Source: Demski, Managerial Uses of Accounting Information, 2nd ed., Ch. 13–16, 18–19
Difficulty: Advanced | Prerequisites: Units 1 and 2 (Ch. 1–12 study notes); comfort with expected value, Bayes' theorem, and basic decision framing
The third pillar of AMIS 3300 turns the lens inward: once a decision has been made and executed, how do you evaluate whether the manager did a good job? And before the decision, how do you design incentives so that the manager wants to do a good job? This is where cost accounting meets agency theory, incentive design, and professional ethics. The unit moves from basic performance measures (Ch. 13) through conditional evaluation and multi‐task settings (Ch. 14–16) to coordination across divisions (Ch. 18) and professional responsibility (Ch. 19). If you are behind, make sure you are comfortable with uncertainty and expected‐value calculations from Ch. 9 before starting here.
Performance evaluation asks whether outcomes reflect skill, effort, or luck. Conditional evaluation filters out factors beyond the manager's control. Incentive design aligns the manager's interests with the firm's, but creates tension when the manager handles multiple tasks or holds private information. Accounting numbers are the primary input to these evaluations, which makes the cost‐measurement choices from Units 1 and 2 directly consequential for how people are judged and paid.
Controllability principle
A manager should be evaluated only on outcomes they can influence. In simple terms, do not blame (or credit) someone for things outside their control.
Variance analysis
The decomposition of the difference between actual and budgeted performance into component causes: price variance, efficiency variance, volume variance, mix variance. Each isolates one source of deviation.
Moral hazard
The situation where one party (the manager) can take unobservable actions (shirk, cut corners) that affect outcomes, and the other party (the firm) cannot directly verify the effort. Incentive contracts exist to mitigate this.
Adverse selection
The situation where the manager holds private information (about costs, demand, or their own ability) that the firm does not have. The firm must design contracts that induce truthful revelation.
Agency theory / principal-agent model
The formal framework for analysing moral hazard and adverse selection. The principal (firm) designs a contract; the agent (manager) chooses effort or reports information.
Conditional performance evaluation
Evaluating a manager's performance by conditioning on information about factors they could not control (e.g. market conditions, exchange rates). If the entire industry suffered, the manager's poor results may reflect the environment, not poor effort.
Relative performance evaluation (RPE)
Judging a manager's results against a peer group or benchmark (e.g. industry average). A form of conditional evaluation that filters out common shocks.
Multi-task problem
When a manager is responsible for several tasks (cost reduction and quality, or short‐term profit and long‐term investment), incentivising one may distort effort on the others.
Transfer price
The price at which one division of a firm sells a product or service to another division. Used for performance evaluation and coordination. Setting it wrong can lead divisions to make decisions that are good for the division but bad for the firm.
Implicit incentives
Motivators beyond the formal pay contract: career concerns, reputation, promotion prospects, intrinsic satisfaction. These can align or conflict with explicit incentives.
Professional responsibility
The ethical and professional obligations of accountants: objectivity, competence, confidentiality, integrity. Demski closes the course here to emphasise that the technical tools are only as good as the people using them.
The fundamental question
Did the manager do a good job? This seems simple, but the answer depends on what "good" means, what information is available, and how much of the outcome was within the manager's control.
Variance analysis as a diagnostic tool
Break the gap between budgeted and actual results into pieces: how much came from price changes, how much from efficiency changes, how much from volume changes.
Each variance tells a different story. A favourable price variance may mean smart purchasing or it may mean the purchasing manager bought inferior materials.
Taxes and incentives
Tax considerations affect both the firm's cost structure and the manager's incentives. A bonus paid in one form may be taxed differently from one paid in another. The firm's after‐tax cost of compensation feeds back into the performance‐evaluation system.
Problem 13‐14 specifically explores this intersection.
Why condition on external information?
A manager's results reflect both their actions and the environment. If the whole market dropped 20%, penalising a manager whose division dropped 15% is unjust and counterproductive.
The informativeness principle
Include a signal in the evaluation if and only if it provides information about the manager's action beyond what the outcome already tells you. In probabilistic terms, condition on a variable if it changes the likelihood ratio of the manager's effort given the observed outcome.
Relative performance evaluation in practice
Compare the manager's results to a peer benchmark. This strips out common shocks (industry downturns, exchange rate movements, commodity price swings) and isolates what the manager contributed.
Problem 14‐14 and Ralph's Responsibility Assignment
These assignments test your ability to compute conditional expectations and decide which signals are informative. The correction in the syllabus (Pr(c,r|H) should read Pr(c,r|L)) is a detail to note: the conditional probability table must match the correct state label.
