Reserves, limits, and payout policy

Estimate the open obligation

Start with the exposure path: unsettled transactions, undelivered sales, pending refunds, and potential returns. Avoid adding overlapping measures that represent the same obligation. Define the event that closes each exposure.

For a simple advance-sale merchant, $200,000 in undelivered sales is a useful starting point. It is not automatically the final loss. Some goods may still be delivered, funds may remain available, and recoveries may occur. Use a gross exposure measure and a separate loss estimate. Keeping both avoids hiding a large obligation behind an optimistic recovery assumption.

Open exposure is a stock measured at a point in time. Payment volume is a flow measured over a period. Multiplying a month’s volume by an arbitrary percentage does not automatically produce the amount at risk. Start with the obligations still open: undelivered services, possible refunds under the relevant assumptions, disputed amounts, and other commitments. Then identify which obligations overlap so the model does not count the same purchase several times.

This distinction becomes visible when a business stops selling. New volume falls to zero, yet old fulfillment and refund obligations remain. A monitoring system that looks only at current processing volume may conclude that risk has disappeared just as the platform loses its easiest source of future offsets.

Inside the mechanism. Open obligation includes value whose related customer promise or recovery risk remains unresolved under the model. Its horizon can follow fulfillment, return, dispute, or repayment timing. State which obligations overlap so they are not added twice. Track the population by cohort and remaining maturity. A current balance is a snapshot; projected exposure also depends on new activity, payouts, and the rate at which old obligations close.

A concrete example. The exposure report measures promises still open today. Current payment volume is a flow and cannot replace the stock of undelivered obligations. The case has $780,000 of exposure. Its stated one-year PD and LGD imply $32,175.00 of expected loss, while the cover analysis leaves $543,000.00 of stress exposure. Monthly cash coverage is 0.86×. These are separate measures: one describes an average under probability assumptions, one describes available cover, and one describes a period’s funding capacity.

When the assumption fails. Processing volume falls to zero after the merchant stops selling, while refunds remain due. Reconstruct remaining obligations by purchase cohort and remove overlap among loss estimates. The following worked sequence shows the reference condition, a stress condition, and a response condition with explicit synthetic data. These are comparative assumptions, not measured causal effects.

Follow a worked case3 conditions · 36 figures

The exposure report measures promises still open today. Current payment volume is a flow and cannot replace the stock of undelivered obligations.

Open fulfillment obligation: reference condition figure preview

01Reference conditionOpen fulfillment obligation

Open fulfillment obligation: failure and stress figure preview

02Failure and stressOpen fulfillment obligation

Open fulfillment obligation: controlled response figure preview

03Controlled responseOpen fulfillment obligation

Estimate the open obligation — the flow

IdentifyList open customer obligations
DeduplicateRemove overlapping exposure records
CloseUse evidence of fulfillment or resolution

Follow the sequence. Use evidence of fulfillment or resolution. Chapter sources · Open image

Estimate the open obligation Estimate the open obligation — the distinction These concepts answer different questions. Read each definition in the context of the section. Gross exposure Amount still subject to the loss path Net loss estimate Exposure adjusted for scenario and recovery — These concepts answer different questions. Read each definition in the context of the section. Chapter sources · Open image

Estimate the open obligation Advance-sale exposure Fictional teaching record. Do not add the subset twice. Advance-sale exposure Illustrative data; not a real customer record or a prescribed policy. Undelivered 200000 USD Open customer promises Pending refunds 20000 USD Subset already included Gross total 200000 USD Do not add the subset twice The same sale must not inflate exposure twice — Fictional teaching record. Do not add the subset twice. Chapter sources · Open image

Estimate the open obligation Estimate the open obligation — control and failure modes The same sale must not inflate exposure twice. The branches show why alternative designs fail. Control design Deduplicate overlapping obligations. The same sale must not inflate exposure twice. Failure mode 1 Add every report total together. Reports can overlap. avoid Failure mode 2 Call exposure a certain loss. Some obligations may be fulfilled. avoid Failure mode 3 Subtract hoped-for recovery as cash. An estimate is not available cover. avoid — The same sale must not inflate exposure twice. The branches show why alternative designs fail. Chapter sources · Open image

Understand reserve structures

A rolling reserve retains part of eligible volume for a defined period. A fixed reserve targets an amount. Other structures can combine conditions. The agreement determines ownership, permitted use, release, and any limits.

