Letter of Credit Discrepancies: Why a Typo Hands the Decision Back to Your Buyer

Summary

A discrepancy is a difference between the documents you presented and the documents the credit told you to present. It is not a finding about your cargo. The goods can be on the water, on specification, on time and fully insured, and the presentation can still be refused because a party's name is spelled two ways across two pieces of paper.

This article sets out why a letter of credit pays against a description rather than against performance, why a refused presentation does not reduce your protection by a little but switches it off and hands the decision to the buyer you bought the instrument to be protected from, why nearly every discrepancy was created on the day the credit arrived rather than on the day the documents were presented, why you are answerable for documents that other organisations write and only they can reissue, why the same field being typed separately into five documents is a data problem rather than a diligence problem, and why the cost that gets recorded — the bank's fee — is the smallest of the three costs a discrepancy actually causes. It is written for the people who live with this: trade finance, documentation, operations and settlement teams, the middle office that has to explain a receivable that has stopped behaving like a bank receivable, and the commercial people who agreed the payment terms without seeing what would later have to be produced under them.

If you want the mechanism that makes the rest follow, start with A discrepancy does not reduce your protection — it transfers the decision. If you already accept that and want to know where the work should move to, start with Almost every discrepancy was created on the day the credit arrived.

The bank is paying against a description of the cargo, not against the cargo

A documentary credit is a bank's promise to pay on the presentation of a specified set of documents. It is not a promise to pay for goods, and the two only coincide when the paperwork happens to be right.

This is the property everything else in the subject follows from, and it is the one that feels wrong the first time a presentation is refused. The instinct is to explain the trade: the cargo loaded, the quality certificate is in the file, the buyer has the goods, nobody disputes that the contract was performed. All of that is true and none of it is responsive. The bank never undertook to assess whether you performed. It undertook to pay if a defined set of documents was presented and those documents, read on their face and against each other, said what the credit said they would say.

Two consequences are worth stating on their own, because teams that have not internalised them lose time in the same place every cycle.

The first is that "the goods are fine" is not an argument, and offering it signals that you have misread what is being examined. The examiner is not empowered to look at the shipment. A discrepancy is a statement about a text.

The second is that the credit, not your sales contract, defines what you have to produce. These are two different documents drafted by different people at different times, and they can disagree. Your contract may allow a tolerance the credit does not. Your contract may name the goods one way and the credit another. Where they differ, the contract governs what you owe your buyer, and the credit governs what the bank will pay against — and the gap between them is a place where a perfectly performed trade produces a perfectly valid refusal.

A discrepancy does not reduce your protection — it transfers the decision

When a presentation is discrepant, the bank's independent undertaking is replaced by your buyer's discretion. The party who then decides whether you are paid is the party whose credit risk the instrument existed to remove.

This is the mechanism most often described as "a delay" and it is not a delay. A conforming presentation obliges the bank; the buyer's opinion of it is irrelevant, and that irrelevance is the entire product you bought. A discrepant presentation does not oblige anyone. What normally happens next is that the bank asks the applicant — the buyer — whether they will waive the discrepancy. A buyer who still wants the cargo has every reason to say yes, and that is precisely why the mechanism is misfiled as an administrative hiccup rather than as what it is.

Look at what the seller now holds. Not a bank undertaking with a paperwork issue attached. An unsecured claim against a counterparty, plus a request that they choose to restore the security you had until an hour ago. Everything that made the instrument worth its cost — that payment did not depend on the buyer's willingness, their solvency, their view of the market, or the state of the commercial relationship — is gone for as long as the waiver is outstanding, and it is gone completely rather than partially.

The consequence that follows is the one to take away, and it is uncomfortable. The buyer's incentive to waive is not constant. It depends on what the market has done since the trade was agreed. A buyer who still wants the cargo waives a spelling error without reading past it. A buyer who has watched the price fall since fixing, who is now holding an expensive cargo and looking for a way not to hold it, has been handed a costless option — and that option was written by your documentation team without anyone deciding to write it. This is why discrepancy rates that look tolerable for years become expensive in a single quarter: the failure correlates with exactly the conditions the instrument was bought to survive. The protection is least reliable at the moment it is most needed, and nothing in the process reports that, because in ordinary markets the waivers come back approved.

It also explains a pattern that otherwise looks like bad luck. Documentation problems on trades that never became contentious are resolved and forgotten. Documentation problems on trades that did become contentious are remembered as disputes about performance. They were the same defect. Only the market was different.

Almost every discrepancy was created on the day the credit arrived

The document set is checked when it is presented, which is the last moment at which anything can be done about it. The credit is checkable on the day it arrives, which is the first moment and the only cheap one.

