Your books say each unit costs $42 to make. That number decides your prices, your quotes, and whether a big order is worth taking — and part of it is a guess you made back in January, before you knew what the year would actually cost. By December, the guess and the reality will disagree. The only question is what you do with the gap.
If you make anything — furniture, packaged food, machined parts, candles you pour in your garage — you live with this gap whether you have named it or not. It has a name: underapplied or overapplied overhead. Understanding it is the difference between product costs you can trust and margins that quietly lie to you all year.
Why You Have to Guess in the First Place
Manufacturing overhead is everything it costs to run production that you cannot trace to one specific unit: the shop rent, the supervisor's salary, equipment depreciation, utilities, insurance, indirect materials like sandpaper and machine oil. You know the year's total only after the year ends.
But you cannot wait until December to cost a job. You quote work in March, close your books monthly, and file taxes on cost of goods sold that has to be reasonable every period. So at the start of the year you set a predetermined overhead rate: estimated annual overhead divided by an estimated activity base, usually machine hours, direct labor hours, or direct labor dollars.
Predetermined overhead rate = Estimated annual manufacturing overhead
/ Estimated annual activity baseSay you expect $200,000 of overhead this year and 10,000 machine hours of production. Your rate is $20 per machine hour. Every time a job runs 100 machine hours, you attach $2,000 of overhead to its cost — $20 times 100 — and move on. That attached amount is called applied overhead, and it flows into work in process like any other product cost.
This is a deliberate trade: timeliness in exchange for accuracy. The rate lets you cost jobs, price orders, and close months without waiting for the last utility bill. The price of that convenience is the variance at the end.
Underapplied vs. Overapplied: The Two Ways the Guess Misses
At year-end (or month-end), you compare two numbers:
- Actual overhead: what you really spent on indirect production costs.
- Applied overhead: what your predetermined rate attached to production (actual activity × the rate).
Underapplied overhead means actual spending exceeded what you applied. You spent more than your rate put into product costs, so your jobs look cheaper than they really were. The manufacturing overhead account carries a debit balance — costs waiting for a home.
Overapplied overhead is the mirror image: you applied more than you spent. Your jobs look more expensive than they really were, and the overhead account carries a credit balance — cost you claimed but never incurred.
Back to the example. Suppose the year ends with $214,000 of actual overhead and only 9,500 machine hours worked. Applied overhead is 9,500 × $20 = $190,000. Actual exceeded applied by $24,000, so you are underapplied by $24,000. Your product costs absorbed $190,000 of overhead while the shop really burned $214,000. That missing $24,000 has to land somewhere before you close the year.
Why the Two Numbers Never Agree
The gap always comes from one or both of two causes, and knowing which one dominates tells you what to fix.
Spending differences. You estimated $200,000 and spent $214,000. Maybe electricity rates rose, a machine needed an unplanned rebuild, or you hired the second-shift supervisor in June instead of "someday." When overhead spending itself moves, no rate set in January could have captured it. In the example, $14,000 of the $24,000 gap is pure spending — the estimate was just wrong.
Volume differences. You estimated 10,000 machine hours and ran 9,500. Every hour you did not run is $20 of overhead your rate never attached to anything. Fixed costs are the sting here: rent and depreciation accrue whether the machines run or not, so low-volume months systematically underapply. In the example, 500 missing hours × $20 accounts for the other $10,000 of the gap.
Notice how the two causes point at different remedies. A spending-driven gap says your budget needs work — get better quotes, revisit the maintenance plan. A volume-driven gap says your denominator needs work — you overestimated how busy the shop would be, and next year's rate should assume fewer hours (which raises the rate, which raises your quoted costs, which is the honest signal even if it stings).
One special case deserves a warning. If the volume shortfall comes from genuinely abnormal idle time — a fire shuts the plant for two months, a key customer cancels half the year's orders — accounting rules do not let you bury that in inventory. Under GAAP, abnormal amounts of idle facility expense are a current-period expense, recognized immediately rather than allocated to products. Capitalizing the cost of an empty factory into the few units you did make would overstate inventory and defer a loss that already happened. Normal ups and downs of demand get allocated; disasters get expensed.
Disposing of the Balance: Close It to COGS or Prorate It
You cannot leave the variance sitting in the overhead account at year-end — it is neither an asset nor a liability, just an unsettled clearing balance. Two accepted methods dispose of it, and materiality decides between them.
When the amount is small: close it straight to cost of goods sold
If the variance would not change anyone's reading of your statements, take the simple path: dump the whole balance into cost of goods sold in one entry.
- Underapplied (debit balance): debit Cost of Goods Sold, credit Manufacturing Overhead. COGS rises, gross profit falls — the year really did cost more than your jobs show.