Model the reserve as a ledger with accruals, uses, and releases. A percentage without a release schedule cannot predict available protection. If 10 percent of $50,000 eligible sales is retained, the new accrual is $5,000 under this example. That does not establish the total reserve balance because earlier releases and uses may also occur. All percentages here are teaching assumptions, not market standards.

Inside the mechanism. A fixed reserve, rolling reserve, and delayed availability structure have different cash effects. State the funding base, release rule, eligible use, and ownership of the funds. A percentage without a time horizon does not explain how much cover exists when a loss arrives. Model reserve releases alongside obligation runoff. The same dollar should not be counted twice as both unrestricted liquidity and separately available loss cover.

A concrete example. The ledger shows a reserve balance, but its usefulness depends on access, currency, legal rights, and competing claims. A reassuring balance may be unavailable when needed. The case has $515,000 of exposure. Its stated one-year PD and LGD imply $12,746.25 of expected loss, while the cover analysis leaves $292,000.00 of stress exposure. Monthly cash coverage is 1.33×. These are separate measures: one describes an average under probability assumptions, one describes available cover, and one describes a period’s funding capacity.

When the assumption fails. The reserve sits with the same counterparty that is under stress. State eligibility assumptions and report gross exposure, recorded reserve, usable cover, and residual exposure. The following worked sequence shows the reference condition, a stress condition, and a response condition with explicit synthetic data. These are comparative assumptions, not measured causal effects.

Follow a worked case3 conditions · 36 figures

The ledger shows a reserve balance, but its usefulness depends on access, currency, legal rights, and competing claims. A reassuring balance may be unavailable when needed.

Recorded reserve versus available cover: reference condition figure preview

01Reference conditionRecorded reserve versus available cover

Recorded reserve versus available cover: failure and stress figure preview

02Failure and stressRecorded reserve versus available cover

Recorded reserve versus available cover: controlled response figure preview

03Controlled responseRecorded reserve versus available cover

Understand reserve structures Understand reserve structures — the flow Follow the sequence. Return funds when conditions are met. Accrue Retain eligible funds under terms Use Apply only to permitted obligations Release Return funds when conditions are met — Follow the sequence. Return funds when conditions are met. Chapter sources · Open image

Understand reserve structures Understand reserve structures — the distinction These concepts answer different questions. Read each definition in the context of the section. Reserve accrual New amount retained this period Reserve balance Opening plus accruals less uses and releases — These concepts answer different questions. Read each definition in the context of the section. Chapter sources · Open image

Understand reserve structures Reserve movement Fictional teaching record. Before other balance movements. Reserve movement Illustrative data; not a real customer record or a prescribed policy. Eligible sales 50000 USD Defined period Retention 10 percent Illustrative agreed rate New accrual 5000 USD Before other balance movements A rate alone cannot describe the balance — Fictional teaching record. Before other balance movements. Chapter sources · Open image

Understand reserve structures Understand reserve structures — control and failure modes A rate alone cannot describe the balance. The branches show why alternative designs fail. Control design Track reserve movements in a ledger. A rate alone cannot describe the balance. Failure mode 1 Assume all retained funds are unrestricted. Terms can limit use. avoid Failure mode 2 Call 10 percent a universal standard. The example is not a prescribed rate. avoid Failure mode 3 Ignore releases. Protection can decline over time. avoid — A rate alone cannot describe the balance. The branches show why alternative designs fail. Chapter sources · Open image

Use limits to bound exposure

A volume limit, transaction limit, payout limit, and outstanding-exposure limit constrain different quantities. A daily sales cap may not control months of undelivered obligations. Choose the limit that matches the loss mechanism.

Enforce limits atomically when concurrent requests can consume the same capacity. If two workers each see $1,000 remaining and each release $800, a read-then-write design can exceed the limit. Use a transactional reservation and release unused capacity after resolution. Monitor both successful enforcement and false blocks caused by stale reservations.

Inside the mechanism. A limit needs a subject, unit, time window, and atomic consumption rule. Per-transaction, daily value, outstanding exposure, and destination limits constrain different risks. Concurrent requests must share the same authoritative capacity state. Include expiry, release, and reconciliation for reservations that never execute. A dashboard threshold that alerts after the limit is exceeded is a monitoring signal; it is not the same as a pre-effect capacity control.

