Why the gap is never just one number
By now you have built two things in this rung. From the standard cost guide you have a tidy expectation: this unit *should* take so much material at so much per kilogram, so much labor at so much per hour. And from the flexible budget guide you have learned to restate that plan for the volume you *actually* produced, so you are never caught comparing the cost of making 1,000 units against a plan that assumed 900. Variance analysis is what happens next: you lay the flexed standard beside the actual result, find they disagree, and refuse to leave it at that.
A single total gap — 'we spent 3,000 more on materials than the standard said we should' — is almost useless on its own, because at least two completely different things can cause it, and they call for completely different responses. Maybe you paid *more per kilogram* than expected. Maybe you *used more kilograms* than the recipe allows. Maybe both, partly cancelling. The whole art of variance analysis is to take that one lump and split it into a price component (did each unit of input cost what we assumed?) and a quantity component (did we use the amount we assumed?). One number that hides two stories explains nothing; two numbers, each telling one story, point straight at a cause.
Favorable and unfavorable: a sign, not a grade
Every variance carries a label — favorable or unfavorable — and the words trip up beginners because they sound like praise and blame when they are really just direction. A favorable variance simply means the actual cost came in *below* the standard, so profit was higher than the flexed plan predicted. An unfavorable variance means actual cost came in *above* standard, denting profit. That is the whole definition: favorable points the right way for profit, unfavorable the wrong way. Neither word, on its own, tells you whether anyone did a good or a bad job.
The honest manager treats both labels with the same suspicion. A favorable materials price variance — beans bought cheap — can be wonderful, or it can be the sound of a buyer grabbing a low-grade batch that will later cause an unfavorable *quantity* variance as workers throw out the ruined pieces. A favorable labor efficiency variance — the job finished fast — might mean a skilled crew, or might mean corners cut that the returns department will pay for next month. Favorable is not a synonym for good, and unfavorable is not a synonym for bad. They are arrows; the question of *why* the arrow points where it does is the part that matters, and a label never answers it.
Splitting the materials gap: price and quantity
Take materials first, because the logic here repeats for everything else. The direct-materials variances come in a matched pair. The price variance asks: for the quantity we actually bought, did each kilogram cost more or less than standard? You hold the quantity fixed at what actually happened and let only the price differ — (actual price − standard price) × actual quantity. The quantity variance (often called the usage variance) asks the opposite: at the standard price, did we use more or fewer kilograms than the recipe allows for the output we made? Now you hold price fixed at standard and let only the quantity differ — (actual quantity − standard quantity allowed) × standard price.
Why hold one thing still while you move the other? Because that is the only way to give each cause a clean number. If you let price and quantity both wander at once, the gap they produce together cannot be honestly handed to any single person. By freezing quantity to isolate price, and freezing price to isolate quantity, you build two figures that *add back up* to the total materials gap — no overlap, nothing lost. That is the quiet beauty of the method: it is not just decomposition, it is decomposition that reconciles. The price variance plus the quantity variance, with their favorable/unfavorable signs, always equals the whole.
BREAD CO. -- MATERIALS FOR ONE WEEK Standard: 0.5 kg flour per loaf at 2.00 / kg Actual output: 1,000 loaves -> standard quantity ALLOWED = 1,000 x 0.5 = 500 kg Actual: bought & used 540 kg, paid 1.90 / kg PRICE VARIANCE (freeze quantity, move price) (1.90 - 2.00) x 540 kg = -0.10 x 540 = 54.00 FAVORABLE (each kg was 0.10 cheaper than standard) QUANTITY VARIANCE (freeze price, move quantity) (540 - 500) x 2.00 = 40 kg x 2.00 = 80.00 UNFAVORABLE (used 40 kg more flour than the recipe allows) CHECK -- the two must rebuild the total gap Actual cost 540 x 1.90 = 1,026.00 Standard cost 500 x 2.00 = 1,000.00 Total gap = 26.00 UNFAVORABLE Price (-54 F) + Quantity (+80 U) = +26 U ok
Read what those two numbers say together, because that is where insight lives. The price variance is favorable: purchasing got the flour cheap. The quantity variance is unfavorable: the bakery burned through 40 extra kilograms. A naive eye sees one win and one loss. A trained eye suspects they are the *same event* — cheap flour was lower-grade, so more of it ended up scraped off the bench. The cheerful price saving (54) was more than swallowed by the waste it provoked (80). This is why you never celebrate a favorable variance without checking what its partner did, and why splitting the gap is worth the trouble: the split is what lets the story surface.
