Cost Concepts, Classification, Behavior and Estimation

Introduction

Cost accounting provides the language through which organizations plan operations, price products, control resources and evaluate performance. A cost is not inherently “fixed,” “variable,” “direct,” or “relevant.” Its classification depends on the purpose of analysis, the cost object selected, the time horizon and the decision being made. The same expenditure may be a direct cost for one cost object, an indirect cost for another, and a sunk cost in a later decision.

For example, factory electricity may be an indirect manufacturing cost when the cost object is a product, but a direct cost of operating a particular machine when the cost object is that machine. Sound management accounting therefore requires disciplined identification of the cost object, the activity measure and the decision context.

Learning Objectives

By the end of this chapter, you should be able to:

1. Classify costs as direct or indirect, fixed, variable or mixed, product or period, and relevant, sunk or opportunity costs.
2. Explain cost behavior, cost drivers, relevant range and cost functions.
3. Estimate costs using account analysis, the high–low method, regression and learning-curve analysis.
4. Interpret estimation results and recognize their limitations.
5. Apply cost information to planning, pricing, outsourcing and operational decisions.

1. Cost Objects and Basic Classifications

A cost object is anything for which cost information is required. It may be a product, service, customer, department, project, contract, distribution channel or activity. A cost pool is an accumulation of costs sharing a common allocation basis. A cost driver is a factor that causes or explains changes in the cost of an activity.

Direct and indirect costs

A direct cost can be economically traced to a particular cost object. Direct materials used in a product and wages of employees working exclusively on a contract are typical examples.

An indirect cost cannot be economically traced to one cost object, or tracing it would cost more than the information is worth. Factory rent, production supervision and general equipment depreciation are usually indirect product costs.

The distinction is relative rather than absolute. A supervisor’s salary is indirect to individual units but may be direct to the production department. Cost assignment should balance accuracy and cost-effectiveness. Excessive tracing can be as inefficient as arbitrary allocation.

Manufacturing and non-manufacturing costs

Manufacturing costs include:

• Direct materials: materials physically incorporated into a product and economically traceable to it.
• Direct labour: labour effort directly associated with production.
• Manufacturing overhead: all other factory costs, including indirect materials, indirect labour, factory utilities, maintenance and factory depreciation.

Product cost is commonly expressed as:

Selling, distribution and administrative costs are generally non-manufacturing costs.

Product and period costs

A product cost is attached to inventory and becomes an expense, usually cost of goods sold, when the related product is sold. A period cost is charged against income in the period incurred because it is not assigned to inventory. Advertising, sales commissions and head-office administration are normally period costs.

The classification depends on the reporting framework and circumstances. Some costs, such as abnormal production waste or storage costs unrelated to production, may be expensed rather than included in inventory under applicable financial reporting standards. Management accounting may also use internal classifications different from external reporting classifications.

2. Cost Behavior

Cost behavior describes how total cost changes as an activity driver changes. The most useful basic model is the linear cost function:

Here, Y is total cost for the period, a is total fixed cost, b is variable cost per unit of activity, and X is the activity level. The equation assumes a relevant range, a stable cost structure and a sufficiently linear relationship between cost and activity.

Fixed costs

A fixed cost remains constant in total within a relevant range and time period, although fixed cost per unit changes inversely with volume:

Examples include factory rent, salaried management and straight-line depreciation. Fixed costs may be committed, such as long-term facility costs, or discretionary, such as training and advertising budgets. Discretionary fixed costs can often be changed more readily in the short term.

A fixed cost is not necessarily fixed forever. Rent may be fixed for one year but change when a lease is renewed. Capacity-related costs may also be step-fixed: they remain constant over a band of activity and jump when additional capacity is required.

Variable costs

A variable cost changes in total in proportion to the activity driver, while variable cost per unit remains constant within the relevant range.

Examples include direct materials and transaction processing fees. A variable cost may be linear, but some costs are nonlinear. Quantity discounts can create a declining unit cost, while overtime premiums can create an increasing unit cost.

Mixed costs

A mixed cost contains both fixed and variable components. Telephone expense with a monthly access charge plus usage charges is an example. Its cost function is Y = a + bX.

A step-variable cost is constant for small changes in activity but increases in steps as activity passes thresholds. Temporary labour hired in groups of ten is an example. It should not be forced into a simple linear model when the step pattern matters for decisions.

Relevant range and time horizon

The relevant range is the activity band over which assumptions about cost behavior remain valid. A cost function estimated for 10,000–20,000 machine hours should not automatically be applied to 50,000 hours. The relevant range incorporates available capacity, wage agreements, supplier terms and technology.