A concrete example. A processing limit controls one source of new exposure, while existing obligations continue to mature. The limit needs the right aggregation key and observation period. The case has $1,120,000 of exposure. Its stated one-year PD and LGD imply $24,640.00 of expected loss, while the cover analysis leaves $750,000.00 of stress exposure. Monthly cash coverage is 1.33×. These are separate measures: one describes an average under probability assumptions, one describes available cover, and one describes a period’s funding capacity.

When the assumption fails. Several related accounts each receive the full limit despite one shared loss driver. Aggregate relevant exposure and enforce reservations before new commitments are accepted. The following worked sequence shows the reference condition, a stress condition, and a response condition with explicit synthetic data. These are comparative assumptions, not measured causal effects.

Follow a worked case3 conditions · 36 figures

A processing limit controls one source of new exposure, while existing obligations continue to mature. The limit needs the right aggregation key and observation period.

Limits on a changing exposure stock: reference condition figure preview

01Reference conditionLimits on a changing exposure stock

Limits on a changing exposure stock: failure and stress figure preview

02Failure and stressLimits on a changing exposure stock

Limits on a changing exposure stock: controlled response figure preview

03Controlled responseLimits on a changing exposure stock

Use limits to bound exposure Use limits to bound exposure — the flow Follow the sequence. Return unused capacity with evidence. Measure Choose the exposure quantity Reserve Consume capacity atomically Release Return unused capacity with evidence — Follow the sequence. Return unused capacity with evidence. Chapter sources · Open image

Use limits to bound exposure Use limits to bound exposure — the distinction These concepts answer different questions. Read each definition in the context of the section. Daily volume limit Caps a period flow Outstanding limit Caps unresolved accumulated exposure — These concepts answer different questions. Read each definition in the context of the section. Chapter sources · Open image

Use limits to bound exposure Concurrent release example Fictional teaching record. Cannot both fit. Concurrent release example Illustrative data; not a real customer record or a prescribed policy. Capacity 1000 USD Available before requests Request A 800 USD Needs a reservation Request B 800 USD Cannot both fit Concurrent decisions must see one consistent limit — Fictional teaching record. Cannot both fit. Chapter sources · Open image

Use limits to bound exposure Use limits to bound exposure — control and failure modes Concurrent decisions must see one consistent limit. The branches show why alternative designs fail. Control design Reserve shared capacity atomically. Concurrent decisions must see one consistent limit. Failure mode 1 Check then update without locking. Both requests can pass against the same capacity. avoid Failure mode 2 Use a daily cap for all future delivery risk. Exposure can accumulate across days. avoid Failure mode 3 Never release expired reservations. Legitimate activity can remain blocked. avoid — Concurrent decisions must see one consistent limit. The branches show why alternative designs fail. Chapter sources · Open image

Match payout timing to customer promises

Payout timing shifts liquidity between the platform and merchant. Faster payouts can improve the merchant experience while increasing platform exposure. Delays can reduce exposure but may harm fulfillment if the merchant needs cash to buy goods.

Analyze both effects. A payout policy that starves a viable merchant can create the failure it was meant to prevent. Use evidence about delivery, funds availability, and the business model. Make the schedule and review process understandable. Any restriction must follow the agreement and applicable duties, with an accountable path for correction.

Inside the mechanism. Payout timing is part of the product promise and the risk structure. Delaying availability can reduce some funding exposure but may also reduce a merchant’s ability to fulfill. Compare the planned delay with the actual uncertainty and the business cash cycle. Explain the customer-visible states and expected next evidence. A hidden delay introduced by a broken control is different from a clearly stated and correctly implemented payout term.

A concrete example. A merchant’s payout schedule interacts with delivery, refunds, and available resources. Delaying funds can reduce one exposure while creating operating pressure for a sound business. The case has $455,000 of exposure. Its stated one-year PD and LGD imply $8,008.00 of expected loss, while the cover analysis leaves $299,000.00 of stress exposure. Monthly cash coverage is 1.12×. These are separate measures: one describes an average under probability assumptions, one describes available cover, and one describes a period’s funding capacity.

When the assumption fails. A blanket delay is applied without measuring fulfillment or cash needs. Compare the loss path, usable cover, timing, and customer impact under stated terms. The following worked sequence shows the reference condition, a stress condition, and a response condition with explicit synthetic data. These are comparative assumptions, not measured causal effects.

