Units-of-Production Depreciation, Explained: How to Turn Lifetime Output Into a Per-Unit Write-Off

Your $120,000 CNC mill ran at full tilt in March, sat idle through a slow summer, then ran double shifts in November. Straight-line depreciation says it wore out by exactly the same amount every single month — which you know is fiction, because the spindle bearings keeping score in metal fatigue have never heard of a calendar. When an asset's wear tracks what it produces rather than how long you have owned it, charging the same depreciation every period quietly overstates your costs in slow months and understates them in busy ones. The units-of-production method fixes that mismatch by tying every dollar of depreciation to a unit of actual output: no units, no expense; double the units, double the expense.

The Method in Two Formulas​

Everything about units-of-production depreciation follows from two short formulas. First you convert the asset's depreciable cost into a rate per unit of output:

Depreciation per unit = (Cost − Salvage value) / Estimated lifetime output

Then each period you multiply that rate by what the asset actually produced:

Period depreciation expense = Depreciation per unit × Units produced this period

Cost is what you paid to acquire the asset and put it into service, including freight and installation. Salvage value is your estimate of what it will be worth when its productive life ends. Estimated lifetime output is the total number of units — widgets, miles, machine-hours, tons — you expect the asset to produce before it is retired. The numerator is the same depreciable base every depreciation method starts from; the denominator is where this method departs from the time-based pack, measuring life in output instead of years.

A Worked Example: The $120,000 Mill​

Suppose your shop buys a CNC mill for $120,000, including delivery and installation. The manufacturer rates the spindle for about 500,000 finished parts before a rebuild that effectively ends this machine's first life, and you estimate you could sell it for $20,000 at that point. Your per-unit rate is:

($120,000 − $20,000) / 500,000 parts = $0.20 per part

Every finished part that comes off the mill now carries twenty cents of depreciation. If the mill produces 130,000 parts in year one, 170,000 in year two, 120,000 in year three, and 80,000 in year four, the depreciation schedule writes itself:

Total depreciation over the four years is $100,000 — exactly cost minus salvage — and the book value lands on the $20,000 salvage value just as the 500,000th part rolls off. Notice what the calendar never got a vote on: year two's heavy production draws the heaviest charge, and a slow quarter draws almost nothing. That is the matching principle doing its job, pairing the cost of using up the machine with the revenue the parts it made will earn.

What Counts as a Unit​

The "unit" is whatever output measure best tracks the asset's wear, and picking the right driver is the most important judgment call in the whole method:

• Parts or pieces produced for manufacturing equipment like presses, mills, and molding machines.
• Miles driven for delivery vans, trucks, and any fleet vehicle — odometer readings make tracking nearly free.
• Machine-hours or flight-hours for equipment whose wear follows running time rather than countable output, from excavators to aircraft engines.
• Tons extracted or board-feet cut for mining and timber equipment, where the asset literally consumes itself against the resource.

The test is correlation: if doubling the measure roughly doubles the wear, it is a good driver. If the asset deteriorates sitting still — rust, rot, obsolescence — no output measure will save the method, and you are looking at the wrong tool. A delivery van's tires and brakes wear per mile, so miles work; the van's onboard computer becomes obsolete per year, but nobody depreciates a van for its computer.

Whatever driver you choose, you need a reliable way to count it every period. Odometers, hour meters, and production counters exist precisely for this. If your shop cannot say with confidence how many parts the mill ran last month, fix the tracking before you adopt the method — an output-based method fed by guessed output is worse than straight-line, because it carries the appearance of precision without the substance.

The Journal Entry Looks Familiar​

For all its structural differences, the units-of-production method books exactly the same journal entry as every other depreciation method. Using year two of the mill example:

• Debit Depreciation expense for $34,000
• Credit Accumulated depreciation — Equipment for $34,000

The calculation of the number changes; the plumbing does not. In a plain-text ledger, the year-end adjusting entry carries the per-unit math in a comment so a future reader can verify the rate instead of trusting it:

2026-12-31 * "CNC mill depreciation, year 2: 170,000 parts x $0.20/part"
  Expenses:Depreciation:Equipment     34000.00 USD
  Assets:Equipment:CNC-Mill:Accumulated-Depreciation  -34000.00 USD

Keep the lifetime estimate itself in your books too — as a dated note next to the asset's opening entry recording the $0.20 rate, the 500,000-part lifetime, and the $20,000 salvage assumption. When the estimate changes in year three, as estimates do, that note is what lets you recompute the rate on the remaining base instead of reconstructing history from memory.

Why Usage-Based Beats Time-Based When Output Swings​

Depreciation under generally accepted accounting principles must allocate cost in a systematic and rational manner, and nothing in that standard says the system has to be the calendar. When an asset's productive life is genuinely measured in output, units-of-production is the most faithful allocation available:

• It matches expense to revenue. The parts expensed this period are the parts sold this period, more or less. Straight-line, by contrast, charges your slow season for wear that happened during the rush.
• It makes unit economics honest. Twenty cents of machine cost per part is a number you can build into pricing, quoting, and make-versus-buy decisions. A flat $25,000 annual charge tells you nothing about what the next thousand parts cost.
• It handles assets with no meaningful time life. A quarry truck rated for 300,000 miles might last four years or nine depending on the jobs it draws. Depreciating it per mile sidesteps a useful-life guess you were going to get wrong anyway.

The difference shows up most clearly in seasonal and cyclical businesses. A snow-removal contractor's loaders earn their keep in four months; spreading their depreciation evenly across twelve charges the idle summer for winter's wear and flatters the winter margins that should be carrying it.