A credit is drafted by the issuing bank from the applicant's instructions. Neither of them is the party who will have to produce what it demands. So a credit can perfectly well call for a document that does not exist in your trade, for a description of the goods that no carrier will put on a transport document, for a certificate that the issuing body will not date the way the credit assumes, for a shipment window and an expiry that leave no room for documents to travel, or for terms that contradict the sale contract they were supposed to implement. None of that is unusual, and none of it is anyone's mistake in a way that helps you. It is the ordinary result of a document being written by people who will not have to satisfy it.

Reading the credit against what you will actually be able to produce is, on any reasonable accounting, the highest-leverage hour of work in the entire chain. It is also routinely the one nobody owns. The credit arrives, it is filed, and it is read closely for the first time by the person assembling the presentation — after the ship has sailed, after the transport documents have been issued, and after every party who could have helped has moved on to other work.

The reason this matters so much is that the cost of fixing the same problem is not merely higher later; it changes shape. An amendment requested before shipment is a request to a buyer who wants their cargo. An amendment requested after presentation is a request to a buyer who has just acquired the right to say no. Early, the buyer's interest and yours point the same way — they want the goods, the amendment is a formality, and their bank will process it. Late, you are asking the same person for the same words at the precise moment they have discovered they do not have to give them.

The practical form of this claim is narrow and testable. If your organisation cannot name the person who read the last credit within a day of its arrival, and cannot say what they checked, then your discrepancy rate is not a documentation outcome. It is a scheduling one.

You are answerable for documents you did not write and cannot reissue

Of the documents in a typical presentation, you author very few. The rest are issued by carriers, surveyors, insurers, chambers of commerce and other third parties — and each correction runs on that party's clock, not yours.

This is the structural reason the discipline that works elsewhere in the back office does not transfer here. A team can standardise its own invoice, its own file naming, its own checklists, its own approvals. It cannot standardise a bill of lading. The carrier issues that, in their format, from data given to them at booking, and altering it later requires their agreement, their process, and in the ordinary case the physical return of the original document — which may by then be in a courier bag between two countries, or already lodged with a bank.

The point is not that third parties are slow. It is that the correction path for a discrepancy passes through organisations that have no interest in your presentation deadline and no obligation to it. A certificate of origin may need to be reissued by a body that works to its own submission cycle. A quality certificate may require the surveyor to re-open a file, and if the disagreement is about what should have been measured, no reissue will help. An insurance certificate can usually be corrected quickly; a transport document usually cannot. These are not equivalent problems and treating them as one queue of "documents to fix" hides the only distinction that predicts whether the fix arrives in time.

What follows is a different allocation of effort than most operations run. The documents you control are the ones you can afford to check last, because you can fix them. The documents you do not control have to be got right at the moment they are being created — at booking, at the survey, at the application for the certificate — because after that you are a supplicant. The instructions you give a carrier at booking are, functionally, part of your document check. They are almost never treated that way.

The same field is typed five times, which is why care does not fix this

A presentation has to be internally consistent across documents produced by different organisations from different copies of the same data. Consistency of that kind is a property of where the data came from, not of how carefully anybody read it.

Consider what has to agree. The description of the goods, the quantity, the marks and numbers, the ports, the dates, the party names and their spelling, the credit reference, the amount, the currency, the shipping terms — restated across an invoice, a transport document, a packing list, one or more certificates and an insurance document. Some of those fields have to match the credit exactly, some have to be consistent with the other documents without being identical, and the rules about which is which are not intuitive to anyone who does not do this work full time.

Now consider where each document's version of those fields came from. The invoice, from your own system. The transport document, from a booking made by email or by phone, possibly weeks earlier, possibly by someone in a different office. The certificate, from an application form typed from a copy of the booking. The insurance document, from a declaration typed from a copy of the invoice. The set is assembled from copies of copies, and the number of places where a difference can enter grows with the number of documents multiplied by the number of fields — while the tolerance for difference is nil.

This is why the standard remedy fails. Checking harder finds more errors, and it does find them; it does not stop them being created, and the checking happens at the end, where the fixing is expensive. The errors are not the product of insufficient care. They are the product of the same fact being entered independently in five places. Careful people do not stop making these mistakes. Processes in which a field is entered once and reused do.

There is a corollary worth stating because it changes what "getting better at this" means. A team that has done thousands of presentations is faster at spotting a discrepancy and no less likely to generate one, because generating one is not a skill deficit. Experience improves detection. Only the data path improves creation.

The cost you record is the smallest of the three you pay

A discrepancy produces a bank fee, a financing delay and a transfer of commercial leverage. The fee is the only one that arrives with an invoice, and it is the one people manage.

Banks charge for handling a discrepant presentation, and the existence of a standard charge for it tells you something on its own: the case is routine, not exceptional. That charge is visible, small enough to be absorbed, and coded to an account that somebody reviews once a year. It is the reason a discrepancy is easy to file as a nuisance with a price on it.