Follow a worked case3 conditions · 36 figures

A merchant’s payout schedule interacts with delivery, refunds, and available resources. Delaying funds can reduce one exposure while creating operating pressure for a sound business.

Payout timing and customer promises: reference condition figure preview

01Reference conditionPayout timing and customer promises

Payout timing and customer promises: failure and stress figure preview

02Failure and stressPayout timing and customer promises

Payout timing and customer promises: controlled response figure preview

03Controlled responsePayout timing and customer promises

Match payout timing to customer promises Match payout timing to customer promises — the flow Follow the sequence. Balance terms controls and review. Exposure Estimate what early release creates Merchant cash Assess fulfillment funding needs Policy Balance terms controls and review — Follow the sequence. Balance terms controls and review. Chapter sources · Open image

Match payout timing to customer promises Match payout timing to customer promises — the distinction These concepts answer different questions. Read each definition in the context of the section. Faster payout Improves merchant liquidity Longer hold May reduce platform exposure but strain delivery — These concepts answer different questions. Read each definition in the context of the section. Chapter sources · Open image

Match payout timing to customer promises Payout tradeoff Fictional teaching record. Illustrative controlled compromise. Payout tradeoff Illustrative data; not a real customer record or a prescribed policy. Supplier due today Merchant funding need Customer delivery next week Open promise Policy staged release Illustrative controlled compromise A cash restriction can change merchant behavior — Fictional teaching record. Illustrative controlled compromise. Chapter sources · Open image

Match payout timing to customer promises Match payout timing to customer promises — control and failure modes A cash restriction can change merchant behavior. The branches show why alternative designs fail. Control design Assess the effect on fulfillment as well as loss. A cash restriction can change merchant behavior. Failure mode 1 Delay every payout indefinitely. That can create avoidable harm. avoid Failure mode 2 Assume fast payout has no exposure. The platform may finance unresolved obligations. avoid Failure mode 3 Hide the review process. Merchants need a correction route. avoid — A cash restriction can change merchant behavior. The branches show why alternative designs fail. Chapter sources · Open image

Stress available cover

Reserve adequacy depends on correlated loss, timing, legal availability, currency, and counterparty access. A reserve held by a failed or frozen partner may not be usable when needed. A reserve in another currency can change value.

Build a stress table that shows gross obligations, eligible cover, expected recoveries, and the remaining gap separately. Test the same reserve against one scenario at a time without allocating it twice. Assign a response to a breach: new exposure limits, funding action, merchant review, or another approved measure. A stress report without an action owner is only a description.

A reserve balance is useful cover only to the extent it is legally and operationally available for the relevant obligation. It may be denominated in another currency, subject to another claim, or held with a counterparty that is also stressed. State the eligibility rules before deducting cover from gross exposure. An exposure report can show gross obligations, recorded reserves, eligible cover, and remaining exposure as separate lines. This makes the assumptions visible and prevents a reassuring net total from hiding a fragile source of protection.

Inside the mechanism. Recognize cover only to the extent it is eligible, accessible, correctly valued, and available at the relevant time. Apply a stated haircut to uncertain recovery values and avoid assuming independence between collateral value and the borrower’s failure. A stressed receivable from the same failing counterparty may offer little practical protection. Show uncovered exposure separately from expected loss and from immediate cash shortfall.

A concrete example. Collateral value, reserve access, and default risk can deteriorate together. Treating them as independent fixed averages can hide the combined loss path. The case has $1,450,000 of exposure. Its stated one-year PD and LGD imply $56,550.00 of expected loss, while the cover analysis leaves $776,000.00 of stress exposure. Monthly cash coverage is 1.14×. These are separate measures: one describes an average under probability assumptions, one describes available cover, and one describes a period’s funding capacity.

When the assumption fails. A supplier failure increases defaults while reducing the value of related collateral. Stress shared causes across PD, severity, cover eligibility, and liquidity together. The following worked sequence shows the reference condition, a stress condition, and a response condition with explicit synthetic data. These are comparative assumptions, not measured causal effects.

Follow a worked case3 conditions · 36 figures

Collateral value, reserve access, and default risk can deteriorate together. Treating them as independent fixed averages can hide the combined loss path.