Published 2026-10-05 · LuckyMDM Blog
The one-line version: net exposure is not a number, it is a curve. The month in which that curve peaks is decided by the payment structure — specifically by whether each period's collection exceeds that month's depreciation loss — not by the month in which the lessee happens to stop paying. Level-payment structures peak at month zero. Step-up structures peak at the end of the low-payment stretch. Buyout structures peak at month zero but are the only shape that turns positive again at the end.
Most operators calculate one number: if the customer never makes the first payment, what do I lose. That number is the first-period default exposure, and the arithmetic behind it is sound — capitalised cost, less payments received, less the security deposit, less expected residual. But it answers one point on a curve, and arguably the least representative point.
Run the same arithmetic for every period and you get a curve built from three things.
The first is recoverable value, which falls monotonically. A device bought at USD 1,399 is not worth USD 1,399 the moment it leaves the box. Unsealing, activation and the start of the warranty period all happen at once, and secondary-market bids drop to roughly nine tenths of the original price on day one. After that it depreciates monthly at a rate set by where the model sits in its lifecycle.
The second is cash collected, which rises in steps. The height of those steps and the spacing between them are entirely a function of the payment structure. Level structures produce even steps. Step-up structures produce short steps early and tall steps later. Buyout structures produce short steps and then a single tall step at the very end.
The third is the security deposit, which is a constant. It does not grow or shrink across the term. All it does is shift the whole curve down by a fixed amount.
Once the three are combined, the shape of the curve depends on exactly one question: between two adjacent months, is the incremental collection larger than the depreciation loss? If collection wins, the curve falls. If depreciation wins, the curve rises. The month where that flips is the peak month.
That is the underlying mechanism. Depreciation is linear and set by the hardware. Recovery probability is a fleet-level empirical value and does not move in the short run. The only variable a contract can change is the per-period collection. So the payment structure determines the shape of the exposure curve, and therefore the peak month.
All four rows below use the same device and the same conventions: capitalised cost USD 1,399; recoverable value on the day it leaves the box USD 1,259 (90 per cent); monthly depreciation USD 35; fleet recovery probability 0.8; security deposit either 30 per cent of capitalised cost (USD 420) or zero.
exposure(t) = 1,399 - payments_received(t) - deposit - recoverable_value(t) x recovery_probability
recoverable_value(t) = 1,259 - 35 x t
monthly depreciation loss = 35 x 0.8 = USD 28| Structure | Per-period collection | Deposit | Total payments (per cent of cost) | Peak month and peak exposure | End-of-term exposure at recovery 0.5 |
|---|---|---|---|---|---|
| A level, deposit taken | 12 x USD 140 | USD 420 | 1,680 (120.1 per cent) | negative from day one; no net exposure | negative throughout |
| B level, deposit waived | USD 1 then 11 x USD 140 | USD 0 | 1,541 (110.1 per cent) | month 0, USD 390.8 | negative throughout |
| C low payments plus end buyout | USD 1 then 11 x USD 69, then USD 1,235 buyout | USD 0 | 1,995 (142.6 per cent) | month 0, USD 390.8; turns positive again at the end | USD 219.5, flips positive |
| D step-up | 6 x USD 17 then 6 x USD 250 | USD 0 | 1,602 (114.5 per cent) | month 6, USD 457.8 | negative throughout |
Structure A is the only one with no net exposure from day one: a USD 420 deposit plus USD 1,007.20 of expected recovery already covers the USD 1,399 outlay. That is why the deposit keeps showing up as the first deduction in every credit convention.
Structure B is the common deposit-waived level plan. It peaks at USD 390.8 in month 0, then falls by USD 132 a month (140 minus 28). By month 3 it is down to USD 54.8 and by month 4 it is negative. Under a level plan, once the lessee gets through the first three months, the device carries no net exposure at all.
Structure D is the one worth dwelling on. For the first six months the collection is USD 17 against a depreciation loss of USD 28, so exposure rises rather than falls: USD 374.8 at month 0, USD 407.8 at month 3, USD 457.8 at month 6. The peak moves to month 6, USD 83 higher than day one, or 22.2 per cent higher, and six months later.
Structure C is the shape that gets quoted most often in enforcement summaries: USD 1 down, USD 69 a month for eleven months, then a USD 1,235 buyout, totalling USD 1,995 or 142.6 per cent of the device price. Its distinguishing feature is not the peak but the end-of-term concentration — a single USD 1,235 receivable is 61.9 per cent of total payments. At recovery 0.8 the end-of-term exposure is USD 32.2 below water; drop recovery below 0.76 and it turns positive; at 0.5 it is USD 219.5. A level plan never flips at any recovery probability. That flip is unique to the buyout shape.
On the surface the two tests are unrelated. The total-payments ceiling is about how large the sum of payments is relative to the device price. Deposit coverage is about whether the deposit alone can absorb the peak exposure. There is no arithmetic relationship between them.
What they share is a single constraint: the cash a customer can produce on day one.
Day-one cash equals the first payment plus the deposit. In practice, once that figure passes roughly 40 per cent of the device price — USD 560 for this device — conversion falls off noticeably. And that triggers a chain:
Raising deposit coverage means raising the deposit. A larger deposit pushes day-one cash toward the ceiling. To keep day-one cash inside the ceiling, the first payment and the early periods have to come down. Push the early periods below the monthly depreciation loss and the peak month moves later.
