A Model of Principles-Based vs. Rules-Based Standards

Transcription

1 A Model of Principles-Based vs. Rules-Based Standards Pingyang Gao, Haresh Sapra, and Hao Xue Minnesota Seminar April 2016

2 Introduction Rules-based standards: rely on bright-line and quantifiable evidence.

3 Introduction Rules-based standards: rely on bright-line and quantifiable evidence. Induce transaction structuring.

4 Introduction Rules-based standards: rely on bright-line and quantifiable evidence. Induce transaction structuring. Principles-based standards: rely on management s professional judgement.

5 Introduction Rules-based standards: rely on bright-line and quantifiable evidence. Induce transaction structuring. Principles-based standards: rely on management s professional judgement. Induce abuse of discretion.

6 Environment Insiders would like to paint a rosy picture of the firm:

7 Environment Insiders would like to paint a rosy picture of the firm: By tampering with the evidence and/or

8 Environment Insiders would like to paint a rosy picture of the firm: By tampering with the evidence and/or By abusing their professional judgement.

9 Environment Insiders would like to paint a rosy picture of the firm: By tampering with the evidence and/or By abusing their professional judgement. Given potential evidence management and abuse of discretion:

10 Environment Insiders would like to paint a rosy picture of the firm: By tampering with the evidence and/or By abusing their professional judgement. Given potential evidence management and abuse of discretion: How should a standard setter design the optimal standard?

11 Environment Insiders would like to paint a rosy picture of the firm: By tampering with the evidence and/or By abusing their professional judgement. Given potential evidence management and abuse of discretion: How should a standard setter design the optimal standard? What is its shape?

12 Environment Insiders would like to paint a rosy picture of the firm: By tampering with the evidence and/or By abusing their professional judgement. Given potential evidence management and abuse of discretion: How should a standard setter design the optimal standard? What is its shape? How does it depend on various features of firm s environment?

13 Main Insights Optimal standard combines rules-based and principles-based elements.

14 Main Insights Optimal standard combines rules-based and principles-based elements. It takes a simple and intuitive form:

15 Main Insights Optimal standard combines rules-based and principles-based elements. It takes a simple and intuitive form: If evidence is sufficiently favorable, rely on professional judgement. Otherwise, use a strict rule.

16 Main Insights Optimal standard combines rules-based and principles-based elements. It takes a simple and intuitive form: If evidence is sufficiently favorable, rely on professional judgement. Otherwise, use a strict rule. Its properties depend on various features of firm s environment:

17 Main Insights Optimal standard combines rules-based and principles-based elements. It takes a simple and intuitive form: If evidence is sufficiently favorable, rely on professional judgement. Otherwise, use a strict rule. Its properties depend on various features of firm s environment: Effectiveness of regulatory enforcement of standard.

18 Main Insights Optimal standard combines rules-based and principles-based elements. It takes a simple and intuitive form: If evidence is sufficiently favorable, rely on professional judgement. Otherwise, use a strict rule. Its properties depend on various features of firm s environment: Effectiveness of regulatory enforcement of standard. Severity of conflict between insiders and outsiders.

19 Main Insights Optimal standard combines rules-based and principles-based elements. It takes a simple and intuitive form: If evidence is sufficiently favorable, rely on professional judgement. Otherwise, use a strict rule. Its properties depend on various features of firm s environment: Effectiveness of regulatory enforcement of standard. Severity of conflict between insiders and outsiders. Nature of the transaction.

20 Timing of Events 1 Standard setter designs standard S to measure ω, the economic substance of a transaction:

21 Timing of Events 1 Standard setter designs standard S to measure ω, the economic substance of a transaction: ω {G, B} with probabilities q H and q B = 1 q H.

22 Timing of Events 1 Standard setter designs standard S to measure ω, the economic substance of a transaction: ω {G, B} with probabilities q H and q B = 1 q H. 2 Manager privately observes initial quantifiable evidence t R about ω:

23 Timing of Events 1 Standard setter designs standard S to measure ω, the economic substance of a transaction: ω {G, B} with probabilities q H and q B = 1 q H. 2 Manager privately observes initial quantifiable evidence t R about ω: t f ω ( ) and a large value of t is good news in the sense of MLRP.

24 Timing of Events 1 Standard setter designs standard S to measure ω, the economic substance of a transaction: ω {G, B} with probabilities q H and q B = 1 q H. 2 Manager privately observes initial quantifiable evidence t R about ω: t f ω ( ) and a large value of t is good news in the sense of MLRP. Manager chooses whether to privately manipulate evidence from t to t m = t + m at a cost C (m).

25 Timing of Events 1 Standard setter designs standard S to measure ω, the economic substance of a transaction: ω {G, B} with probabilities q H and q B = 1 q H. 2 Manager privately observes initial quantifiable evidence t R about ω: t f ω ( ) and a large value of t is good news in the sense of MLRP. Manager chooses whether to privately manipulate evidence from t to t m = t + m at a cost C (m). Evidence t m could be an input for the standard.