- Overapplied (credit balance): debit Manufacturing Overhead, credit Cost of Goods Sold. COGS falls, gross profit rises — you charged your jobs for overhead you never spent.
This is what most small manufacturers do every year, because the variance is usually small relative to total COGS. One entry, done, and next year's rate gets the lesson.
In a plain-text ledger, the year-end close looks like this — a single transaction moving the unsettled overhead balance into COGS, dated with everything else you true up on December 31:
; Applied overhead fell $2,500 short of actual — close the debit balance to COGS
2026-12-31 * "Close underapplied manufacturing overhead"
Expenses:Cost-of-Goods-Sold 2500.00 USD
Expenses:Manufacturing-Overhead -2500.00 USDAn overapplied balance would flip the signs: credit COGS down by the excess. Either way, the overhead account ends the year at zero, which is exactly where a clearing account belongs.
When the amount is material: prorate it across inventory and COGS
If the variance is big enough to matter — a common rule of thumb is anything that would visibly move gross margin — dumping it all into COGS distorts the current period. The theoretically correct fix is to spread it across every account that absorbed overhead at the wrong rate: work in process, finished goods, and cost of goods sold.
The proration base is usually each account's ending balance (a common shortcut) or, more precisely, the amount of overhead applied within each balance. Take the shortcut version with our $24,000 underapplied balance:
Account Ending balance Share of total Share of $24,000
Work in process $30,000 7.5% $1,800
Finished goods $70,000 17.5% $4,200
Cost of goods sold $300,000 75.0% $18,000
Total $400,000 100.0% $24,000The entry debits each account for its share and credits manufacturing overhead for $24,000, zeroing it out. For an overapplied balance, the same table would reduce each account instead. The effect is to restate every inventory balance and COGS as though the rate had been right all along — ending inventory carries its fair share of the true overhead, and the current period's profit takes only its portion.
Which proration base should you use? Ending balances are simpler and usually close enough. Overhead-applied-in-each-balance is more defensible when labor and material mixes differ a lot between WIP, finished goods, and sold goods — but if your variance is large enough for the choice to matter, the deeper problem is a rate that needs rebuilding, not a base that needs refining.
The Mid-Year Correction Most Shops Skip
You do not have to wait until December to react. If by June you can see the rate is materially off — the new compressor doubled the power bill, or the big contract that justified 10,000 hours fell through — revise the rate for the remaining months. Recompute with updated full-year estimates of both spending and activity, and apply the new rate going forward. The first half stays as costed; the second half self-corrects.
This is standard practice, not creative accounting: the whole point of a predetermined rate is reasonable interim costing, and a rate you know is wrong stops being reasonable the moment you know. What you must not do is retroactively restate the months you already closed — adjust prospectively and disclose the change if your statements go to a lender.
Mistakes That Turn a Routine Variance Into a Real Problem
Leaving the balance on the balance sheet. During the year, an unsettled overhead balance is a normal work-in-progress artifact. At year-end it must be gone — closed to COGS or prorated. Parking it as a prepaid asset or burying it in inventory overstates assets and defers expense recognition into a year that did not cause it.
Prorating an immaterial variance. Spreading $800 across three accounts buys nothing but reconciliation work and rounding noise. Match the method to the magnitude: immaterial goes straight to COGS.
Burying abnormal idle time in the rate. A plant that ran at 40% capacity all year should not allocate a full year of fixed overhead to the units it produced. Expense the abnormal portion, fix next year's denominator, and let your margins tell the truth about the bad year.
Running one plantwide rate when departments differ wildly. If machining is machine-hour-intensive and assembly is labor-intensive, a single rate overcosts one department's jobs and undercosts the other's. Departmental rates — each with its own base — cost more to maintain but produce variances you can actually diagnose. A plantwide variance that never resolves into a clear spending-or-volume story is often two departmental stories cancelling each other out.
Forgetting the denominator next January. The most common repeat error: updating the overhead budget but copying last year's activity base. If volume is trending down and you keep dividing by the old, higher hours, you will underapply again by construction. Rebuild both halves of the fraction every year.
What to Do This Week
Pull your last closed period and compute the two numbers: actual overhead spent against overhead applied at your current rate. If you have never set a rate — many small shops just expense overhead as it hits and guess at product cost — estimate one now from last year's spending and this year's expected activity, and start applying it to new jobs. Either way, split the gap into its spending half and its volume half before you decide anything. The spending half is a budgeting conversation; the volume half is a capacity conversation; and the disposition entry at year-end is just bookkeeping once you know which conversation you are having.