The multi-task problem
When a manager is responsible for several tasks, strong incentives on one (say, cost reduction) can draw effort away from others (say, quality or innovation).
The optimal contract balances incentives across tasks. If one task is hard to measure, you may need to weaken incentives on the measurable task to avoid distorting effort.
Continuous vs. discrete performance measures
Accounting numbers are often discrete (quarterly earnings, annual budget variances). Continuous measures (stock price, real‐time production data) can provide richer information but may also be noisier.
Ralph's Task Balance
This problem requires you to think about how to allocate effort across tasks when the incentive weights differ. The core trade‐off: sharpening incentives on one dimension blurs them on another.
Using accounting numbers to evaluate managers
Most firms use accounting measures (divisional profit, return on investment, residual income) as the primary performance metric. These are available, verifiable, and understood by managers.
Limitations of accounting-based measures
Accounting earnings are backward‐looking, subject to allocation choices (the same issues from Unit 1), and can be manipulated by timing revenue recognition or deferring maintenance.
A manager evaluated on short‐term profit may under‐invest in R&D, training, or equipment maintenance because these reduce current earnings while their benefits arrive later.
Problem 16‐19 and the probability correction
Note the syllabus correction: the probability table labelled Pr(c,r|H) = [.15, .15, .35, .35] should read Pr(c,r|L). Getting the conditioning state right is essential; reversing H and L produces the wrong posterior and the wrong evaluation.
Why coordination is hard
In a decentralised firm, divisions make their own decisions. Each division manager optimises for their own unit, which may not be optimal for the firm as a whole.
Transfer pricing as a coordination tool
The transfer price determines how profit is split between a selling division and a buying division. If set too high, the buying division purchases less than the firm‐wide optimum. If set too low, the selling division under‐produces.
Common methods: market‐based (use the external market price), cost‐based (variable cost, full cost, cost‐plus), and negotiated.
Ralph's Implicit Incentives and Ralph's Excess (Part B)
These problems explore how non‐contractual incentives (reputation, career concerns) interact with the formal compensation system. A manager may work hard even without explicit pay‐for‐performance if promotion prospects depend on observed results.
Problem 18‐16
Tests your ability to analyse coordination in a setting where two divisions must agree on a transfer price and both have private information about their costs or demand.
The ethical dimension
The course closes by revisiting Chapter 1's themes: accounting aids management's stewardship responsibilities. The technical tools (cost measurement, decision analysis, performance evaluation) are means to an end. If the people using them lack integrity, the system fails.
Professional codes and standards
Accountants are bound by professional codes (IMA Statement of Ethical Professional Practice, AICPA Code of Professional Conduct) that require competence, confidentiality, integrity, and credibility.
Why this is in a cost accounting course
Every cost allocation, every performance evaluation, every incentive contract involves judgement. The person making those judgements has both the power to inform and the power to mislead. Demski's point: the accounting system is only as trustworthy as the people running it.
Problem 19‐8
Asks you to think through a scenario involving conflicting professional obligations. There is typically no clean answer; the point is to reason through the trade‐offs.
Price variance (materials)
(Actual price – Standard price) × Actual quantity purchased.
Efficiency variance (materials)
(Actual quantity used – Standard quantity allowed) × Standard price.
Spending variance (overhead)
Actual overhead – Budgeted overhead (at actual activity level).
Volume variance (overhead)
Budgeted overhead (at actual activity) – Applied overhead.
Residual income
Divisional profit – (Cost of capital × Divisional investment). Positive residual income means the division earns more than the firm's required return.
Return on investment (ROI)
ROI = Divisional profit / Divisional investment. Widely used, but a manager can improve ROI by rejecting projects that earn above the cost of capital but below the division's current ROI.
Informativeness condition (conceptual)
A signal is informative about the manager's effort if its conditional probability differs across effort levels. Formally, include variable y in the evaluation if P(outcome | effort, y) differs from P(outcome | effort) for at least some values.
Variance analysis is the daily language of manufacturing management: plant managers review price and efficiency variances at the start of every shift in many factories. Relative performance evaluation is standard in executive compensation (CEO bonuses benchmarked against industry peers). The multi‐task problem explains why teachers evaluated solely on test scores may narrow the curriculum: measurable outcomes crowd out unmeasurable ones. Transfer pricing is a perennial headache for multinational firms, where it intersects with tax planning across jurisdictions.
Students assume a favourable variance is always good. A favourable price variance from buying cheap materials may cause an unfavourable efficiency variance from higher waste rates. Variances interact.
Students think the controllability principle is straightforward to apply. In practice, almost no outcome is entirely within or entirely outside a manager's control. The question is always one of degree.