The second cost is time. Payment that was due against a conforming presentation is now due when a waiver comes back, and the interval is spent financing a cargo you have shipped. That cost is real, it scales with the value of the shipment rather than with the size of the error, and it is usually invisible because it lands as an interest cost in a different department, attributed to nothing in particular.

The third cost is the one described above: for the duration, you are unsecured, and the buyer is holding an option. That cost is nothing at all whenever the buyer wanted the cargo anyway, and very large when they did not — which is the worst possible distribution for a risk being managed on the basis of average experience. A control whose failures are cheap almost every time and catastrophic occasionally will always look adequate right up until it is not, and the fee-based view of discrepancies is exactly that kind of measurement.

The practical implication is about reporting rather than about process. An operation that tracks discrepancy fees is tracking the one component that does not vary with what is at stake. An operation that tracks how many presentations were discrepant, on what value, and how long each one stood unwaived, is looking at the exposure. Those are two different reports and most organisations produce only the first.

Who issues what, and what a correction actually requires

Two things are visible in this table that are invisible in an exception list. The first is that the column headed "who issues it" is mostly not you, which is the whole subject in one column. The second is the last column: it separates the corrections that are a task from the corrections that are a negotiation with someone who owes you nothing, and it is the column that predicts whether a fix lands before the credit expires. Most document checklists do not carry it, because a checklist is organised by document rather than by who controls it.

The same discrepancy at three moments

The bottom two rows carry most of the meaning. Discrepancies caught early leave no trace, so an organisation that has got good at this has no evidence that it did — while the ones that were not caught early are remembered, and misremembered, as commercial disputes. And the ownership row shows the defect: the moment where the work is cheapest is the moment with no owner.

Three questions your own file should be able to answer

These are not audit questions. They ask whether the facts are stored or reconstructed.

One: for the last credit you received, who read it against what you would have to produce, and on what date relative to its arrival? If the answer is that it was read properly for the first time while the presentation was being assembled, that is the finding, and it is upstream of every discrepancy you had that year.

Two: for the presentations refused in the last period, can you say which document was discrepant, who issued that document, and how many days passed before a waiver came back? Not how much you were charged. If the only field you can retrieve is the fee, you are measuring the smallest of the three costs and none of the exposure.

Three: for the credits currently outstanding, can you produce — as a query, not as a piece of work — the list of latest shipment dates, expiry dates and presentation deadlines, sorted by which comes first? If building that list takes an afternoon, then between afternoons it does not exist, and the dates that matter most are the ones nobody is looking at.

Questions people ask about this

Why was our presentation refused when the goods arrived exactly as contracted?

Because the examination is of documents, not of performance. A credit obliges a bank to pay against a specified set of documents that agree with the credit and with each other on their face. Whether the cargo was correct is a question between you and your buyer under the sale contract; it is not what the bank undertook to assess. This is why explaining the shipment does not move a refusal, and why a perfectly performed trade can produce a perfectly valid discrepancy.

Our buyer will just waive it. Why does this matter?

Because the waiver is a decision they get to make, and until they make it you are unsecured. The whole point of the instrument was that payment did not depend on your buyer's willingness. A discrepancy suspends that, and it does so completely rather than partially. A buyer who still wants the cargo has every reason to waive — which means ordinary practice teaches you that discrepancies are harmless, and the lesson holds until a market move gives a buyer a reason to look for a way out. The risk is not that waivers get refused often. It is that whether one is refused depends on the same conditions that decide whether you needed the protection.

Who should be checking the credit when it arrives, and what are they looking for?

Somebody who knows what your operation can actually produce — which is usually documentation or operations, not the commercial desk that agreed the terms. They are checking three things: that every document the credit demands can exist for this trade and be obtained in the form demanded; that the descriptions, quantities, tolerances and party names in the credit match what your own systems and your carrier will emit; and that the shipment window, expiry and presentation period leave enough room for documents to be produced, corrected and delivered. Anything that fails those tests is an amendment request, and the day the credit arrives is the cheapest day it will ever be.

What is the difference between a discrepancy and a dispute?

A discrepancy is a mismatch between the documents and the credit, decided by a bank reading paper. A dispute is a disagreement about the underlying trade, decided between you and your counterparty under the sale contract. They are separate systems and a discrepancy can arise with no dispute in sight. They become entangled when a buyer who has a commercial motive uses a documentary defect to act on it, which is the case worth designing against even though it is not the common one.

Can we stop using letters of credit and avoid this?

That trades one risk for another and the choice is commercial rather than operational. Open account removes the document examination and removes the bank undertaking with it. What is worth separating is the two reasons an organisation ends up on documentary terms: because the counterparty or the market genuinely requires it, or because it was on the last contract. The first is a decision; the second is inertia carrying a real administrative cost, and it is worth knowing which one you are paying for.

