentations (obviously) were import transactions (because thebeneficiary-seller came straight to the issuer without using an intermediaryexport-side bank). Excluding those transactions (five of which involved con-


actually was significantly higher than the rate of discrepanciesin export-side files (67.6%, 169 out of 250 files).29 On the otherhand, my data about the activities of banks on the export sidedoes suggest that curing of discrepancies by export banks is animportant aspect of the system. As mentioned above, thebeneficiary’s banks identified discrepancies in 169 of the 250sets of documents submitted to them. But the export-sidebanks were able to obtain complying documents in 68 (40.2%)of those 169 files, leaving discrepancies in only 101 files. Thus,the beneficiary’s banks were able to forward documents thatcomplied in a total of 149 (59.6%) of their 250 transactions.

TABLE ONEDISCREPANCY RATES

NUMBER OFDISCREPANCIES

PERCENTAGE OFDISCREPANCIES

IMPORTTRANSACTIONS

196/250 78.4%

EXPORTTRANSACTIONS

169/250 67.6%

OVERALL 365/500 73.0%

The most salient explanation for those figures is thatAmerican banks and exporters are much more concernedabout documentary compliance than the average overseasexporter and bank transmitting documents toward thiscountry.30 That is seen most starkly by comparing the 22% forming documents), the import discrepancy rate would have been 79.1% (186out of 235).

29 The correlation between the likelihood of discrepancy and the side ofthe transaction is significant at the 99% level. See infra Statistical Appendix.

30 Indeed, more than one banker suggested that in Asia banks offer astandard product in which the seller agrees up front that its bank will not exam-


rate31 of incoming conforming documents (documents thatAmerican issuers receive from overseas export-side banks) tothe 60% rate of outgoing conforming documents described inthe preceding paragraph (documents that American export-side banks transmit to overseas issuers). The higher concernthat American banks and sellers have for producingcompliant documents is illustrated by a comparison of thedomestic figures to the total number of complying documentsforwarded by overseas banks. The 22% figure (54 out of 250files), which includes both initially compliant documents and

ine the documents, but instead will forward them immediately to the issuerwithout determining whether they comply. See Notes from Site Visit to ForeignBank, at 1 (Aug. 26, 1999 – Aug. 27, 1999) [copy on file with author] [hereinafterForeign Bank Site Visit Notes]; Notes from Site Visit to Midwest Bank, at 4 (July28, 1999 – July 29, 1999) [copy on file with author] [hereinafter Midwest Bank SiteVisit Notes]. It is not likely that the pattern that I discern is permanent. Morethan one banker suggested that a reverse pattern – more compliant documentscoming into the United States than going out – was characteristic in earlier years.See Telephone Interview with Manager, Trade Service Issues, Second MajorNortheast Bank (Sept. 21, 1999) [transcript on file with author] [hereinafterSecond Major Northeast Bank Telephone Interview] (transcript at 5-6); TelephoneInterview with Vice President and Operations Manager, West-Coast Bank (Aug.12, 1999) [transcript on file with author] [West-Coast Bank Interview] (transcriptat 4); see also Interview with Yutaka Abe, Senior Manager, Overseas BusinessDivision, The Fuji Bank, Limited, Tokyo (June 15, 1999) [transcript on file withauthor] [hereinafter Fuji Bank Interview] (transcript at 3) (suggesting thatdiscrepancies formerly were much higher in Japanese import transactions thanthey are now).

31 That rate is simply the flip side of the 78% discrepancy rate in import-side transactions. I also should emphasize that the rate differs considerably frombank to bank. In my data, it varied from 72% to 82%. One other American bankerwith a large portfolio told me that the discrepancy rate in his import portfoliowas only 60%. Second Major Northeast Bank Telephone Interview, supra note 30(transcript at 2). It also plainly varies significantly from customer to customer.One banker explained that although some of his customers submitted documentsthat complied 99.9% of the time, others submit documents that are discrepant90% of the time. Second Major Northeast Bank Telephone Interview, supra note30 (transcript at 15); see also Maulella, supra note 5 (“[S]ome exporters report thatover 95% of the document presentations are in order; other exporters report a 95%frustration rate.”).


initially defective but cured documents, is less than either (I)the number of initially defective documents cured byAmerican export-side banks (40%, 68 out of 169) or (II) thetotal number of initially compliant documents submitted byexporters in this country (32%, 81 out of 250).32

TABLE TWOINGOING AND OUTGOING COMPLIANCE

NUMBER OFDISCREPANCIES

PERCENTAGE OFDISCREPANCIES

INCOMINGCOMPLIANCE RATE

54/250 22%

TOTAL OUTGOINGCOMPLIANCE RATE

149/250 59.6%

Initial Compliance(by Customers)

81/250 32%

Cured (by Bank) 68/169 40.2%

Although a variety of considerations doubtlesscontribute on a country-by-country basis to the differing ratesof attention to discrepancies, my interviews with bankers sug-gest that the relative reliability of the American commercialand banking systems is the leading general cause. The idea isthat – at least as compared to the median trading partner –the party sending goods into the United States tends to be lessconcerned about the likelihood of malfeasance by the UnitedStates purchaser than a corresponding United States exporter

32 The 32% figure is the flip side of the 68% discrepancy rate in American

export-side transactions.


considering the likelihood of malfeasance by an overseaspurchaser.33

That certainly is not true with respect to all countries inwhich the trading partners of American companies arelocated, but it is true with respect to many less-developedcountries that have relatively unstable economic conditionsand relatively undeveloped legal systems. Together, thoseconditions can make the reliability of the letter of credit lesscrucial for the party selling goods into the United States thanit is for the party selling goods out of the United States.34Accordingly, it should be no surprise that the Americanexporter sending goods overseas (and its bank) will workharder to ensure that they have complied with the conditionsof the letter of credit than an overseas exporter sending goodsinto the United States (and its bank).35

33 See West-Coast Bank Interview, supra note 30 (transcript at 7) (“Our

whole setup is based on not sending documents out of here which a bank over-seas can find discrepancies with … because we know that in some areas of theworld it’s a tendency to find discrepancies for the sole purpose of coming upwith a discount.”). Interestingly, one banker suggested that banks followcountry-by-country conditions so closely that they step up the vigilance of theirdocument checking in countries (such as Asian countries in recent years) thatappear to be undergoing particular crises. See Telephone Interview with VicePresident, Bank Number 1 (Aug. 6, 1999) [transcript on file with author] [here-inafter Midwest Bank Interview] (transcript at 12-13). As he explained, althoughthe banks in the country under stress might honor discrepant documents 99% ofthe time under normal conditions, “in hard times they might reject half of them.”Id. (transcript at 13).

