🛡️ Monitoring a merchant portfolio

An acquirer's exposure, and why it is not the same as volume

An acquirer's exposure is the amount it will have to repay issuers if a merchant in its portfolio goes out of business. It covers both services sold but not yet delivered and sales already delivered that can still be disputed. The acquirer is the institution that signs up the merchant, accepts card payments on its behalf, and pays the funds out to it. The cardholder is the person who holds the card. Most acquirer losses come from honest merchants that took payment for a service they will never deliver; fewer come from fraudulent merchants. The day the business stops, cardholders dispute their payments. Issuers claw the funds back from the acquirer, which pays with no solvent debtor to pursue. Measuring this exposure comes before any collateral policy and any pricing negotiation.

Exposure breaks down into two pools with different maturities. The first is the balance of sales already paid for where the customer is still owed something. An airline ticket for next month, an annual subscription paid on day one, and a concert ticket sold nine months ahead all belong to it. The second is the dispute tail that stays open on sales that were in fact delivered. Cardholders keep the right to dispute long after receiving their order. Together, the two pools give the amount the acquirer would have to pay from its own funds if the merchant went out of business. That amount is recalculated every month.

Gross exposure at time t
   services sold and not yet delivered
 + delivered sales still within their dispute window
 + refunds promised to customers and not yet made

Net exposure
   gross exposure
 - collateral available to draw on (reserve, deposit, personal guarantee, bank guarantee)
 - share of the order book an industry protection scheme will refund to customers

The only figure to track is the second. The first is how you build it.

The dispute window is the period during which a cardholder can still dispute a payment with the issuer. In the ordinary case, it runs 120 days from the transaction. For dispute reasons covering a service not provided, it runs from the expected delivery or performance date instead. The absolute limit remains 540 days after the transaction (Visa Rules and Mastercard Rules). A cruise line that sells a December sailing in January therefore stays exposed until the following spring, 15 months after collecting the fare. An old sale thus remains disputable on these grounds when performance is deferred, because the clock starts on the expected performance date, not the payment date.

The prudential framework for payment institutions does not measure this exposure. A payment institution licensed for acquiring in the EU must hold initial capital of €125,000. Its own funds are then calculated using one of the three methods in Article 9 of Directive (EU) 2015/2366. Method A is based on the previous year's overhead, Method B on the volume of payments executed, and Method C on an operating income indicator. None of the three accounts for what the portfolio's merchants have sold but not yet delivered. The regulatory floor is therefore unrelated to the balance the acquirer carries. An acquirer that is also a credit institution captures the same exposure in its credit risk framework, which makes it visible but sets no limit on it.

Underwriting, or deciding how much you will lose

Underwriting is the review through which an acquirer decides whether to take on a merchant and on what terms. It estimates what the acquirer would lose if the merchant ceased to exist six months from now, which is a different question from whether the merchant is honest. Identity and beneficial ownership checks meet a separate obligation, anti-money laundering, and say nothing about the delivery gap. The two reviews should not be conflated. Merchant risk is assessed with different documents and through different questions.

The merchant category code (MCC) identifies a merchant's line of business in card systems. These codes are standardized in ISO 18245. The code assigned determines the applicable interchange rates, the network programs the merchant falls under, and its internal risk rating. A travel seller registered under a generic retail code pays less and escapes the checks designed for its sector. Misrepresenting the line of business is among the grounds for immediate termination in acquiring agreements. The code on file is still only the merchant's own declaration. Verification means checking what the website actually sells, page by page, including sections the application form never anticipated.

The term “high-risk sector” covers two different ideas: brand risk and delivery risk. Brand risk arises when the sale itself exposes the network, as with gambling, regulated products, or prohibited content. Visa and Mastercard require these categories to register with the network in advance and subject them to dedicated monitoring, which includes Mastercard's Business Risk Assessment and Mitigation program. Delivery risk arises solely from the gap between payment and performance, and it hits perfectly lawful sectors. The two risks call for different treatment. Registering with the network does nothing to reduce the balance sold but not delivered. A gym and a tour operator raise no brand issue at all, yet they carry some of the largest undelivered balances in the portfolio.

• The gap between payment and delivery, as a distribution rather than an average. A 30-day average can hide a fifth of sales performed nine months out, and that fifth carries all the risk.
• The share of sales collected in advance in total revenue, with the declared seasonality and the trend over the last three fiscal years.
• The cancellation and refund policy, which sets the volume of refunds the acquirer will have to process once the business has stopped.
• Whether an industry protection scheme exists that will refund customers in place of the network, with its exact scope and exclusions.
• The counterparty's financial strength: filed annual accounts, equity, and dependence on a single supplier or a single main client.
• The acceptance history: previous contracts, reasons for termination, and any listing in the networks' terminated merchant databases.