Is a discrepancy fee the real cost of getting this wrong?

No, and treating it that way is the most common measurement error in this area. The fee is fixed and visible. The financing cost of the delay scales with the value of the shipment. The transfer of leverage to the buyer is worth nothing in most cases and a great deal in a few. Only the first arrives with an invoice, which is why it is the one that gets managed.

Why do the same discrepancies keep coming back after we have trained everyone?

Because most of them are not knowledge failures. They come from the same fact being typed independently into documents produced by several organisations working from different copies of the data. Training improves detection at the end of the process; it does not change how many differences are created at the start. The number of opportunities for a mismatch is set by how many times a field is re-entered, and that number does not fall because people are trying harder.

We only do a handful of these. Is it worth building a process?

Low volume makes this worse rather than better in one specific way: nobody has a routine, the last credit was long enough ago that its lessons are gone, and the person who handled it may have changed roles. The proportionate answer is not a system. It is two habits — read the credit against your own capability on the day it arrives, and give the carrier, the surveyor and the certifying body the exact wording the credit requires at the moment they are creating the document rather than after. Both are free. Most of the benefit described in this article comes from those two.

Where should the deadlines live?

Somewhere that can be queried rather than assembled, and somewhere that is not one person's calendar. The dates that decide this — latest shipment, expiry, the period allowed for presentation — are known the day the credit arrives and are almost never in the same place as the trade they belong to. That gap is where an expiry that everybody knew about arrives without warning.

Where this lands in a trading system

Nothing above is an argument for buying software. Two of the three habits it ends on cost nothing and involve talking to people earlier. But the conclusions do map onto system capabilities, and it is worth being specific about which, and about where the mapping stops.

The first conclusion is that a documentary credit is part of the trade, not an administrative appendix to it — its dates, its required documents and its terms have to be visible next to the shipment they govern and the receivable they will settle. That is what a CTRM/ETRM system providing complete trading lifecycle management for commodity traders, integrating physical trade, financial hedging, risk control, and settlement in one platform is for; Time Dynamics' Fusion is one, and the capabilities that bear on this article are complete physical trade lifecycle management and execution, end-to-end logistics coordination and execution management, and automated financial settlement and payment processing.

The second conclusion is that the fields which have to agree across a presentation originate outside any one system, and that the copies which disagree are sitting in spreadsheets, in booking correspondence and in other parties' portals. That is a different problem, and it is the one X-Ray addresses: a non-invasive data processing and analysis platform, whose collection toolkit XDK performs non-invasive automated data collection from databases, Excel files, and web interfaces, and which is built to collect data without disrupting existing systems or workflows. The last property is the one that matters here, because the spreadsheet your documentation team runs on is working, and no plan that begins by replacing it will survive its first busy week.

The third conclusion is the deadline question, and this is the one place where the mapping is unusually direct. X-Eagle is described as real-time risk alerts and monitoring for trade finance field warnings — a watched field, a condition, and a warning raised while there is still time to act. Latest shipment date, expiry and presentation period are trade finance fields, and an approaching one is an ordinary shape for a warning rather than an exotic one. Whether it is pointed at your own credit register is a configuration question rather than something a platform decides for you.

The fourth is the reporting point: how many presentations were discrepant, on what value, which document was at fault, who issued it, and how long each stood unwaived. Those are re-cuts of the same underlying items, and they have to be producible often enough to act on rather than assembled once a year. X-Sheet does one-click generation of personalized reports with Excel-like interface.

And here is where it stops. None of this is a trade finance or document preparation system. It does not draft your documents, it does not examine a presentation, it does not tell you whether a discrepancy is well founded, and it has no view on whether a bank was right to refuse — that is a judgement made by an examiner reading a text, under rules two banks and your buyer agreed to. It will not obtain an amendment, it will not negotiate a waiver, and it will not make a carrier reissue anything. It does not talk to your bank. What a system can settle is narrower and duller than the problem in this article: whether the credit's terms and dates are stored next to the trade rather than in a folder, whether the fields your documents must share came from one place or five, and whether anybody has to build a spreadsheet to find out which deadline is next. That is a useful share of the distance, and it is not the whole of it.

See the products → · Talk to us about trade documentation data →

A note on the evidence in this article

This article contains no statistics, and that is a decision rather than an omission. There are figures in circulation for how often documentary presentations are refused on first examination, and we are not in a position to publish one we can source properly. A number we cannot stand behind would make this argument look stronger and would make it worth less.

So the argument is structural instead. Every claim here is made from the mechanism — what the instrument is, who writes which document, who can change it, and what changes hands when a presentation is refused — and each is stated in a form you can test against your own last twelve credits rather than take on our authority. The three questions above are the intended way to use it. If your own file contradicts what is written here, your file is the better evidence.