34 Part III explains in more detail the ways in which a letter of credit is“reliable” or “unreliable.”

35 Some loose evidence of the less stringent document-review practicesoverseas seems to be apparent from the perception of Japanese banks that there isa very low rate of discrepancy in the documents they receive from their clients inexport transactions. See Interview with Deputy General Manager (SpecialAssignments), Deputy General Manager, and Senior Manager, AnonymousJapanese Bank, Tokyo (June 17, 1999) [transcript on file with author] [hereinafterAnonymous Japanese Bank Interview] (reporting a discrepancy rate on export


Indeed, my interviews suggest that in many countriesthe export-side bank transmitting goods into the United Stateswill not even bother to examine the documents before for-warding them.36 Thus, in those cases, the bank makes noeffort at all to cure discrepancies; that is a far cry from theAmerican banking practice that (in my data set) cured about40% of the discrepant documents submitted by theircustomers.

Because I did not examine export-side transactions inthe files of any overseas banks, I have no direct observationsof their effort (or lack of effort) to cure defective documents. Ican, however, compare the frequency with which documentswith curable defects were not cured in transactions cominginto and going out of the United States.37 My 250 exporttransactions included 236 transactions outbound from to the

transactions of only 35%); Interview with Hiroshi Higuma, Assistant Manager,Corporate Planning Dept., The Sumitomo Bank, Limited, Tokyo (June 21, 1999)[transcript on file with author] [hereinafter Sumitomo Interview] (transcript at 3)(reporting a discrepancy rate on export transactions of only 15% {albeit after cureefforts at the branch-bank level}).

Those rates suggest to me that a considerably different standard fordocument examination prevails in Japan than the one that prevails in Americanbanks reviewing documents received from Japan. On that point, my limitedinformation does not suggest that documents coming into this country fromJapan are significantly more likely to comply than documents from other coun-tries: only 4 of the 13 Japan-based import transactions (31%) included complyingpresentations. That rate is not significantly different from the overall import rateof 21.6% compliance (78 out of 250 transactions). It is difficult to test thathypothesis more strictly, however. The most obvious test would require usingAmerican-trained examiners to study a set of documents also submitted toJapanese examiners; that type of test is beyond the resources and time availablefor this study.

36 See supra note 30 and accompanying text (mentioning references to thatpractice).

37 At the time I examined the files I included for each file a notationreflecting my judgment as to whether it was easily practicable for the beneficiaryor the beneficiary’s bank to cure the defect in question.


United States, 36 of which (15.3%) included curablediscrepancies that were not cured. By contrast, my 250 importtransactions included 178 transactions inbound to the UnitedStates, 79 of which (44.4%) included curable discrepanciesthat were not cured.38

Another explanation offered by one banker is thediffering types of credits that are typical for the two sets oftransactions: incoming shipments into the United States aremuch more likely to be part of substantial long-term relation-ships (as to which discrepancies are less important) while out-going shipments from the United States are much more likelyto be “one-off” transactions as to which the shipper’srelational protections are limited.39 Another banker offered asimilar transaction-based explanation. He explained thatprice changes that make opportunism more beneficial aremore likely to occur on commodities shipments (whichcharacterize transactions going out from the United States)than on manufactured-goods shipments (which characterizetransactions coming into the United States).40 My data on thetype of good covered by the shipment are insufficientlyobjective to be analyzed statistically, but it does seemplausible that the outgoing (commodities) shipments that hedescribes would be larger than the incoming (finished goods)

38 I use the total number of transactions as the denominator, rather than

the number in which discrepant documents initially were presented, because I donot know the number of the import transactions in which discrepant documentsinitially were presented.

39 See Midwest Bank Interview, supra note 33 (transcript at 11-12); see alsoInterview with Vice President and Manager, International Operations, BankNumber 2 (Aug. 5, 1999) (location not disclosed to protect anonymity) [transcripton file with author] [hereinafter Mid-Sized Northeast Bank Interview] (transcriptat 12) (explaining that relational considerations are the prime motivation forbuyers to waive discrepancies in documents presented for payment on letters ofcredit).

40 See Second Major Northeast Bank Telephone Interview, supra note 30(transcript at 6).


shipments that he describes. As I explain below,41 my datashow a significant inverse relation between amount and therate of discrepancy. That data could be interpreted to supporthis explanation. It seems to me, however, that the sizecorrelation is explained more aptly by the simple notion thatparties expend more effort on compliance when more moneyis at stake.

41 See infra Statistical Appendix (reporting an inverse correlation

statistically significant at the 99% level between the amount of the letter of creditand the likelihood of discrepancy).


Summarizing, Figure Two illustrates what I take to bethe most likely explanation of the data, the existence of fourdifferent rates of discrepancy. In transactions exportinggoods from the United States, my data suggest that thebeneficiaries submit discrepant documents about 68% of thetime, and that the cure efforts of banks lower the discrepancy


rate to about 40%.42 In transactions importing goods into theUnited States, there is relatively little effort to cure defects, sothat the initial presentation discrepancy rate probably43 isabout the same as the 78% discrepancy rate in documentsforwarded to United States issuers.44

B. The Types of Discrepancies

My data also provide considerable detail about thetypes of discrepancies that appear in commercial letter-of-credit transactions. That detail is important because itprovides the context necessary to gain a balancedunderstanding of the significance of the discrepancies.