The time horizon matters. In the short run, factory rent may be fixed; in the long run, facilities can be expanded or closed. “Fixed” and “variable” are therefore conditional descriptions, not permanent properties of costs.

3. Decision-Relevant Cost Concepts

Relevant and irrelevant costs

A relevant cost is a future cost that differs between alternatives. Relevance has two tests: the cost must occur in the future, and it must differ among the options being compared.

An irrelevant cost does not change between alternatives. It may still be important for reporting or control, but it should not influence a particular incremental decision.

For a special-order decision, relevant items may include additional materials, incremental labour, packaging and extra shipping. Allocated corporate overhead is irrelevant if it will be incurred regardless of acceptance.

Sunk costs

A sunk cost is a past cost that cannot be changed by current or future decisions. The purchase price of an existing machine is sunk. Its original cost should not determine whether the machine is used, replaced or abandoned. However, the machine’s disposal value, operating costs and replacement cost may be relevant.

Opportunity costs

An opportunity cost is the benefit forgone by selecting one alternative instead of the next-best alternative. If a factory uses capacity for Product A and thereby gives up contribution from Product B, the lost contribution is an opportunity cost of Product A.

Opportunity costs are normally absent from financial accounting records but essential in managerial decisions. They are economic costs, not necessarily cash payments.

Avoidable and unavoidable costs

An avoidable cost can be eliminated by selecting an alternative; an unavoidable cost continues regardless of the decision. The distinction is often more useful than the direct–indirect distinction for short-term decisions. A direct cost may be unavoidable under a binding contract, while an allocated indirect cost may be avoidable if a department closes.

4. Cost Drivers and Cost Functions

A cost driver should have a logical causal relationship with the cost. Common drivers include units produced, machine hours, labour hours, number of setups, purchase orders, inspections and delivery kilometres. Traditional volume-based systems may assign all overhead using labour or machine hours, whereas activity-based costing uses multiple activity drivers.

A useful cost model has four elements:

1. Dependent cost: the cost to be estimated.
2. Independent driver: the activity believed to cause cost changes.
3. Functional form: linear, step, curvilinear or another relationship.
4. Relevant range and time period: the conditions under which the model applies.

For several drivers, a multiple-cost model may be written as:

Y is total estimated cost; a is fixed cost; Xᵢ is driver i; and bᵢ is the estimated cost per unit of driver i. The model assumes that the selected drivers are measured consistently and that their effects can be estimated without unacceptable multicollinearity or omitted-variable bias.

A high coefficient of determination, R², indicates that the model explains much of the observed variation, but it does not prove causation. Managerial knowledge and operational logic remain essential.

5. Cost Estimation Methods

Account analysis

Account analysis classifies each ledger account as variable, fixed or mixed based on managerial judgment and operational knowledge. The estimated cost function is then formed by combining the classifications.

Suppose monthly maintenance costs are estimated as:

Estimated maintenance cost is:

Account analysis is quick and practical, but its reliability depends on the expertise and objectivity of the analyst. It can be improved by reviewing invoices, contracts and operational interviews rather than relying solely on account labels.

High–low method

The high–low method uses the highest and lowest activity observations, not necessarily the highest and lowest costs. The variable rate is:

Fixed cost is then:

Worked example. A delivery department reports:

Thus:

Using January:

The cost function is therefore Y = $8,000 + $3X. At 4,000 deliveries, estimated cost is $20,000.

The method is simple but uses only two observations and may be distorted by unusual months. It should be avoided when the high and low observations are outliers or when the relationship is visibly nonlinear.

Regression analysis

Regression uses all available observations to estimate the relationship between cost and one or more drivers. In simple least-squares regression, the estimated slope is:

and the intercept is:

Xᵢ and Yᵢ are observed activity and cost values; X̄ and Ȳ are their sample means. The method minimizes the sum of squared residuals, where a residual is actual cost minus estimated cost.

Regression is generally superior to high–low because it uses all observations and provides diagnostic statistics such as R², standard errors and residual plots. Nevertheless, a statistically good fit may be economically meaningless if the driver is not causal, the data include structural changes or the sample lies outside the relevant range.

Best practice is to inspect a scatter plot, investigate unusual observations, test economically plausible drivers, distinguish correlation from causation and validate the model with later data.