Terminated merchant databases are lists kept by the networks, where acquirers record merchants whose acquiring agreement was terminated for cause. MATCH is Mastercard's, and a listing stays visible there for five years. Visa runs an equivalent screening service. Three reasons for listing matter directly here: excessive chargebacks, excessive fraud, and going out of business with amounts left unpaid. A listing covers both the company and its principals, so setting up a new legal entity does not clear it. The check happens during the review, before signing. A listing found at that stage leads the acquirer to decline the application outright.

Collateral required at onboarding is sized on the expected undelivered balance, not on expected volume. A merchant with no delivery gap needs no reserve, and its revenue plays no part in the calculation. A merchant that sells 12 months ahead needs collateral from its very first transaction, before it has built up any dispute history. The common practice of requiring collateral only from merchants that have already deteriorated therefore leaves sound sales models with a structural delivery gap uncovered.

Accepting a merchant is a decision made at a point in time, and the acquirer revisits it on a schedule set in advance. An annual review is enough for merchants with no undelivered balance, while those that sell ahead call for quarterly reviews. The review uses the same documents as onboarding: filed accounts, ownership, website content, and the distribution of delivery times. Above all, it compares what was declared with what is observed, because a gap between the two is the first sign that a merchant has changed its business without telling anyone. That gap shows up in the transaction data the acquirer already has, not in the impression the business relationship leaves.

Sizing, monitoring, and releasing the reserve

A reserve is the portion of the funds owed to a merchant that the acquirer holds back rather than paying out immediately, to cover future disputes. The holdback is released as the risk expires. The acquirer builds it on its own, without relying on a third party, which sets it apart from security provided by a bank or by a principal. Three forms of holdback coexist in the market: the rolling reserve, a deposit paid at signing, and a full holdback until delivery. They do not cover the same situations.

A rolling reserve holds back a percentage of each payout and returns it after a fixed period. At steady state, its balance levels off at roughly the holdback rate times the volume processed over the rolling period. A 10% holdback released after 180 days thus ties up nearly 5% of the merchant's annual volume. That exceeds the net profit of a low-margin retailer. This calculation should come before any negotiation of the holdback rate, since it puts a number on the cash the merchant gives up for the entire life of the contract.

A security deposit is a sum the merchant pays at signing, which the acquirer holds for the duration of the relationship. It is available from day one, whereas a rolling reserve takes several months to reach its steady-state balance. The deposit amount, however, stays fixed. A deposit sized for a business processing one million a year covers only a small share of the exposure once the same business processes 10 million. Revising it requires either a clause providing for it or a renegotiation, which a relationship that is going well never prompts.

A rolling reserve has a built-in lag between when it builds up and when exposure peaks. It builds at the pace of sales already settled and is released at the pace of the rolling period, while exposure rises at the pace of sales not yet delivered. A ticketing company that sells an entire season in three weeks sees its exposure peak before the matching holdback has even been collected. The coverage ratio therefore hits its low point just as the undelivered balance hits its high point. Acquirers that handle seasonal merchants correct this lag by holding back the full amount of any sale for a distant date until the service is delivered.

net exposure      =  undelivered sales balance
                  +  delivered sales still open to dispute
                  +  announced refunds not yet made

collateral        =  rolling reserve balance
                  +  deposit paid at signing
                  +  callable personal or bank guarantee

ratio             =  collateral / net exposure

  ratio < 1   the acquirer covers the gap from its own funds
  ratio = 1   nominal coverage, before any haircut on collateral
  ratio > 1   overcoverage, to be returned to the merchant or justified

The coverage ratio is collateral held divided by estimated net exposure. It measures the share of exposure the acquirer would not have to absorb with its own funds. The amount held back, the estimated exposure, and their ratio are recalculated every month, merchant by merchant, and presented to the risk committee in that form. A ratio that deteriorates while the dispute rate is flat signals growth in prepaid sales. Raising the holdback at that point comes before any dispute, since what has deteriorated is the undelivered balance, not the loss experience.

The reserve is released on two separate schedules, depending on whether the sale has been delivered. Delivered sales are released at the end of their dispute window, counted from the transaction. Undelivered sales are released only after their performance date plus the same window. The two dates do not coincide. Releasing the reserve six months after the last transaction leaves the acquirer uncovered if the merchant has sold departures for the following year. A release clause that covers both schedules therefore keys off the last service date sold, not the last payment date.

Monitoring with and without network thresholds

Network monitoring works at two levels: the individual merchant, and the acquirer's portfolio as a whole. Acquiring agreements pass the merchant-level thresholds on to merchants. Portfolio thresholds are much lower, because a portfolio by design dilutes individual excesses. An acquirer can therefore breach its own threshold without any of its merchants breaching theirs.

The ratio tracked by Visa's acquirer program adds fraud reports and disputes together and divides them by the number of settled card-not-present transactions. The portfolio ratio is the average of its merchants' ratios, weighted by transaction count. A merchant that accounts for 1% of transactions and runs 20% fraud and disputes therefore adds 20 basis points to the acquirer's overall ratio. The excessive threshold is 70 basis points, and the warning threshold is 50. Both levels also require a minimum number of monthly events, below which the portfolio is not enrolled in the program. Three merchants of that size are enough to use up all the headroom of a portfolio that is healthy everywhere else. An acquirer therefore offboards a merchant long before that merchant nears its own threshold, because the constraint the acquirer faces is the portfolio ratio.