42 Candor requires me to point out that the evidence I collected in Japan

does not robustly support that understanding of the American data. The anec-dotal information from Japanese bankers indicated that the rate of discrepanciesin transactions coming into Japan, although much less uniform, was relativelylow. See Fuji Bank Interview, supra note 30 (transcript at 3) (20-30% discrepancyrate on Japanese import transactions); Sumitomo Interview, supra note 35(transcript at 2) (25% discrepancy rate on Japanese import transactions).Somewhat inconsistently with the other data, another Japanese bank (on condi-tion of anonymity) reported a discrepancy rate on import transactions of 67%.Anonymous Japanese Bank Interview, supra note 35 (transcript at 2-3). Assuggested above, however, the lower Japanese discrepancy figures also might beattributable to differing standards of document examination. See supra note 35.

43 Because I did not examine export-side transactions in the files of anyoverseas banks, I have no direct observation of that rate.

44 That two-tiered understanding of my data is corroborated by a similar(although slightly more discrepant) set of figures provided me under condition ofanonymity from a banker at a major Northeastern bank. His data suggested thatthe highest rate of discrepancy (60% in his portfolio) is on incoming importtransactions, that there is a lower discrepancy rate (50%) on outgoing exporttransactions, and that his bank cures about half (25%) of the export discrepanciesbefore the documents are forwarded overseas. See Telecopy from Second MajorNortheast Banker 1 (Sept. 21, 1999) [copy on file with author].


The most important thing that the data about thediscrepancies does is to rebut any suggestion45 that the highdiscrepancy rates reported by anecdote and found in the filesthat I examined can be attributed to hypertechnical documentexamination practices.46 For one thing, more than a quarter ofthe presentations that were found to be discrepant (98 out ofthe 343 files, 28.6%)47 appeared to suggest a contractualdefault by the seller: not just a failure to comply with thetechnical provisions of the letter of credit, but a failure tocomply with the substantive provisions of the underlyingsales contract.48 Moreover, although the relatively subjective

45 See, e.g., Boris Kozolchyk, The UNIDROIT Principles as a Model for theUnification of the Best Contractual Practices in the Americas, 46 AM. J. COMP. L. 151,160-61 (1998) (arguing that “arbitrary” judicial evaluations of discrepancies“had encouraged bad faith practices by bankers and their customers wherebymeaningless defects in the documents tendered by beneficiaries were used tojustify non-payment of letters of credit”).

46 Although the examination practice might seem hypertechnical to theoutsider, American banks have expended considerable effort to enhance theobjectivity of document examination. The most recent effort is the promulgationof a detailed document describing the items that banks should check on the mostcommonly presented types of documents. See U.S. COUNCIL ON INTERNATIONALBANKING, INC., STANDARD BANKING PRACTICE FOR THE EXAMINATION OFLETTER OF CREDIT DOCUMENTS (1996); see also Boris Kozolchyk, The “BestPractices” Approach to the Uniformity of International Commercial Law: The UCP 500and the NAFTA Implementation Experience, 13 ARIZ. J. INT’L & COMP. L. 443, 446-48(1996) (discussing development of the USCIB standards). As doubts about thestandards for document examination overseas suggest (see, e.g., supra notes 33 &35), it is not at all clear that those efforts have made significant progress overseas.

47 Although 365 files were not found in compliance, 22 of those were notexamined: in those cases the applicant approved payment without the need forexamination of the documents. Thus, the information that I report about thetypes of defaults states percentages as a share of the 343 examined files found tobe discrepant.

48 It is not always clear whether the discrepancies suggest a contractualdefault. I made a judgment call at the time I examined the file as to whether Ithought the discrepancies collectively raised a serious doubt about the per-formance by the seller. Because I was interested in the frequency with whichtechnical defects that do not go to the seller’s performance appear asdiscrepancies, I tried to err on the side of assuming that there might be a default.


problem of defective documents was quite common,49 a largenumber of the defaults that were not contractual defaults areplainly objective defects about which no informed documentexaminer could disagree. For example, 75 of the presentations(21.9%) were missing a document required by the letter ofcredit; 62 (18.1%) involved a shipment later than the periodspecified in the letter of credit; 48 (14.0%) involved a latepresentation against the letter of credit50; in 36 (10.5%) theletter of credit had expired51; and in 16 (4.7%) the documentssought payment for an overdraft (an amount that exceededthe balance remaining on the letter of credit).52

49 Defects in documents collectively constituted the largest category of

discrepancy, appearing in 293 (85.4%) of the files. It is particularly difficult todetermine whether defective documents suggest a contractual default, because itis rarely possible to tell from the file whether the defect reflects inadequateperformance or inadequate documentation of adequate performance.

50 Unless the letter of credit stipulates otherwise, documents must bepresented no later than 21 days after the date of shipment. UCP art. 43(a). It wasnot common in the letters of credit that I examined to alter that 21-day period.

51 In addition to the implied deadline for presentation mentioned in theprevious note, each letter of credit includes an express date on which the creditexpires. Any later presentation is defective. UCP art. 42(b); see UCP art. 44(a)(implied extension of expiration date to next business day).

52 In assessing those numbers, it is important to remember that manypresentations contained multiple defects. Thus, the figures in the text count mostof the files multiple times (because most files contained multiple defects).


TABLE THREETYPES OF DISCREPANCIES

Defective Documents 293 (85.4%)

Missing Documents 75 (21.9%)

Late Shipment 62 (18.1%)

Late Presentation 48 (14.0%)

Expired 36 (10.5%)

Overdraft 16 (4.7%)

Incorrect Shipment 14 (4.1%)

Partial Shipment 7 (2.0%)

Other 2 (0.6%)

Total Discrepancies 554

Total Discrepant Files 365

Files Not Examined 22

Files Examined 343

That is not to say, however, that the discrepanciesgenerally suggested a serious failure of performance by theseller on the underlying sales contract. On the contrary, in201 of the files (58.6%), the defects found on examination didnot suggest a contractual default by the beneficiary.53 Thatcould be true either because the defect was a minordocumentary defect – an inadequate signature on a bill oflading or a technical inaccuracy in describing the collateral toname two common examples – or for several other reasons.