Labor and overhead: the same idea, two more times
Once the materials pattern clicks, labor is the very same shape with the nouns swapped. The direct-labor variances split into a rate variance — the labor twin of price — and an efficiency variance — the labor twin of quantity. The rate variance freezes hours and lets the wage rate differ: (actual rate − standard rate) × actual hours. It signals whether you paid your workers more or less per hour than planned, which usually traces to who was scheduled — overtime, a senior hand on a junior task, a raise the standard never absorbed. The efficiency variance freezes the rate and lets the hours differ: (actual hours − standard hours allowed) × standard rate. It signals whether the work took longer or shorter than the standard allows for that output.
Watch how labor variances braid together, just as the materials pair did. Schedule a senior baker on routine loaves and the rate variance turns unfavorable — you are overpaying for the hour — but the efficiency variance may turn favorable, because the veteran works fast and wastes little. Whether the trade was wise depends on whether the hours saved were worth more than the premium paid. This is the recurring lesson of variance analysis: the variances rarely live alone. Each one is a thread, and you understand the cloth only by seeing how the threads cross — a favorable here often born of an unfavorable there.
Overhead is the trickier cousin, and at this level you only need its shape. Recall the predetermined overhead rate from the costing rung: because overhead is indirect, you applied it to products with a planned rate per labor hour or machine hour rather than tracing it. The overhead variances measure two distinct slips in that scheme. A spending variance asks whether the overhead you actually incurred — the factory's electricity, supervision, supplies — was more or less than expected for the activity level. A volume variance is stranger: it asks whether you ran the factory at the activity level your fixed-overhead rate assumed, because that rate was built by spreading a fixed pool of cost over a planned number of hours. Make fewer hours than planned and the fixed overhead is under-applied; the volume variance is the accounting footprint of unused capacity, not of overspending.
What each variance signals — and to whom
Variances earn their keep only when they change a conversation, and that means handing each one to the person who can actually explain it. This is the purpose of responsibility accounting: you organize the business into responsibility centers — units a named manager controls — and you route each variance to the center whose decisions plausibly caused it. A materials *price* variance lands on purchasing, because purchasing chooses suppliers and negotiates prices. A materials *quantity* variance lands on production, because the factory floor controls how carefully the input is used. Splitting the gap was step one; this routing is what the split was *for*.
The most common responsibility center for these production variances is a cost center — a unit judged only on the costs it incurs, because it neither sells anything nor controls its own revenue. A bakery's mixing room, a factory's machining bay, a hospital's laundry: each is handed a budget of standard costs and asked to explain its variances against it. The guiding principle is *controllability* — a manager should answer only for what their decisions could move. Holding a production supervisor responsible for an unfavorable price variance caused entirely by a global wheat shortage is not accountability; it is scapegoating, and it teaches managers to game the numbers rather than improve the work.
Read holistically, the variances also grade the *planning* itself. If every center reports large variances month after month, all in the same direction, the likeliest culprit is not a roomful of incompetent managers but a standard cost that no longer reflects reality — set when flour was cheaper, or when the old slow machine was still running. Variances that are small and scattered, some favorable and some unfavorable, are the sign of a healthy standard and a steady operation. Persistent one-directional variances are the system quietly telling you to *update the plan*, not punish the people. A good analyst always asks, before blaming anyone, whether the zero they are measuring from is still the right zero.
Using variances well — and their limits
Put to honest use, variance analysis runs a simple loop. First compute the variance against the *flexed* standard, never the original static budget, so volume changes do not masquerade as efficiency failures. Then investigate by exception — chase the variances large or persistent enough to matter and let the trivial ones lie, because a manager who investigates every two-dollar wobble drowns. Finally, ask not just *what* the number is but *why*, and feed the answer back into either better operations or a better standard. The variance is never the destination; it is a question mark that earns its place only by starting the right conversation.
Be candid about what this tool cannot do. It is backward-looking — it dissects a period already gone, never guaranteeing the next one will behave. It can breed perverse behavior if managers are squeezed too hard on a single variance: a buyer chasing a favorable price variance hoards cheap, shoddy material; a supervisor chasing a favorable efficiency variance overproduces inventory nobody ordered just to keep the machines 'busy'. And it sees only what the standards measure — it is mute on quality, on safety, on a customer quietly lost. Variances are one instrument on the dashboard, sharp at what they read and silent on everything else; the wise manager reads them alongside the rest, never instead of it.
And with that, the budgeting rung closes a full circle. You built a standard to say what a unit should cost, flexed the budget to the real volume so the comparison would be fair, and now you have learned to split each gap into price and quantity, label its direction, route it to a responsibility center, and read the whole set as a verdict on both the operation and the plan. That loop — plan, measure, explain, adjust — is the heartbeat of management accounting. Everything past here, from richer performance measures to investment decisions, is this same loop turned on bigger and bigger questions.