Learning-curve analysis

A learning curve models declining average labour time as cumulative production increases. The cumulative-average model is:

Y is cumulative average labour hours per unit after producing X units; a is the hours for the first unit; and b is the learning exponent. If the learning rate is r, then:

The learning rate is the ratio of cumulative average time at a doubled output level to the prior cumulative average time. For an 80% learning rate, b ≈ −0.3219. The negative exponent means average time falls as cumulative output rises.

Example. The first unit requires 100 labour hours and the learning rate is 80%. At 2 units, cumulative average time is 80 hours, so total time is 160 hours. At 4 units, cumulative average time is 64 hours, so total time is 256 hours. The additional time for units 3 and 4 is 96 hours.

Learning curves are most useful for new, labour-intensive projects with repetitive tasks. They are less reliable when automation, design changes, worker turnover or interruptions alter the learning process. The cumulative-average model must not be confused with a unit-time model; the two produce different estimates.

6. Integrated Applied Case: Northstar Appliances

The following is an illustrative case.

Background

Northstar Appliances is considering a one-year contract to assemble 12,000 countertop devices. Existing factory capacity is limited. Management has collected monthly assembly-hour and maintenance-cost data and wants to estimate incremental cost and assess the contract.

Problem

The finance manager proposes using the high–low method. The operations manager argues that maintenance is driven by machine hours, while the contract may displace a profitable product. The company must avoid treating allocated fixed overhead as automatically relevant.

Analysis

The highest activity month involved 9,000 machine hours and $42,000 maintenance cost; the lowest involved 5,000 hours and $30,000. High–low gives:

Estimated maintenance at 8,000 hours is $39,000. However, management discovers that the high month included an emergency repair, so it may be an outlier. Regression using all twelve months produces a more stable estimate of $14,200 fixed cost and $3.05 per machine hour, with residual review showing no major pattern. The model should be validated before use.

The contract requires 9,600 additional machine hours. Incremental maintenance estimated from the regression model is approximately:

The fixed maintenance component is not automatically incremental if the factory will operate within existing capacity. Direct materials, contract labour, special packaging and shipping are relevant. The opportunity cost of using capacity is also relevant: if displaced production would generate $48,000 contribution margin, that amount is an economic cost of accepting the contract.

Approach and outcome

Northstar should compare the contract price with incremental costs plus the opportunity cost of capacity, rather than with full allocated product cost. It should document the cost-driver choice, test capacity assumptions and perform sensitivity analysis for machine hours and material prices.

Learning points

1. High–low is a screening tool, not automatically the best final model.
2. Outliers should be investigated rather than mechanically retained or deleted.
3. Fixed cost is relevant only when it changes between alternatives.
4. Opportunity cost can make a seemingly profitable order unattractive.
5. Cost estimates require operational validation as well as statistical analysis.

7. Limitations, Governance and Contemporary Practice

Cost estimates are uncertain because data may be incomplete, accounting classifications may be inconsistent and technology may change the cost structure. Managers may also manipulate cost assumptions to justify budgets or performance targets. Governance requires documented assumptions, version-controlled models, review by operational experts and clear separation between actual results and forecasts.

Spreadsheets remain useful for small analyses, while enterprise systems and business-intelligence tools can combine general-ledger, production and operational data. AI and automated analytics can identify potential drivers and anomalies, but outputs require human review. Users should check data quality, explainability, privacy, access controls and model drift. An algorithm that predicts cost accurately but uses a biased or unstable driver may produce poor managerial decisions.

Best practice includes defining the decision and cost object first; selecting causal drivers; separating fixed, variable and step components; checking units and relevant range; examining outliers; comparing methods; validating forecasts against subsequent actual costs; and communicating uncertainty rather than presenting estimates as facts.

Chapter Summary

Cost classification is purpose-dependent. Direct and indirect costs concern traceability to a cost object; fixed, variable, mixed and step costs concern behavior over an activity range; product and period costs concern financial reporting treatment. Relevant costs are future and differential, whereas sunk costs are past and unavoidable. Opportunity costs measure benefits forgone and are vital even though they are usually absent from accounting records.

Cost behavior is commonly represented by Y = a + bX, subject to relevant-range and stability assumptions. Account analysis uses informed classification, high–low estimates a line from two observations, regression uses all observations and provides diagnostics, and learning curves model improving efficiency as cumulative production rises. No method replaces judgment: causal reasoning, data quality, capacity analysis and validation determine whether an estimate is useful.

Key Terms

Cost object: Anything for which cost information is required.

Cost driver: A factor that causes or explains cost changes.