53 As mentioned above (supra note 48), I treated ambiguous cases as

suggesting a default. Thus, that 59% figure is, if anything, understated.


For example, the defects of late presentation (48 files, 14.0%),54expiration (36 files (10.5%), and overdraft (16 files, 4.7%) maybe objectively indisputable problems with the presentation,but they do not suggest any default by the seller on theunderlying contract.

TABLE FOURDISCREPANCIES AND CONTRACTUAL DEFAULT

DISCREPANCIESINDICATING DEFAULT

98/343 (28.6%)

DISCREPANCIESINDICATING DEFAULT

201/343 (58.6%)

RESIDUAL (UNABLE TOCLASSIFY)

44/343 (12.8%)

The data highlight one structural difficulty with theletter-of-credit system. If the system worked perfectly, itwould use documentary presentations to sort transactionsbased on the beneficiary’s performance: the documents wouldcomply when the beneficiary had performed as agreed andthe documents would not comply when the beneficiary hadnot performed as agreed.55 As the data show, however, myfiles indicate that more than half of the examined filesincluded defects that vitiated the beneficiary’s right to collectpayment, even though those same defects did not call into

54 Late presentation (which I did not treat as a clear contractual default)must be distinguished from late shipment, which I treated as a contractualdefault.

55 I leave to one side the question whether the performance that thesystem seeks is performance up to standards set by unenforceable industrynorms or performance up to the standards of judicially enforceable contractualprovisions. For discussion of the distinction between those two different types ofstandards, see Bernstein, supra note 3.


question the caliber of the beneficiary’s performance on thecontract. As industry observers recognize, the poor fitbetween discrepancies and default suggests a problem withthe letter-of-credit system.56

C. Waiving Discrepancies

Because the frequency of discrepancies motivated mystudy, the information described above was not at all surpris-ing; it only confirmed a variety of statements made to me pre-viously. What did surprise me, however, was the response tothe defaults. Being cynical by nature, I expected to find thatapplicants seize on the discrepancies in a significant numberof cases – including many cases in which the discrepancies didnot suggest a contractual default – as a basis for delaying orwithholding payment to the beneficiary on the letter of credit.

My data suggest that my expectation was completelywrong: applicants almost always waive the discrepancies andpermit full payment to the beneficiaries under the letter ofcredit. In the 365 files that I examined with discrepancies, theapplicants waived the discrepancies and permitted full pay-ment in every file but one (99.7% of the files). And in that filethe applicant did not refuse payment; it permitted payment of94% of the agreed amount.57 Thus, in the 500 letter-of-credittransactions that I examined, I did not find a single case in

56 See Shaw, supra note 5, at 11-12. Bob Rasmussen has pointed out to me

that the poor fit is less relevant if you accept the idea (proposed in Part III of thisArticle) that letters of credit operate primarily as a device for verifying thereliability of the applicant rather than as a device for assuring payment. In anyevent, analysis of that problem is far beyond the scope of this Article.

57 Profile 457. That transaction included some defects that did notdemonstrably indicate contractual defaults: expiration, late presentation, amissing document (one counterpart of a purchase order), and a defectivedocument (technically inaccurate shipping terms). It also included, however, alate shipment, which normally would reflect a default on the underlyingcontract.


which the applicant refused payment on the letter of credit.58Nor does my experience seem to be atypical. Rather, everyinterview subject with whom I discussed that topic expressedthe view that the rate of refusal would be less than onepercent of the discrepant files.59

58 In all of my files, the banks immediately paid upon receipt of the

waiver. It is, however, the view of the industry that the bank is not bound by thewaiver issued by its customer the applicant. See ICC, MORE QUERIES ANDRESPONSES ON UCP 500: 1997, at 14 (1997) (Response 254) (“In the event thatdiscrepancies are observed in a presentation of documents and the issuing bank[gives adequate notice of the discrepancies], the issuing bank is under no obli-gation to take up the documents, even if a proper waiver of the discrepancies isreceived from the applicant.”); ICC, MORE QUERIES AND RESPONSES ON UCP500: 1997, at 28 (1997) (Response 267) (“The receipt of a waiver from the appli-cant, either direct or via the beneficiary, does not bind the issuing bank to acceptthe documents. The decision of whether or not to comply with the waiver is onefor the issuing bank to decide in its sole judgment.”). It appears that the onlysubstantial reason that a bank would decline to accept such a waiver would bedoubts about its ability to obtain reimbursement from the applicant. See SecondMajor Northeast Bank Interview, supra note 30 (transcript at 6-7).

59 See Foreign Bank Site Visit Notes, supra note 30, at 1 (estimating threerefusals each year out of 100,000 presentations, for a rate of .03%); TelephoneInterview with executive from Bank Number 5 (Oct. 8, 1999) [transcript on filewith author] [hereinafter Major Northeast Bank Interview] (transcript at 10) (“Iwould have said it was a small fraction of one percent.”); Telephone Interviewwith Vice President, Technical Consultant for Global Trade Services, MajorMidwest Bank (July 19, 1999) (interview conducted on condition of anonymity)[transcript on file with author] [hereinafter Major Midwest Bank TelephoneInterview] (transcript at 7) (“I would say ninety-nine percent of the documents[that are discrepant are paid].”); Second Major Northeast Bank TelephoneInterview, supra note 30 (transcript at 5) (“[A]t the end of the day ninety-ninepoint nine percent of the documents they present, whether they carrydiscrepancies or not, are paid.”); Second Major Northeast Bank TelephoneInterview, supra note 30 (transcript at 17) (banker who does 100,000 transactionsa year suggesting that “I would be very comfortable in just guessing [that thenumber of rejected documents is] less than a hundred. It may be less than ten.”);West-Coast Bank Interview, supra note 30 (transcript at 7) (agreeing with myexpectation of finding only one or two refused presentations in my five-hundredfile study).

