INTRODUCTION TO ANALYTICAL RESEARCH IN ACCOUNTING

Transcription

1 INTRODUCTION TO ANALYTICAL RESEARCH IN ACCOUNTING Christian Hofmann LMU Munich 1. USE OF MODELS IN INFORMATION EVALUATION (FELTHAM/DEMSKI 1970) 2. ACCOUNTING FOR STEWARDSHIP OR DECISION-MAKING (GJESDAL 1981) 3. CONTRACTING THEORY AND ACCOUNTING (LAMBERT 2001)

2 Feltham, G.A. and J.S. Demski (1970) The Use of Models in Information Evaluation The Accounting Review 45 (4), C. Hofmann - Analytical Research in Accounting - 2

3 GENERAL PROBLEM What information should be supplied to a particular decision maker in a particular decision context? Accounting information = consumption good. Basic economic calculus: Marginal cost of producing accounting information = marginal benefit of using accounting information C. Hofmann - Analytical Research in Accounting - 1

4 GENERAL FRAMEWORK OF THE ANALYSIS Designing the supply of information requires a decision setting under uncertainty. decision theory - decision making under uncertainty Information system choice depends on the use of the information: Demski/Feltham decision-facilitating role decision-influencing role Gjesdal decision-making demand stewardship demand ex ante-information ex post-information Examples C. Hofmann - Analytical Research in Accounting - 2

5 PARTICIPANTS Decision maker - who uses the information for decision making Decision: a 0 A Information evaluator - who decides what information to supply Events: x 0 X Preferences: u P = w P (x) Prediction of future events: φ P (x a) Expected payoff: E[u P a] = 3 x0x w P (x) φ P (x a) C. Hofmann - Analytical Research in Accounting - 3

6 REPRESENTATION OF INFORMATION Information system η signal y 0 Y with probability φ P (y η) Conditional probability Prediction of future events: φ P (x y,η,a) Note: probability distributions must be consistent C. Hofmann - Analytical Research in Accounting - 4

7 SEQUENTIAL INFORMATION EVALUATION PROCESS 1. Specification of a particular information system η by the evaluator 2. Supply of signals y to the decision maker 3. Use of the signal by the decision maker in selecting his action a(y) 4. Action a(y) determines the events of the subsequent period Y Optimal information system by backward induction C. Hofmann - Analytical Research in Accounting - 5

8 DECISION RULES Relationship between signal and action? Information evaluator = decision maker E * [u P y,η] = max a0a 3 x0x w P (x) φ P (x y,η,a) Y a = α(y,η) Information evaluator s choice of signal-dependent actions Information evaluator decision maker E [u A y,η] = max a0 3 x0x Y a = α(y,η) Information evaluator s prediction of the results of the decision maker s model C. Hofmann - Analytical Research in Accounting - 6

9 INFORMATION SYSTEM SELECTION Selection of information system = choice under uncertainty signal y 0 Y is uncertain Expected net payoff for a given information system η E[u P η] = 3 y0y {3 x0x w P (x) φ P (x y,η,α(y,η))! wn(y,η)} φ P (y η) Choice among information systems η 1 and η 2 η 1 η 2 ] E[u P η 1 ] > E[u P η 2 ] Marginal value of system η 2 V(η 2 η 1 ) = E[u P η 2 ]! E[u P η 1 ] C. Hofmann - Analytical Research in Accounting - 7

10 APPLICATION OF THE ANALYSIS Information evaluator Accountant Firm owner Regulator Tax authority Decision maker C Manager C Investor C Investor C Taxpayer Information systems Segment information Variance analysis Alternative opportunity costs Customer satisfaction, scrap rate C Marginal product cost C Inventory valuation C Depreciation Y Generic results with limited reference to accounting information C. Hofmann - Analytical Research in Accounting - 8

11 Gjesdal, F. (1981) Accounting for Stewardship Journal of Accounting Research 19 (1), C. Hofmann - Analytical Research in Accounting - 9

12 OBJECTIVES OF FINANCIAL REPORTING Decision-making demand versus stewardship demand for (accounting) information. Financial statements may be of value... < for making investment decisions. < for controlling the allocation of resources by managers. Ranking of information systems, e.g., < Cashflow- versus accrual accounting (matching of revenues and expenses) < IAS 11 - Percentage-of-Completion vs. Completed contracts < IAS 38 - intangible assets C. Hofmann - Analytical Research in Accounting - 10

