Cost Estimation
What this domain tests
Cost estimation builds the cost functions that feed CVP, budgeting and pricing. The exam tests your ability to choose an estimation method, compute fixed and variable components, interpret R-squared and standard error, and apply learning curve theory.
Methods and rules
- Engineering method: build costs from technical specifications; accurate for new processes but costly and it ignores behavioural inefficiency.
- Account analysis: classify each ledger account as fixed or variable; judgement-based.
- High-low: variable rate = (cost at highest activity − cost at lowest activity) ÷ (highest activity − lowest activity). Choose the two points by activity level, not by cost level.
- Least squares regression uses all observations; R-squared measures goodness of fit (0 to 1); the standard error of the estimate indicates reliability; t-statistics test the significance of each cost driver.
- Learning curve: the cumulative average time falls by a fixed percentage each time cumulative output doubles; y = ax^b where b = log(learning rate) ÷ log 2.
Worked example
Overheads: 10,000 units cost Rs. 480,000; 18,000 units cost Rs. 640,000. Variable rate = (640,000 − 480,000) ÷ (18,000 − 10,000) = 160,000 ÷ 8,000 = Rs. 20 per unit. Fixed cost = 480,000 − (10,000 × 20) = Rs. 280,000. Predicted cost at 15,000 units = 280,000 + (15,000 × 20) = Rs. 580,000. Learning curve check: if the first unit takes 100 hours at an 80% rate, the cumulative average for 8 units = 100 × 0.8³ = 51.2 hours, so total time = 8 × 51.2 = 409.6 hours, not 8 × the time of the eighth unit.
Traps
Picking the highest and lowest cost observations instead of activity; ignoring step costs that break the linearity; mixing periods with different price levels; using the incremental unit time rather than the cumulative average in learning curve questions.