The pattern appears to be similar in Japan. See Anonymous JapaneseBank Interview, supra note 35 (transcript at 7) (suggesting rejection of about ten


Even more surprising than the rate of waiver was thepromptness with which the waivers are forthcoming from theapplicants. Several of my interview subjects suggested thatapplicants – even if they ultimately will permit payment –commonly delay payment for a significant period of time toreflect dissatisfaction with the beneficiary’s performance inthe transaction. But in my files waiver generally was quiteprompt: Of the 196 import files in which there werediscrepancies, the applicant in more than half of the files –103 (52.6%) – waived the discrepancies on or before the firstbusiness day after the date on which the issuer contacted theapplicant about the discrepancy. By one week after the issuercontacted the applicant, discrepancies had been waived in 165(84.1%) of the files.60 By four weeks (twenty business days)after the issuer contacted the applicant, only six files (3.1%)remained unaccepted.

documents out of a monthly volume of 17,000 transactions); Fuji Bank Interview,supra note 30 (transcript at 9) (reporting rejection of documents or reduction ofamount to be paid in about ten out of every 18,000 transactions); SumitomoInterview, supra note 35 (transcript at 4) (suggesting rejection of ten documentsout of a monthly volume of one thousand noncomplying transactions).

60 One week (five business days) serves as a rough guide of a timelyresponse, because the issuer generally needs to respond to the bank that pre-sented the documents within seven business days after it receives the documents.See UCP art. 13(b) (calling for a response within “a reasonable time, not to exceedseven banking days”). If the issuer takes two business days to examine thedocuments, that would leave five business days for the applicant to decidewhether it wishes to waive any discrepancies without preventing the issuer fromtransmitting a timely acceptance of the documents. Of course, if the issuer doesnot receive a waiver by the seventh business day, it still can pay later, by sendinga notice rejecting the documents on the seventh business day, followed by a laternotice accepting the documents with discrepancies. For a thorough discussion oftypical practice, see International Finance Services Association, Statement ofPractice: Reasonable Time for Examination & Notice of Dishonor , in THE 1999ANNUAL SURVEY OF LETTER OF CREDIT LAW & PRACTICE 311 (James E. Byrneed.) .


As a practice of accommodation in transactions amonglong-time partners, the rate and pace of waivers might seemcommonplace. But most commercial letters of credit probablyare used in other contexts, one-shot transactions or other con-texts where relational constraints are less forceful.61 In thosecontexts, that rate of waiver seems truly startling.

III. WHY LETTERS OF CREDIT?

The data presented in Part I display an odd andpuzzling picture. Commercial parties pay substantial fees tobanks to use letters of credit in their transactions.62 Thebeneficiaries then usually submit documents that do notconform – which means that their right to payment under theletter is jeopardized– but the applicants then almostuniversally waive the defects with startling promptitude,notwithstanding the frequent contractual defaults displayedon the face of the documents presented by the seller.

The remainder of this Article offers some tentativeexplanations for that pattern. Given the worldwide use of thecommercial letter of credit, and the wide variety of contexts inwhich it is used, no single explanation can capture all of themotivations for its use. I do think, however, that the informa-tion collected in this study allows me to make some progressin understanding the transactions. Thus, subpart A starts by

61 See infra pp. 36-37 (discussing reasons why parties select letters ofcredit instead of other payment mechanisms).

62 The fees differ significantly from market to market, and from customerto customer (with better customers paying much less). As a general matter,however, the total fees for the banks issuing and processing the letter of credit arelikely to approximate ¼ of one percent of the amount of the letter of credit. On a$1,000,000 sale of goods, then, use of a letter of credit would require about $2,500to the total transaction costs. See MANN, supra note 2, at 217. Because allmethods of payment in cross-border transactions involve some out-of-pockettransaction costs, the excess cost of the letter of credit is some portion of that$2,500.


rejecting some possible explanations – most prominently irra-tional path dependence and mistake by the businesses thatbuy letters of credit. Subpart B then presents three partialexplanations that seem plausible to me. The first two arerelated: I argue that the principal general motivation for theletter of credit is to allow the buyer or some other financial orregulatory institution to verify the applicant’s assertionsabout its own reliability and the legitimacy of the transaction.The last, less significant, explanation is use by a firm to controlthe behavior of its low-level agents.

A. Roads Not Taken: Rejected Explanations

1. Assuring a Source of Payment

The payment-assurance story is the classicunderstanding of letters of credit. In that story, the keybenefit that the letter of credit provides to the seller is that itlargely removes the risk that the seller will ship its goods firstand that the buyer, having the goods in a distant locale, willwithhold payment from the seller.

It should be clear from Part II, however, that thepayment-assurance method cannot provide a robust explana-tion for the general use of letters of credit.63 The payment-assurance story makes sense only if the seller at the time itacquires the letter of credit generally expects to be in aposition to use the letter of credit to force the buyer to pay.But as my data suggest, an experienced seller would have tounderstand that it usually will not be able to force paymentfrom the buyer, because it usually will not submit documentsthat comply with the letter of credit. Thus, in most cases, the

63 That is not to say that it has no explanatory force. As I discuss below,

a similar motivation does lead sellers to buyers in weak-currency countries toinsist on letters of credit to assure the availability of hard currency to pay for thetransaction. See infra text accompanying note 116.


seller’s right to payment will depend entirely – at least as alegal matter – on the grace of the buyer in waiving thediscrepancies in the documents submitted by the seller.

To be sure, it is possible to articulate a weaker version ofthe payment-assurance story that fits better with a high rateof discrepancies. For example, if the costs of letters of creditare quite small, if sophisticated sellers expect buyers to renegeand try to withhold payment quite rarely, and if it isexpensive to submit documents that comply, then it might berational to use letters of credit generally, but accept a highrate of discrepancy, just to have the letter of credit there insome of the rare cases in which buyers try not to pay. Indeed,the existence of substantial cure efforts, illustrated by mydiscussion above of American export-side banks,64 shows thatthere are some contexts in which that story has some force.