13 SIMPLE MAPPING EXERCISE # Valuing assets at historical cost, replacement cost, or current value? # Expense or capitalize R&D-investments? # Expense or capitalize investments to increase brand awareness? # How to calculate pension obligations? C. Hofmann - Analytical Research in Accounting - 11

14 DIFFERENT COSTS FOR DIFFERENT PURPOSES? <... the stewardship objective implies a preference for historical cost over current value alternatives... (Rosenfield 1971) < If accounting for stewardship is the first purpose... emphasis on completed transactions is appropriate. Conversely future-oriented statements are less appropriate (Financial Executives Research Foundation 1973) <... measures useful for assessing earnings power are also useful for holding management to account. (AICPA 1973) < Historical cost valuation provides data that are less disputable than data provided under other valuation methods currently being proposed, an essential reuirement in equity accounting. (Ijiri 1971) C. Hofmann - Analytical Research in Accounting - 12

15 THE AGENCY INFORMATION PROBLEM - ASSUMPTIONS Individual i 0 I action a i 0 A i outcome x i (θ,a), where a = (a 1,..., a I ) state variable θ 0 Θ preference u i (x i,a i ) Information system Ex post signal y(θ,a,η), with η 0 H Pareto optimality An allocation of goods is Pareto optimal when no further Pareto improvements can be made. A movement from one allocation to another that can make at least one individual better off, without making any other individual worse off, is called a Pareto improvement. C. Hofmann - Analytical Research in Accounting - 13

16 THE AGENCY INFORMATION PROBLEM choose (s *, a *, η * ) to be Pareto-efficient subject to s(y) = {s 1 (y),..., s I (y)} 0 S with 3 i s i (y) = 0, for all y η * 0 H a * = (a 1 *,..., a I * ) 0 A a i * 0 argmax a i 0 A E[u i (x i (θ,a) + s i * (y)], where a j = a j * for j i C. Hofmann - Analytical Research in Accounting - 14

17 KEY FEATURES OF THE AGENCY INFORMATION PROBLEM Actions are chosen non-cooperatively - a * is a Nash equilibrium Information system and sharing rule are chosen cooperatively Decentralization of decision making is key characteristic of stewardship demand for accounting information. Value of information - the agency is willing to pay for it (a) action-informativeness: M f(y,a) / M a / 0 is not true for a = a 0 (b) insurance-informativeness: Cov[x(θ,a 0 ),y(θ,a 0 )] 0 There is a stewardship demand for information because the use of such information is Pareto-optimal for the organization. C. Hofmann - Analytical Research in Accounting - 15

18 COMPARISONS OF INFORMATION SYSTEMS Proposition 2: Information system η 1 is weakly preferred to η 2 for any agency information problem if η 2 is a garbling of η 1. Example: η 1 : y 1 = a + g 1, with g 1 ~ N(0,σ 2 ) η 2 : y 2 = y 1 + g 2, with g 2 ~ N(0,σ 2 ) C. Hofmann - Analytical Research in Accounting - 16

19 HARDNESS REQUIREMENT A hard measure is one constructed in such a way that it is difficult for people to disagree. (Ijiri 1975) η 1 : agent privately learns the signal y - compensation depends on m η 2 : principal and agent observe y Pareto-optimal information system: η 2 C. Hofmann - Analytical Research in Accounting - 17

20 COMPARISONS OF INFORMATION SYSTEMS Corollary: The incentive (stewardship) objective and the decision objective do not give rise to identical rankings. Practical interest: < Do the objectives rank proposed accounting alternatives differently? < Compare a given a set of information systems. C. Hofmann - Analytical Research in Accounting - 18

21 Lambert, R.A. (2001) Contracting Theory and Accounting Journal of Accounting and Economics 32, C. Hofmann - Analytical Research in Accounting - 19

22 AGENCY THEORY Motivation for accounting and auditing has to do with the control of incentive problems. Key questions: (a) How do features of information, accounting, and compensation systems affect incentive problems? (b) How does the existence of incentive problems affect the design and structure of information, accounting, and compensation systems? C. Hofmann - Analytical Research in Accounting - 20