When Straight-Line Wins Instead​

Units-of-production is a specialist tool, and most assets want the generalist. Reach for straight-line — or another time-based method — when any of these describe the asset:

• It delivers steady benefit over time. Buildings, office furniture, and warehouse racking serve you about equally every month. Output-based depreciation on a desk is complexity with no payoff.
• Obsolescence, not wear, sets the life. Computers, software-driven equipment, and vehicles in image-sensitive businesses retire because they are outdated or off-brand, not because they are worn out. A server does not depreciate per query.
• Output cannot be tracked reliably. If counting units requires a manual log nobody maintains, the method's accuracy is theoretical and its administrative cost is real.
• You need predictable expense. Time-based methods give lenders and budget reviewers a smooth, forecastable number. Units-of-production expense bounces with production, which complicates covenants and budgets built on stable ratios.

There is also a hard boundary at the bottom: total depreciation can never push book value below salvage value. If the mill runs past 500,000 parts and is still cutting, you stop depreciating — the asset sits on the books at salvage value until disposal. Depreciation allocates cost; it does not appraise value, and a fully depreciated machine that still runs is worth whatever it is worth, not zero.

The Idle-Year Trap: Zero Units Means Zero Expense​

The method's signature feature is also its sharpest edge. If the mill produces nothing in a year — a canceled contract, a long retooling — depreciation expense that year is zero dollars. That is correct under the method's logic: no wear, no allocation. But consider what it implies:

• The asset still aged. Seals dry out, software falls behind, and the resale market moved on while the machine sat. None of that appears in a zero-depreciation year.
• Profits look better in the idle year than economics justify, because a real cost of holding the asset went unrecognized.
• If idleness stretches on, the lifetime-output estimate itself is probably stale — a machine that runs half its expected pace may become obsolete before it ever reaches 500,000 parts.

The fix is not to abandon the method but to revisit the estimate, which is exactly what accounting rules expect. A year of idleness is information: fold it into a revised lifetime-output figure and, if the machine will now retire with parts unmade, shorten the denominator so the remaining cost lands on the parts actually left to run.

Revising the Estimate: Prospective, Never Retroactive​

Estimates change — a rebuild extends the spindle's life, a lost contract shortens it — and accounting treats a revised lifetime-output figure as a change in estimate, applied prospectively. You never go back and restate prior years. Instead, you spread the remaining depreciable base over the revised remaining output.

Return to the mill. After year two, accumulated depreciation is $60,000 and book value is $60,000. Suppose a routine inspection reveals the spindle will manage only 400,000 lifetime parts instead of 500,000, meaning 100,000 remain after the 300,000 already run. The new rate is:

($60,000 remaining book value − $20,000 salvage) / 100,000 remaining parts = $0.40 per part

Year three's 120,000 parts cannot happen now — the machine only has 100,000 left in it — so the remaining $40,000 of depreciable cost lands on whatever those final parts turn out to be, and the schedule ends early. The mirror case works the same way: if a rebuild extends life to 700,000 parts, the remaining $40,000 spreads over 400,000 remaining parts at $0.10 each, and current-year expense drops. Either way, years one and two stay exactly as booked. Document the revision with its date and reason alongside the original estimate note, because "the rate changed and here is why" is the first thing an auditor, a buyer, or your future self will ask.

Book Versus Tax: Expect Two Different Numbers​

Here is the point that surprises owners most: units-of-production is a book method, not a tax method. For financial statements under GAAP it is perfectly acceptable. For your federal tax return, tangible business property generally must be depreciated under MACRS — the Modified Accelerated Cost Recovery System — which assigns assets to fixed recovery periods and prescribes its own rates. MACRS has no per-unit option for equipment.

The practical consequence is a book-tax difference you should plan for rather than discover. Your books may show $34,000 of mill depreciation in year two while the tax return shows whatever MACRS prescribes for that year — often larger early on, since MACRS front-loads deductions. That timing difference is routine: many businesses run one depreciation schedule for the books and another for the return, with the gap recorded as a deferred tax item. What you must not do is let the tax tail wag the book dog — switching your internal books to MACRS just to avoid keeping two schedules destroys the unit-cost information that made units-of-production worth adopting. Keep the books honest for management, take the deductions the law allows on the return, and reconcile the two at year-end with your preparer.

Five Mistakes That Break the Method​

Forgetting salvage value in the numerator. Depreciating the full $120,000 instead of the $100,000 depreciable base overstates every period's expense by 20 percent and drives book value below what you expect to recover. Subtract salvage first, every time.

Setting the lifetime estimate once and never revisiting it. A 500,000-part guess made at purchase is stale by year three. Review remaining-output estimates annually, the same cadence as the physical reality they describe.

Choosing a driver that does not track wear. Depreciating a forklift per pallet moved sounds clever until you realize battery cycles and idle hours drive its decline, not pallets. Pick the measure that predicts maintenance bills, not the one that is easiest to count.

Ignoring the zero-production year. Booking "about what we booked last year" during an idle period to keep the schedule smooth defeats the method's entire purpose. Zero units means zero expense — and a prompt to reconsider the lifetime estimate.

Applying it where obsolescence rules. The per-unit rate on a 3D printer looks rigorous right up until a new model makes yours a paperweight at 40 percent of its rated output. Time-based depreciation would at least have recognized the decline while it was happening.

Your Next Step: Rate One Asset This Week​

Pull the fixed-asset register and find one asset whose wear you can count — a delivery van with an odometer, a press with a counter, a copier on a per-page lease that already tracks volume for you. Write down its cost, your best salvage guess, and the lifetime-output figure from the manufacturer or your maintenance history. Divide, and you have a per-unit rate. Compare what this year's actual output implies against the straight-line charge you booked: if the two numbers tell different stories about what the year cost, you have found an asset that was miscast in a time-based role. Book the difference going forward, note the estimate where you will find it next year, and let the calendar go back to minding its own business.