Network thresholds rely on monthly data published with an additional lag, so they flag a situation that has already been in place for several weeks. Internal monitoring runs daily on the acquirer's own data. It tracks metrics that move before disputes do. The table below lists those metrics, what each one measures, and the matching response.

External signals often come before internal ones, because a struggling merchant stops paying its suppliers before it stops delivering to its customers. The clearest is a credit insurer withdrawing cover on the sector or on the business itself, and suppliers know about it before the acquirer does. Missing the legal deadline for filing annual accounts, liens registered by social security agencies, the opening of a pre-insolvency procedure, and removal from a trade register deserve the same attention. Monitoring customer reviews that mention late deliveries is cheap and pays off quickly. Each of these signals can have an explanation unrelated to failure, so a single signal is not grounds for a decision. Two signals that corroborate each other, however, justify an immediate review of the account, without waiting for the next scheduled review.

The merchant risk policy is the internal document that specifies, for each signal, the threshold that triggers it, who decides, and how quickly they must act. It is adopted before the signals it covers appear. Without it, suspending a payout gets debated at the very moment it should be carried out. The debate then pits a salesperson defending their client against a lawyer reading the contract for the first time. A risk committee decides much faster when it is applying a decision it made itself six months earlier. Its review then focuses on whether the threshold has been crossed, not on whether acting is advisable.

The default playbook, step by step

The default playbook is the internal procedure that sets out what to do when a merchant goes out of business, written long before the event it covers. The first decisions are made within hours of the shutdown being confirmed. The acquirer will have to answer for them before a court, the insolvency administrator, and the network, each of which may later review whether they were properly taken. The playbook comes down to six steps. Their order matters as much as their content.

Suspending payouts is the most legally exposed step in the playbook. The acquirer owes the merchant money. It chooses to keep that money to cover disputes that have not yet arisen. The contract is the only basis for the holdback, and the seriousness of the situation adds no further right. An acquiring agreement that specifies neither the trigger, nor the cap, nor the duration of the suspension exposes the acquirer to an adverse judgment, even when its reading of the risk proves correct.

Notifying cardholders is not the acquirer's job, yet the acquirer bears all the consequences. Customers learn of the shutdown from the news or from an email sent by the merchant, then contact their bank within hours. The wave of disputes arrives within a few days. It concentrates on the reason codes for services not received: 13.1 at Visa and the Cardholder Dispute category at Mastercard. Bulk handling prepared in advance absorbs this wave without stretching response times. Those times do stretch when the disputes team is staffed for business as usual.

Coordinating with industry protection schemes avoids both paying customers twice and provisioning twice. An industry fund that refunds a tour operator's customers publishes its procedure, scope, and timetable. From that information, the acquirer works out which share of the order book the fund will handle and which share will come back to it as disputes. The scope should be requested in writing, because provisioning the whole order book on both sides distorts the loss estimate for months. Underestimating the residual loss is riskier still, because a guarantee scope narrower than announced sends the entire balance back to the network.

Once insolvency proceedings open, the case changes in nature. The acquiring agreement becomes an executory contract whose fate depends on local law, and the administrator may want to keep it in force to preserve a business that can be sold. The reserve stops being an accounting line and becomes a subject of litigation in its own right. The acquirer argues that it is setting off related claims; the insolvency representative argues that it is holding money owed to the estate. The outcome of that dispute drives most of the final recovery rate whenever no third-party collateral was taken at onboarding.

Absorbing disputes takes far longer than the liquidation itself. The merchant's acceptance IDs stay open to receive disputes, even though no more sales are possible. The acquirer pays each dispute, charges it against a dwindling reserve, then writes it off. Fighting a dispute through representment requires proof of delivery, which lives in the merchant's systems and becomes inaccessible as soon as they are shut down. The preservation step in the playbook addresses that constraint. A copy of the order database taken on day one still makes it possible to defend cases six months later. Without it, disputes get paid with nothing to submit.

The acquirer's remedies after a failure rank by how much they recover, and that ranking varies little from case to case. The most effective are put in place at underwriting, since collateral not required at signing can no longer be obtained from a failing merchant. Underwriting and the reserve therefore weigh more on the amount recovered than all crisis measures combined. The list below follows that order, from the remedy that can be drawn on fastest to the claim least likely to be recovered.

• The reserve and the deposit can be drawn on immediately and return their face value, depending on how setoff is treated under the applicable law.
• An on-demand bank guarantee is called on simple demand, up to its cap and within its validity period.
• A personal guarantee from the principal has to be argued and negotiated, and it rarely recovers most of the loss, because its amount has almost never been sized to the undelivered balance.
• Credit insurance, where a policy was taken out on the merchant, pays according to its own notification rules and waiting period.
• Filing a claim in the proceedings comes last, as an unsecured creditor, and the recovery rate is usually negligible.