But the large frequency of easily curable defects in thefiles that I examined suggests that it is unlikely to providemuch of a general explanation. Defects that the bank cancause the seller to cure generally are defects that could havebeen avoided more cheaply in the first instance. To use acommon example from the files that I examined, surely itwould be cheaper for a seller to write a draft correctly the firsttime than it would be to write a defective draft, pay the bank’sdiscrepancy fee,65 and then take the time and effort a weeklater (after prodding by the bank) to produce a compliantdraft. Even experienced and careful sellers would makemistakes from time to time, which banks would catch, but auniverse in which banks are able to cure defects in more thanone-fourth (27%, 68 out of 250 export files) of the

64 See supra pp. 13-18.

65 At the banks that I visited, discrepant presentations universally wereassessed a fee, which ranged from a low of $25 to a high of $75.


presentations made to them is unlikely to be a universe inwhich sellers are trying hard to avoid obvious mistakes.66

In sum, my general impression – which hardly can bemade more definitive than that – is that assurance of paymentis at most an incomplete reason for the use by businesses ofletters of credit in their overseas sale-of-goods transactions. Ifthere is a reason for the wide use that businesses make ofletter of credit, then, I am convinced that it must include somebenefit that accrues at the time of issuance, not in theassurance of payment at the end of the transaction.

2. Irrational Habit and Path Dependence

There is, of course, another possibility, that there is norational explanation for the use of letters of credit: businessesuse them not because of the benefits they provide, but becauseof a combination of practical factors such as a failure tounderstand how letters of credit work in practice, coupledwith some path-dependent habit of usage. To put it moredirectly, that perspective suggests that businesses buy lettersof credit from banks by mistake – because they always have –and that if they fully understood the costs and benefits ofletters of credit they would not buy them.

As with the payment-assurance story from the previoussection, it surely would be wrong to reject the mistake storyout of hand: businesses cannot have perfect comprehension ofeverything that they do; from time to time they surely enterinto transactions on terms attributable to imperfections in

66 That is not to say that sellers do not try at all. The strong correlation

(significant at the 99% level) between the size of the credit and the likelihood ofconformity (see infra Statistical Appendix) does suggest that sellers try harder toproduce compliant documents when more money is at stake. Indeed, asmentioned above, the existence of any effort at all – even in response to the bank’sprodding – shows that sellers perceive some value in compliance.


their understanding.67 But two factors make me doubt thegeneral applicability of that mistake theory.

The first is the most obvious: the sophistication of theparties involved. Many of the users of letters of credit arelarge and sophisticated companies. In my study, for example,I reviewed files in which several prominent discount retailersand department stores obtained letters of credit to pay theiroverseas suppliers. And the uses were not occasional or hap-hazard; on the contrary, it was clear from the files that thosecompanies have large letter-of-credit relationships that covera substantial portion of their sales activity.68 Absent some evi-dence that I have not seen, it is difficult to believe that thosecompanies would organize such a large number oftransactions in a way that systematically, repeatedly, andpointlessly increases the cost of the transactions.

The larger framework of the institutions for providingpayment in cross-border sale transactions also casts doubt onthe mistake explanation, because it shows that businessesmake a conscious and deliberate choice to select the letter ofcredit from among a variety of competing paymentinstitutions. Generally, sellers and buyers can choose from

67 For comments to that effect in this context, see Fuji Bank Interview,

supra note 30 (transcript at 13-14) (discussing the lack of sophistication bysmaller companies doing international trade and explaining that “there’s a per-ception for the Japanese company that the LC is very credible, reliable. And oncethey receive an LC they feel like they have completed the transaction.”); SecondMajor Northeast Bank Telephone Interview, supra note 30 (transcript at 15, 17)(suggesting that customers focus on the fact that their letter-of-credit transactionsget paid rather than the risk of nonpayment).

68 The size of such a relationship was particularly evident at two of thebanks (Banks 1 and 4), whose file-numbering systems included a separate filingsystem – with separate numbers and file locations – for the transactions of twoprominent retailers that are their largest letter-of-credit customers. Each of thosecustomers provided, on a conservative estimate, more than 20% of the bank’sletter-of-credit work.


four significantly different methods of providing payment incross-border transactions. Ranging from the most favorableto the seller to most favorable to the buyer, the options includeprepayment, payment by letter of credit, payment bydocumentary collection, and open account. 69

The first and the last are the simplest and cheapest, butcreate the greatest possibility for opportunistic misconduct bythe trading partner. In the prepayment transaction, the buyerforwards payment before the seller ships the goods. In theopen-account transaction, the seller ships the goods withoutany formal assurance that the buyer will forward paymentwhen the goods arrive. Thus, in each of those transactions oneparty performs completely in the first instance, trusting theother party to respond by performing in turn.70

In between those two polar choices are twointermediate choices, in which the performance of the partiesis intertwined, with each party taking substantial stepstoward performance before either party completes itsactivities. In the letter-of-credit transaction, as discussedabove, the seller waits to ship until it receives a letter of creditissued on behalf of the buyer. The buyer, in turn, withholdspayment until it receives adequate evidence that the shipmenthas occurred (the evidence being provided by the documentsrequired for payment under the letter of credit).