Direct cost: A cost economically traceable to a cost object.

Indirect cost: A cost not economically traceable to one cost object.

Fixed cost: A cost constant in total within a relevant range.

Variable cost: A cost changing in total with activity.

Mixed cost: A cost containing fixed and variable components.

Relevant range: Activity interval over which cost assumptions hold.

Product cost: Cost attached to inventory until the product is sold.

Period cost: Cost expensed in the period incurred.

Relevant cost: Future cost differing among alternatives.

Sunk cost: Past cost unaffected by current decisions.

Opportunity cost: Benefit forgone from the next-best alternative.

Avoidable cost: Cost eliminated by selecting an alternative.

Cost function: Mathematical relationship between cost and driver activity.

High–low method: Estimation using the highest and lowest activity observations.

Regression analysis: Statistical estimation using observed data.

Learning curve: Model of improving efficiency with cumulative production.

Residual: Actual cost minus estimated cost.

Review Questions

Short Answer Questions

1. Why is cost classification dependent on the cost object?
2. Distinguish product costs from period costs.
3. What two tests determine whether a cost is relevant?
4. Define the relevant range.
5. State one advantage and one limitation of regression analysis.

Long Answer Questions

1. Explain direct, indirect, fixed, variable, mixed and step costs, using examples.
2. Discuss why sunk costs and opportunity costs should be treated differently in decisions.
3. Compare account analysis, high–low and regression methods of cost estimation.
4. Explain the assumptions and limitations of the linear cost function.
5. Describe the cumulative-average learning-curve model and its managerial uses.

Analytical / Application Questions

1. A department incurs $52,000 at 10,000 machine hours and $37,000 at 5,000 hours. Estimate its variable rate, fixed cost and cost at 8,000 hours using high–low.
2. A special order uses spare capacity but requires special packaging and displaces a product contribution of $12,000. Identify the relevant and irrelevant cost categories that should be considered.
3. A regression model has a high R² but uses advertising expenditure as the only driver of maintenance cost. Evaluate whether management should accept the model without further analysis.

Multiple Choice Questions with Answers

Question 1

Which cost is normally a product cost in a manufacturing company?

A. Sales commission

B. Factory depreciation

C. Head-office advertising

D. Legal department salary

Correct Answer: B. Factory depreciation

Question 2

A cost is relevant to a decision when it is:

A. Past and allocated

B. Fixed in total

C. Future and different between alternatives

D. Included in the general ledger

Correct Answer: C. Future and different between alternatives

Question 3

Using the high–low observations of 5,000 hours and $30,000 and 9,000 hours and $42,000, the variable cost per hour is:

A. $2.00

B. $3.00

C. $4.00

D. $6.00

Correct Answer: B. $3.00

Question 4

In the cost function Y = a + bX, a represents:

A. Total fixed cost

B. Total variable cost

C. Activity level

D. Residual error

Correct Answer: A. Total fixed cost

Question 5

Which item is an opportunity cost?

A. The original purchase price of an existing machine

B. A depreciation charge already recorded

C. Contribution margin forgone by using capacity for another product

D. A cost incurred last year

Correct Answer: C. Contribution margin forgone by using capacity for another product

Question 6

The high–low method uses:

A. The highest and lowest cost amounts regardless of activity

B. The highest and lowest activity observations

C. Only the average activity observation

D. Only budgeted observations

Correct Answer: B. The highest and lowest activity observations

Question 7

A learning rate of 80% means that when cumulative output doubles, cumulative average time becomes:

A. 80% of its previous amount

B. 20% higher than its previous amount

C. Exactly 80 hours in every case

D. Unchanged

Correct Answer: A. 80% of its previous amount

Question 8

Which statement about fixed costs is correct?

A. Fixed cost per unit is constant at all volumes

B. Total fixed cost changes proportionally with activity

C. Total fixed cost is constant within a relevant range

D. Fixed costs are always irrelevant

Correct Answer: C. Total fixed cost is constant within a relevant range

Question 9

A high R² in a regression model proves that the selected driver causes the cost.

A. True because statistical fit proves causation

B. True only when the intercept is positive

C. False because fit does not by itself establish causation

D. False because regression cannot estimate costs

Correct Answer: C. False because fit does not by itself establish causation

Question 10

Which cost is most likely irrelevant to a short-term decision to use an existing machine?

A. Additional direct materials

B. Extra electricity

C. The machine’s original purchase price

D. Disposal value forgone

Correct Answer: C. The machine’s original purchase price