Standard costing works when the standards are re-tested against what production actually consumed. Set at the start of the year and left alone, it produces a precise number that is quietly wrong, and prices set from it inherit the error.
Standard costing is a good system. It lets a manufacturer value inventory, price work and see performance without recomputing everything from first principles each month. It depends on one thing: that the standards are periodically tested against what production actually consumed.
In practice the standards are set at the start of the year, material prices move, yields change, the product mix shifts, and the variance account absorbs the difference. By the fourth quarter the standard describes a factory that no longer exists.
The variance account is where information goes to die
A single variance line tells you the standard was wrong by some amount. It does not tell you whether that was price, usage, yield, mix or absorption — and those have completely different responses. Price is procurement. Usage is the shop floor. Absorption is a volume assumption made in an office.
- Material price variance
What was paid against what was assumed. Procurement's to explain, and often the fastest to correct.
- Material usage and yield variance
What was consumed against what should have been. This is the shop floor, and it is where scrap and rework hide.
- Labour rate and efficiency variance
Separate, because paying more per hour and taking more hours are different problems.
- Overhead absorption variance
Usually the largest and least examined. Absorption assumes a level of activity; when actual volume differs, the rate is wrong for every unit made.
A variance you cannot decompose is not information. It is the size of your ignorance, reported monthly.
Scrap and rework that never became a cost
Material that is scrapped, reworked or taken as a sample frequently moves without a document. Physically it has gone; in the records it is still there until a stock count finds the difference, at which point it is written off as a stock adjustment with no cause attached.
The cost was real, it belonged to a specific product and batch, and by the time it appears it belongs to nothing in particular. Yield loss recorded as it happens is one of the higher-return changes available in a plant, and it is mostly a paperwork discipline rather than a systems project.
Work in progress valued by formula
Work in progress is often valued by applying a completion percentage to a standard cost. If the standard is stale and the completion estimate is a convention rather than an observation, the valuation is two assumptions multiplied together — and it sits on the balance sheet.
Testing it periodically against what is physically on the floor, and what has actually been booked to it, is the only way to know whether the number means anything.
How often to re-test
Standards do not need resetting monthly; that defeats the purpose. But they should be reviewed against actuals on a stated cycle, with a rule for when a standard is revised rather than argued with. Quarterly works for most manufacturers, more often where material prices are volatile.
The test is simple: if the variance on a line is consistently in one direction, the standard is wrong, not the factory.