The documentary collection transaction (or, commonly,simply a “collection” transaction) is another intermediateoption, cheaper but less protective of the seller than the letter-of-credit transaction.71 The seller ships the goods without any

69 See Mid-Sized Northeast Bank Interview, supra note 39 (transcript at 6-7).

70 See Mid-Sized Northeast Bank Interview, supra note 39 (transcript at 6).

71 For a more detailed summary, see MANN, supra note 2, at 457-66. Fordescriptions by a banker, see Mid-Sized Northeast Bank Interview, supra note 39


previous action by the buyer to commit to payment. But thegoods are covered by a document of title transmitted throughbanking channels. In the typical (though not universal) wayof arranging the transaction, the buyer cannot obtain thedocument of title, and thus cannot obtain the goods that thedocument covers, until it pays the bank for the goods.72 Thetransaction is less favorable to the seller than the letter ofcredit because the buyer has no obligation to take up the docu-ments.73 Hence, the buyer cannot get the goods withoutpaying for them, but the seller cannot force the buyer to pay;that means, if the buyer chooses not to pay, that the seller isleft with goods that it has shipped to an overseas location,with no local buyer for them.74 That is better than an open

(transcript at 6-7). Because two of the banks that I visited maintained records ondocumentary-draft transactions at the same sites as they maintained letter-of-credit records, I collected information on documentary-draft transactions at thosesites (50 records at each bank for a total of 100). Those records are in the samedatabase as the other data.

72 See Mid-Sized Northeast Bank Interview, supra note 39 (transcript at 7(“If the documents are titled properly then no pay, no documents for mer-chandise.”). Two common variations use nonnegotiable documents of title. Inone, the goods are consigned to the collecting bank; that has substantially thesame effect as a negotiable shipment, because the buyer usually must pay toacquire the goods. See Mid-Sized Northeast Bank Interview, supra note 39(transcript at 8-9). The other common variation, particularly in shipments by air,uses nonnegotiable documents and ships directly to the buyer. In thattransaction the buyer can obtain the goods without paying the bank for them. SeeMid-Sized Northeast Bank Interview, supra note 39 (transcript at 8). Thus, thattransaction provides the seller little more protection than the open-accounttransaction discussed above. The transactions proceeded in that less protectivefashion in 33 out of the 96 (34%) collection transactions for which I couldexamine the relevant documents.

73 See Midwest Bank Interview, supra note 33 (transcript at 8) (“[A]collection raises the obligation of absolutely nobody to do anything that theydon’t want to do.”); West-Coast Bank Interview, supra note 30 (transcript at 13)(“[Y]ou’re completely putting yourself at the mercy of that party [i.e., the overseasbuyer].”).

74 See Mid-Sized Northeast Bank Interview, supra note 39 (transcript at 7).Buyers declined to pay the banks in 12 out of the 100 collection transactions.


account transaction (where the buyer can get the goodswithout paying for them), but not much. It is, however,significantly cheaper than a letter-of-credit transaction, withbank fees typically fixed in the range of $100-$300.75

Although I have not located definitive statistics, theavailable information makes it clear that all of those optionsare realistic choices. And that is significant because it showsthat the letter of credit is not the automatic response of acustom-bound business. Businesses do not use letters of creditindiscriminately out of habit; rather, they select them fortransactions in which they do not have a relationship with theoverseas party that is good enough to justify engaging in col-lection or open-account transactions. As one banker put it,“there has to be trust between the two before you send thedocuments on collection.”76 Indeed, the best information I That 12% non-payment rate is striking compared to the non-payment rate in theletter-of-credit transactions of less than one-tenth of one percent. The higher non-payment rate would not surprise the bankers to whom I spoke about collectiontransactions. See, e.g., Mid-Sized Northeast Bank Interview, supra note 39(transcript at 12) (discussing difficulties of obtaining payment in collectiontransactions); West-Coast Bank Interview, supra note 30 (transcript at 13-17)(same).

75 See Mid-Sized Northeast Bank Interview, supra note 39 (transcript at15-16) (discussing the different types of charges in letter-of-credit transactionsand collection transactions); West-Coast Bank Interview, supra note 30 (transcriptat 12) (“[O]bviously the cost of a letter of credit is very much higher than acollection. Our collection fees are low – very low in comparison. [Collectionsa]re not a big money maker and they’re looked at more as a service to a customerinstead of an actual money-making product.”); see also supra note 62 (discussingfees for letters of credit). [My estimate of the fees for collection transactions isbased on my review of the files in my database.]

76 Midwest Bank Interview, supra note 33 (transcript at 8); see Mid-SizedNortheast Bank Interview, supra note 39 (transcript at 19-20) (discussing reasonswhy parties choose collection transactions instead of letters of credit); West-CoastBank Interview, supra note 30 (transcript at 12) (“[T]he only reason [collectiontransactions] exist is because there’s a great deal of … trust between the partiesconcerned or in some cases people just would rather take the risk than pay theinitial letter-of-credit fees, which can get pretty expensive. “).


have located indicates that letters of credit are used in onlyabout one-fifth of cross-border sale-of-goods transactionscoming into or out of the United States.77

What that suggests is not a market in which businessesbuy letters of credit for no good reason. On the contrary, itsuggests a market in which businesses use cheaper methods ofpayment whenever the letter of credit is not worth the cost,and select the letter of credit only when it is worth the addedexpense. Generally, it suggests, businesses use letters of creditin one-shot transactions where relational concerns areunlikely to arise, or in the opening stages of what mightdevelop into a long-term relationship, before relationalconstraints become effective. The question that still remainsis why the letter of credit is ever worth the added expense.