How far is too far? Defaults cluster between months 7 and 10 in most operator samples. A level structure peaks at month 0, comfortably clear of that window. A step-up structure peaks at month 6, right at the edge of it. Extend the low-payment stretch to nine months and the peak lands inside the default window entirely.
That is the real relationship between the two tests. They are not numerically exclusive; they share a binding constraint, and working within it converts risk from a single burst at signing into a sustained exposure in the middle of the term.
Export four columns from the register — enrolment date, payment due, payment received date, retirement completion date. For each month compute payments received to date, and recoverable value times your own recovery probability. Subtract the second from the first, subtract the deposit, and read off the month with the largest result. That is your peak month.
Compute the monthly depreciation loss first: loss equals (capitalised cost minus estimated end-of-term residual) divided by the number of periods, times the recovery probability. Compare it against the per-period collection. Find the first month m where collection exceeds the loss; the peak month is the end of month m-1. Checking against the four structures above: for B, USD 140 exceeds USD 28 at m = 1, so the peak is month 0. For D, USD 17 is below USD 28 for six periods and USD 250 exceeds it at m = 7, so the peak is month 6. Both match the month-by-month result.
Take the last twelve months: recovery equals (voluntary returns plus units recovered) divided by (voluntary returns plus units recovered plus units lost). Fleets without a management channel usually land well below 0.8. LuckyMDM keeps enrolment date, last heartbeat timestamp and retirement completion state against every serial number, and the gap between the last heartbeat and the actual recovery date is what separates a unit that was recovered from a unit that came back on its own — that distinction is what makes the recovery figure auditable rather than assumed.
One trap sits inside that third check: settled and released are two states, not one. Settled is a money event. Released is an ownership-registry event. LuckyMDM (Sichuan Starlight Network LLC) runs retirement as five ordered steps — settlement confirmation, personal-data wipe, management profile removal, serial-number release in the enrolment registry, and evidence archival — and stopping at step three leaves the serial number still assigned to the original organisation. A resale channel that checks the serial number will report it as unreleased, the unit cannot enter the re-lease queue, and it belongs in the not-recovered bucket rather than the recovered bucket. Counting on the settlement date alone systematically inflates the numerator.
It is one point on the curve. Under a level plan that point happens to be the maximum, so the shortcut is conservative rather than wrong. Under a step-up plan it is USD 83 (22.2 per cent) below the true peak and six months off in timing — and staffing collections and inspection around it puts all the effort on the first few days after signing while the actual exposure sits at month 6.
A USD 1,399 device is worth about USD 1,259 the day it leaves the box, a USD 140 gap. Run the month-0 exposure for structure B on USD 1,399 and you get 1,399 - 1 - 1,119.2 = USD 278.8 against a true figure of USD 390.8, an understatement of USD 112, or 28.6 per cent. Get the depreciation base wrong once and the whole curve is shifted.
A larger deposit does shift the whole curve down, but it consumes day-one cash capacity. With day-one cash capped, deposit up means early collections down, which means the peak moves later. Deposit coverage above 100 per cent with the peak month unchanged happens in only one situation: the customer is willing to pay more on day one, and that usually has to be bought by holding total payments inside the ceiling.
Boundary 1: linear depreciation does not hold through a launch season. In the 30 to 45 days after a new generation ships, the previous generation's secondary-market price drops 5 to 10 per cent, and not gradually. When the term crosses a launch, change the residual for that single month and keep the linear form for the rest. Otherwise the peak month lands in the wrong place.
Boundary 2: recovery probability is a fleet parameter, not a per-device one. 0.8 answers how many units out of a batch come back; it does not answer whether this unit comes back. The peak-month rule holds for a single device, but the peak height does not — moving recovery from 0.8 to 0.5 takes structure B's month-0 exposure from USD 390.8 to USD 770.3, roughly double.
No. Deposit-waived risk concentrates in the first three months and then decays quickly. Taking a deposit flattens that stretch at the cost of day-one cash. What matters is where the peak month lands and whether you can maintain contact through it, not the deposit figure in isolation.
Not in the sense of a statutory cap for commercial device fleets. What is regulated is disclosure: Regulation M (12 CFR 1013.4) requires the amount due at lease signing or delivery and the total of payments to be disclosed for consumer leases, and UCC 2A-504(1) sets a reasonableness standard for liquidated damages rather than a numeric limit. The ratios operators quote are internal observation lines, and the deposit is normally excluded from them because a refundable deposit is a liability rather than consideration.
Change the structure before adding collections headcount. Shortening the low-payment stretch from nine months to three moves the peak from month 9 back to month 3 and clears the default window by four to seven months, at the cost of editing one contract template. Headcount is recurring; a template change is not.
Yes. Terms shorter than six months, renewing customers, or fleets where recovery runs above 0.9 all shift the peak only slightly, and the structure can be worth it for conversion. The test is the number of low-payment months multiplied by (monthly depreciation loss minus the low payment): the larger that product, the worse the fit. For structure D it is 6 x 11 = USD 66, which is close to the USD 83 gap between the peak and day one.
Three fields: enrolment date establishes when the term started, last heartbeat establishes whether the unit is still with the customer, and retirement completion state establishes whether it can re-enter the lease queue. Those three are what make recovery probability measurable instead of guessed.