23 INFORMATION ECONOMICS The most meaningful way to compare accounting/performance measurement systems is by comparing each system when it is used optimally. For example, in order for there to be a role for additional accounting information, it must be the case that the incentive problems being studied cannot be completely resolved via other means. Y What conditions are necessary for a non-trivial incentive problem? Reasons for conflicts of interest (i) effort aversion by the agent (ii) agent can divert resources for his private consumption (iii) differential time horizons (iv) differential risk aversion C. Hofmann - Analytical Research in Accounting - 21

24 PRODUCTION OF INFORMATION SIGNALS 1. Signals are generated by the actions and automatically observed by the parties. 2. Principal decides whether to conduct an investigation to obtain more information (variance analysis - informed principal at a cost). 3. Agent privately observes performance and reports it to the principal (information acquisition - informed agent at a cost). 4. Information production by third party (auditor - independence and audit effort). 5. Security market aggregates information into stock price. 6. Product market aggregates information into product prices. 7. Labor market aggregates information into outside employment opportunities. C. Hofmann - Analytical Research in Accounting - 22

25 CATEGORIZATION OF PRINCIPAL/AGENT-SETTINGS # of performance measures - single report versus multiple reports settings # of tasks - single-task versus multi-task settings # of periods - single-period versus multi-period settings # of agents - single-agent versus multi-agent settings # of principals - single-principal versus multi-principal settings Information asymmetry pre-contract information asymmetry post-contract, pre-decision information asymmetry post-decision information asymmetry C. Hofmann - Analytical Research in Accounting - 23

26 SINGLE-ACTION AGENCY MODELS Principal s choice problem max s(x,y),a E[u P (x! s(x,y))] subject to E[u A (s(x,y),a)] $ U A (participation constraint, agent s acceptable utility constraint) E[u A (s(x,y),a)] $ E[u A (s(x,y),an)] for all an 0 A (incentive compatibility constraint) Characterization of the optimal contract C. Hofmann - Analytical Research in Accounting - 24

27 SELECTED RESEARCH QUESTIONS Incentives and agency cost: < Contract choice trades-off risk sharing and incentives to provide effort. Value of addition performance measures: < Any performance measure that is (conditionally) informative about the agent s action is valuable. Aggregation of performance measures: < Linear aggregation of performance measures is optimal if the performance measure belongs to the exponential family of distributions. < Relative aggregation weight is proportional to the signal-to-noise ratio. Relative performance evaluation / controllability principle: < Compare agent s performance with the performance of a peer group. < While peer performance is not controllable, it may be informative. C. Hofmann - Analytical Research in Accounting - 25

28 MULTI-TASK PRINCIPAL/AGENT-SETTINGS Congruency between performance measure and payoff to the principal Y Motivating the right effort allocation Issues in multi-task principal/agent-settings: Window dressing x = b 1 a 1 + g x versus y = m 1 a 1 + m 2 a 2 + g y Myopic performance measures x = b 1 a 1 + b 2 a 2 + g x versus y = m 1 a 1 + m 2 a 2 + g y, with m 1 / m 2 b 1 / b 2 Stewardship versus valuation uses of information Valuation: correlation between the performance measure and the outcome. Stewardship: signal-to-noise ratio C. Hofmann - Analytical Research in Accounting - 26

29 PRIVATE INFORMATION AND COMMUNICATION Timing of acquiring private information by the agent < before signing the contract < after signing the contract but before selecting actions < after selecting actions Can the agent leave the firm after observing the information signal? Can the agent communicate the signal to the principal? C. Hofmann - Analytical Research in Accounting - 27

30 CONSEQUENCES OF PRIVATE INFORMATION Selection of an action strategy a = α(y) or a = α(m) Revelation principle: The cost of motivating the truth is no greater than the cost of motivating a nontruthful reporting strategy. Extension of the agency model by considering truthtelling-constraints C. Hofmann - Analytical Research in Accounting - 28

31 APPLICATIONS OF MODELS WITH PRIVATE INFORMATION Capital budgeting Charge for capital Transfer pricing Cost allocation C. Hofmann - Analytical Research in Accounting - 29