B. The Letter of Credit as a Verification Institution

If the letter of credit adds real value to the transactionsin which it is used, and if the value does not come fromproviding an enforceable mechanism of payment to the seller,then the value must come from something else that isprovided by the mere issuance of the letter of credit on behalfof the buyer. The information that I have collected suggeststhat the key is the willingness of the issuing bank to issue theletter of credit: that bank’s issuance of the letter of credit

77 One banker reported to me two sources of data in his possession. Thedata that he considered more reliable suggested that 13% of such transactionswere done by letters of credit, 72% by open account, 4% by documentarycollections, and 2% by cash in advance. Another source (that he considered lessaccurate) reported 29% letters of credit, 52% open account, 12% cash in advance,and 7% documentary collections. See Interview with Group Vice President andHead of Trade Services Product Management, International Trade & AdvisoryGroup, Bank Number 4 (Sept. 2, 1999) (location not disclosed to protectanonymity of bank) [transcript on file with author] [hereinafter Foreign BankInterview] (transcript at 13). Whichever figures are closer to reality, however,both figures suggest that letters of credit are not used routinely without regard tocost.


provides an implicit verification of the reputation of theapplicant and the legitimacy of the transaction. To put itgenerally, that bank’s issuance of the letter of credit verifiesinformation about the buyer and the transaction moreeffectively than direct assertion by the buyer could: the partiesuse the bank as an intermediary and the bank’s issuance of theletter of credit as an institution to resolve informationproblems about the reliability of the buyer and the legitimacyof the transaction.78

1. Verifying the Reputation of the Buyer

In my view, the most general explanation of thecommon usage of letters of credit is the implicit verification ofthe applicant’s reliability and probity that the issuer makeswhen it issues the letter of credit. Structurally, thatverification is a classic example of reputationalintermediation: the applicant/buyer “rents” the issuer’sreputation to allow the beneficiary/seller to verify theapplicant/buyer’s character.79

The key to that arrangement, of course, is theavailability of a reputational sanction against the bank thatissues the letter of credit. On that score, I have little to add,because it is so well known that banks feel a strongreputational interest in their letter-of-credit businesses.80 For

78 For a general discussion of those kinds of “third-party” verification

institutions, see Mann, supra note 16, at 2265-71.

79 The classic explanation of reputational intermediation appear inRonald J. Gilson & Reinier H. Kraakman, The Mechanisms of Market Efficiency , 70VA. L. REV. 549, 618-21 (1984) (discussing reputational intermediation in theissuance of securities); see also Mann, supra note 16, at 2269-71 (generalizing thatidea).

80 The best anecdote I heard about the significance of bank reputation inthe letter-of-credit context involved Chinese banks that were barred by theChinese government from using hard currency to honor previously issued lettersof credit. Notwithstanding the plenary control of the government over bank


that explanation to make sense, however, I need to establishtwo separate characteristics of the transactions in which theletter of credit is used: the relative ease of verifying thereliability of a foreign bank as opposed to a foreign tradingpartner; and the plausibility of treating a bank that issues aletter of credit as vouching for the quality of its client theapplicant. I address those problems in turn.

(a) Evaluating the Buyer’s Bank

The first point is the simpler one, and also the easier todefend. The relevant information problem is the problem thatfaces a seller of goods to a foreign buyer when the sellerattempts to estimate the likelihood that the buyeropportunistically will attempt to withhold payment in thetransaction after the seller ships the goods. Efforts to assessthe reputation of the buyer directly are likely to be costly, ifnot entirely fruitless.81 For one thing, the buyer’s location in aforeign country makes it more costly to collect informationthan it would be if the buyer were located in the same countryas the seller.82 As one banker put it, letters of credit are not asuseful in same-country transactions because “it’s easier to get

activities in China, the officers at those banks still used a variety of arrangements– offsets against overseas funds and the like – to do the best they could to providetimely payment on the letters of credit that they had issued. See Foreign BankInterview, supra note 77 (transcript at 3-4).

81 For a theoretical discussion of reasons why it is difficult for manycountries to develop effective systems for disseminating credible informationabout their businesses, see Bernard S. Black, The Legal and Institutional Pre-conditions for Strong Stock Markets: The Nontriviality of Securities Law (unpublishedSept. 1999 manuscript) [copy on file with author].

82 See Foreign Bank Interview, supra note 77 (transcript at 4) (“If you’reselling to somebody outside the United States it’s very difficult to get good creditinformation so you try to go to [Dun & Bradstreet] or whatever you can. But, youget very sketchy information.”); Major Northeast Bank Interview, supra note 59(transcript at 7).


a credit report cheaper than to force somebody to give them aletter of credit.”83

Also, even if the information is available, it may be lessreliable to the seller than information collected in the seller’sown country.84 Similarly, differences in accounting systemsfrom country to country make it difficult to assess anyobjective financial information that the seller might obtain.Moreover, the sheer number of potential trading partnersworldwide makes the task of maintaining any knowledge offinancial strength and probity daunting at best.

Although those problems make it difficult to assess thefinancial strength and credibility of foreign banks, they aremuch more tractable for foreign letter-of-credit banks thanthey are for foreign trading partners.85 For one thing, theuniverse of reputable letter-of-credit banks is much smallerthan the universe of trading partners: in most countries therewill be only a few banks that plausibly could participate in theglobal letter-of-credit arena. Thus, it should not be surprisingthat there is a substantial amount of information availableabout the activities of the banks in question.86 Second, large

83 Midwest Bank Interview, supra note 33 (transcript at 5).

84 See Foreign Bank Interview, supra note 77 (transcript at 5) (pointing outthat Dun & Bradstreet collects information on foreign companies, but suggestingthat “companies in another country are not so concerned [as Americancompanies] about [the validity of the information that they provide Dun &Bradstreet]”).

85 See Second Major Northeast Bank Telephone Interview, supra note 30(transcript at 14) (“When we’re asked to confirm a credit – we’re making thatdecision for the most part based on the bank. … We understand the bank and weknow the bank and we have a relationship with the bank and we may not have arelationship or know the importer.”).

86 See Foreign Bank Interview, supra note 77 (transcript at 5) (“[T]here’s anawful lot publicized. There’s a lot you can read in The Economist or otherperiodicals, magazines, things about the creditworthiness of various banks andthe countries that they’re in.”).


banks of the relevant markets are much more likely to followinternationally comprehensible accounting conventions thanthe great mass of trading businesses in the foreign country;thus, it is much easier for analysts in the seller’s country toassess information about the buyer’s bank than it is for themto assess information about the buyer itself.87 Third, especiallyfor smaller countries in which financial conditions are lessstable, regulatory authorities are much more likely to provideclose supervision of the affairs of banks than the affairs of thegreat mass of trading businesses in the country.88

Taken together, those conditions all work together