Students equate higher‐powered incentives with better incentives. Stronger incentives on measurable tasks distort effort away from unmeasurable ones. The optimal contract often looks weaker than students expect.
Students overlook that accounting numbers used for evaluation are the same numbers the manager helps produce. A manager evaluated on divisional profit has both the motive and the means to influence the measurement. This is not a bug; it is the central tension of accounting‐based evaluation.
⚠️ The final exam is comprehensive and cumulative (20% of the grade). Expect problems requiring variance decomposition, conditional evaluation calculations, and short essays on incentive design.
⚠️ Be able to decompose a total variance into price and efficiency components for materials and labour, and into spending and volume components for overhead.
⚠️ Understand the informativeness principle well enough to explain, in a short essay, why a particular signal should (or should not) be included in a manager's evaluation.
⚠️ Transfer pricing questions may ask you to identify the firm‐wide optimal outcome and show that a particular transfer price leads divisions to choose differently.
⚠️ The professional responsibility material (Ch. 19) is assigned alongside a re‐reading of Ch. 1. This framing choice is deliberate: the exam may ask you to connect the ethical dimension back to the course's opening themes.
True or false: A manager should always be held responsible for unfavourable variances in their division. (False. The controllability principle says evaluate only what the manager can influence.)
Fill in the blank: Residual income equals divisional profit minus _______ times _______. (Cost of capital; divisional investment.)
True or false: A market‐based transfer price always leads to firm‐wide optimal decisions. (False. It works well when there is a competitive external market, but fails when the market is thin or when internal trade has different cost characteristics.)
Fill in the blank: The multi‐task problem arises when strong incentives on one task _______ effort on other tasks. (Distort / crowd out.)
True or false: A signal that is correlated with outcomes but uncorrelated with the manager's effort should be included in the evaluation. (False. The informativeness principle requires the signal to carry information about effort, not just about outcomes.)
Q: A division budgeted 10,000 units of material at $5 each. It actually used 11,000 units at $4.80 each. Compute the price and efficiency variances.
A: Price variance = ($4.80 – $5.00) × 11,000 = –$2,200 (favourable, since actual price was lower). Efficiency variance = (11,000 – 10,000) × $5.00 = $5,000 (unfavourable, since more material was used than budgeted).
Q: Why might a favourable price variance and an unfavourable efficiency variance occur together?
A: The purchasing manager bought cheaper materials, which were lower quality, leading to more waste and rework in production. The price saving was more than offset by the higher usage.
Q: Division A can sell a component externally for $50. Its variable cost is $30. Division B needs the component and can buy it externally for $48. What transfer price range works for both divisions?
A: Division A will not sell internally for less than $50 (its opportunity cost from the external sale). Division B will not pay more than $48 (its external purchase option). There is no mutually acceptable range, so the internal transfer should not happen: Division A sells externally at $50 and Division B buys externally at $48. The firm is better off.
Q: Explain in two sentences why ROI can lead a manager to reject a value‐creating project.
A: If the division's current ROI is 20% and a new project earns 15%, accepting it lowers the division's ROI. Yet if the firm's cost of capital is 10%, the project creates value (residual income is positive). The manager's incentive to protect their ROI metric conflicts with the firm's interest.
Q: What does Demski mean by returning to Chapter 1 at the end of the course?
A: The course opened with the premise that accounting aids management's stewardship of resources. After studying how costs are measured, decisions are made, and performance is evaluated, the closing message is that these tools require professional judgement and ethical conduct. The system's integrity depends on the people operating it.
Performance evaluation depends on the cost numbers produced in Unit 1: if overhead is allocated arbitrarily, the variance analysis built on those allocations is equally arbitrary. The uncertainty framework from Ch. 9 underpins conditional evaluation (filtering out luck) and the principal‐agent model (the manager's effort is uncertain from the firm's perspective). The strategic framing of Ch. 10 connects to coordination (Ch. 18): divisional managers are strategic actors who respond to the incentive structures they face, just as competitors respond to pricing.
Performance evaluation, variance analysis, price variance, efficiency variance, volume variance, spending variance, controllability principle, moral hazard, adverse selection, agency theory, principal-agent, incentive contract, conditional performance evaluation, relative performance evaluation, RPE, informativeness principle, multi-task problem, transfer pricing, market-based transfer price, cost-based transfer price, residual income, return on investment, ROI, divisional performance, coordination, decentralisation, implicit incentives, career concerns, professional responsibility, ethics, IMA, AICPA, stewardship, AMIS 3300, Demski, cost accounting, managerial accounting