ACC2200 · Introduction to Management Accounting
Cost Behaviour, Drivers and Estimation
Cost behaviour is the analytical spine of ACC2200 Introduction to Management Accounting at Monash University: almost every later tool — product costing, overhead rates, ABC, CVP and relevant-cost decisions — assumes you can say how a cost moves as activity changes. This chapter works through three linked ideas: cost behaviour (the real cost-to-activity relationship), cost estimation (using data to model it), and cost prediction (forecasting the dollar cost at a planned level). You will meet the cost driver and its hierarchy, the five behaviour patterns and the relevant range, the linear cost function Y = a + bX, and the two workhorse estimation methods — the high-low method and regression (read through R²). Monash flags Week 3 as a primary emphasis of the closed-book final exam, where marks are split roughly 60% calculation and 40% discussion.
What this chapter covers
- 011. The estimation chain — cost behaviour → estimation → prediction, and why the three differ
- 022. Cost drivers — volume-based (units, labour hours, machine hours) vs non-volume-based (setups, inspections)
- 033. The cost-driver hierarchy — unit, batch, product-sustaining and facility-level costs
- 044. Behaviour patterns — variable, fixed, step-fixed, semivariable (mixed) and curvilinear costs
- 055. The relevant range — the activity band over which an assumed pattern (and your a and b) holds
- 066. The linear cost function — Y = a + bX, with a = fixed component and b = variable rate
- 077. The high-low method — splitting a mixed cost from the two extreme-activity observations
- 088. Regression and R² — the least-squares line through all points and reading its fit
- 099. Estimation pitfalls — outliers, inflation, mismatched periods and allocated fixed costs
High-low method — split a mixed cost, write the equation, predict
- +1Pick the two extreme ACTIVITY levels (the km column), not the extreme costs: HIGH = 18,000 km ($9,400), LOW = 6,000 km ($5,800).
- +1Variable rate b = (cost at high − cost at low) ÷ (km at high − km at low) = (9,400 − 5,800) ÷ (18,000 − 6,000) = 3,600 ÷ 12,000 = $0.30 per km.
- +1Fixed cost a = cost at high − (b × activity at high) = 9,400 − (0.30 × 18,000) = 9,400 − 5,400 = $4,000 per month.
- +1Check a at the low point: 5,800 − (0.30 × 6,000) = 5,800 − 1,800 = $4,000 ✓ — the two points agree, so a is correct.
- +1Write the cost equation: Total cost = 4,000 + 0.30 × (km driven), i.e. Y = 4,000 + 0.30X.
- +1Predict at 14,000 km: 4,000 + 0.30 × 14,000 = 4,000 + 4,200 = $8,200 — valid only within the relevant range of 6,000–18,000 km.
Key terms
- Cost driver
- An activity or factor whose change causes a change in total cost — the X in the cost function. Volume-based drivers (units, labour hours, machine hours) scale with output; non-volume-based drivers (setups, inspections) reflect complexity.
- Relevant range
- The band of activity over which an assumed cost-behaviour pattern, and the estimated fixed cost a and variable rate b, remain valid. Predictions outside this range are unreliable.
- Variable cost
- A cost whose total rises in proportion to activity while the cost per unit stays constant — for example direct materials.
- Fixed cost
- A cost whose total stays constant over the relevant range while the cost per unit falls as volume rises — for example factory rent.
- Semivariable (mixed) cost
- A cost with a fixed base plus a variable element, modelled as Y = a + bX — for example electricity with a standing charge plus usage.
- Step-fixed cost
- A cost that is fixed across a range of activity, then jumps to a higher level once a capacity threshold is crossed — for example adding a second supervisor.
- High-low method
- A quick two-point estimation technique that fixes the variable rate from the highest- and lowest-activity observations, then solves for the fixed cost. Simple, but hostage to those two points.
- R² (coefficient of determination)
- The proportion, from 0 to 1, of the variation in cost explained by the driver in a regression. A measure of fit — high R² means the driver tracks cost well, but it never proves causation or licences extrapolation.
Cost Behaviour, Drivers and Estimation FAQ
What is the difference between cost behaviour, cost estimation and cost prediction?
Cost behaviour is the real underlying relationship between a cost and its driver; cost estimation is the analytical step that models that relationship as a function (finding the fixed component a and variable rate b); and cost prediction plugs a planned activity level into the estimated function to forecast the dollar cost. Behaviour is the reality, estimation is the model, and prediction is the output.
How does the high-low method work, and why does it use activity rather than cost?
High-low takes the observations with the highest and lowest levels of the ACTIVITY (the cost driver), computes the variable rate as the change in cost divided by the change in activity, then solves for the fixed cost using either point. It uses activity, not cost, because activity is what drives the cost — picking the highest or lowest dollar figure instead is the single most common error in this topic, since a spike in cost can occur at a mid-range activity level.
What does R² tell me in a regression cost estimate?
R² is the proportion of the variation in cost that the driver explains, on a scale from 0 to 1. An R² of 0.92 means about 92% of the movement in cost is explained by the driver, which signals a strong fit. But R² does not prove the driver causes the cost, and it does not justify predicting outside the historical relevant range — always state both the strength and the limits.
Is a cost driver the same as the root cause of a cost?
Not necessarily. The driver you use to estimate a cost (often an easy-to-measure volume driver such as number of samples or machine hours) can differ from the root cause that actually generates it (such as staff skill mix or product complexity). Use volume drivers for prediction, but hunt root-cause drivers when the goal is cost reduction.
How is cost behaviour examined in ACC2200 at Monash?
Monash flags Week 3 as a primary emphasis of the closed-book, calculator-only final exam, and it also underpins the early-semester group presentation. Expect a multi-part question that asks you to run the high-low method (or read regression output), write and use a Y = a + bX equation, predict a cost, and discuss the limitations — the discussion marks are worth about as much as the arithmetic, so practise the full arc, not just the calculation.
Can AI help me with cost behaviour, drivers and estimation?
Yes — ask Sia to walk through any cost behaviour, drivers and estimation problem or concept step by step, the way Monash University tests it. Sia is an AI tutor that explains each move — how to pick the high-low extremes, why the fixed cost is checked at the other point, or how to phrase an R² interpretation — so you understand the method and can re-derive the numbers yourself under exam conditions.
Studying with AI? Sia — free AI accounting tutor works through ACC2200 step by step.
Exam move
Cost behaviour is a calculation-plus-discussion topic, so drill the whole arc rather than just the arithmetic. Take any six-point cost table, sort it by the activity column, run high-low, write the Y = a + bX equation, predict a cost, check your fixed cost at the other point, and add two limitations — all inside about six minutes. Then say aloud what a, b and R² mean in the firm's own words, because the exam awards the interpretation marks for meaning, not for re-quoting statistics. Keep a short list of the estimation pitfalls (outliers, inflation, mismatched time periods, allocated fixed costs, learning-curve effects) and tie two of them to the scenario whenever a question asks how confident you are. Use the revision/SWOTVAC week before the end-of-semester exam period to repeat this with fresh tables until picking the extremes by activity — never by cost — is automatic.