26 Timing of Events (continued) 3. The manager observes additional non-quantifiable information about ω:

27 Timing of Events (continued) 3. The manager observes additional non-quantifiable information about ω: She perfectly learns ω.

28 Timing of Events (continued) 3. The manager observes additional non-quantifiable information about ω: She perfectly learns ω. However, unlike quantifiable evidence t m, her knowledge of ω cannot be directly prescribed in the standard.

29 Timing of Events (continued) 3. The manager observes additional non-quantifiable information about ω: She perfectly learns ω. However, unlike quantifiable evidence t m, her knowledge of ω cannot be directly prescribed in the standard. If standard calls for professional judgement:

30 Timing of Events (continued) 3. The manager observes additional non-quantifiable information about ω: She perfectly learns ω. However, unlike quantifiable evidence t m, her knowledge of ω cannot be directly prescribed in the standard. If standard calls for professional judgement: Manager may abuse her discretion at a private cost of φ > 0.

31 Timing of Events (continued) 3. The manager observes additional non-quantifiable information about ω: She perfectly learns ω. However, unlike quantifiable evidence t m, her knowledge of ω cannot be directly prescribed in the standard. If standard calls for professional judgement: Manager may abuse her discretion at a private cost of φ > 0. φ has a cumulative density function K (φ; τ).

32 Timing of Events (continued) 3. The manager observes additional non-quantifiable information about ω: She perfectly learns ω. However, unlike quantifiable evidence t m, her knowledge of ω cannot be directly prescribed in the standard. If standard calls for professional judgement: Manager may abuse her discretion at a private cost of φ > 0. φ has a cumulative density function K (φ; τ). K τ (φ; τ) < 0: τ captures effectiveness of regulatory enforcement.

33 Timing of Events (continued) 3. The manager observes additional non-quantifiable information about ω: She perfectly learns ω. However, unlike quantifiable evidence t m, her knowledge of ω cannot be directly prescribed in the standard. If standard calls for professional judgement: Manager may abuse her discretion at a private cost of φ > 0. φ has a cumulative density function K (φ; τ). K τ (φ; τ) < 0: τ captures effectiveness of regulatory enforcement. Based on prevailing accounting standard, report r {b, g}.

34 Timing of Events (continued) 3. The manager observes additional non-quantifiable information about ω: She perfectly learns ω. However, unlike quantifiable evidence t m, her knowledge of ω cannot be directly prescribed in the standard. If standard calls for professional judgement: Manager may abuse her discretion at a private cost of φ > 0. φ has a cumulative density function K (φ; τ). K τ (φ; τ) < 0: τ captures effectiveness of regulatory enforcement. Based on prevailing accounting standard, report r {b, g}. 4. ω becomes common knowledge and payoffs are realized.

35 Preferences Standard setter designs standard S to minimize q G Pr(r = b ω = G ) L }{{ G } expected cost of false alarm + q B Pr(r = g ω = B) L B }{{} expected cost of undue optimism where S determines Pr(r = b ω = G ) and Pr(r = g ω = B).

36 Preferences Standard setter designs standard S to minimize q G Pr(r = b ω = G ) L }{{ G } expected cost of false alarm + q B Pr(r = g ω = B) L B }{{} expected cost of undue optimism where S determines Pr(r = b ω = G ) and Pr(r = g ω = B). Firm s manager always prefers the favorable report, r = g.

37 Preferences Standard setter designs standard S to minimize q G Pr(r = b ω = G ) L }{{ G } expected cost of false alarm + q B Pr(r = g ω = B) L B }{{} expected cost of undue optimism where S determines Pr(r = b ω = G ) and Pr(r = g ω = B). Firm s manager always prefers the favorable report, r = g. Conflict of interest captured by δ > 0.

38 Preferences Standard setter designs standard S to minimize q G Pr(r = b ω = G ) L }{{ G } expected cost of false alarm + q B Pr(r = g ω = B) L B }{{} expected cost of undue optimism where S determines Pr(r = b ω = G ) and Pr(r = g ω = B). Firm s manager always prefers the favorable report, r = g. Conflict of interest captured by δ > 0. Induces evidence management m and/or abuse of discretion at a cost of φ.

39 An Example: Revenue Recognition Has the majority of risks and rewards associated with products been transferred from firm to the buyer?

40 An Example: Revenue Recognition Has the majority of risks and rewards associated with products been transferred from firm to the buyer? A rules-based standard would rely on quantifiable evidence such as product shipment.

41 An Example: Revenue Recognition Has the majority of risks and rewards associated with products been transferred from firm to the buyer? A rules-based standard would rely on quantifiable evidence such as product shipment. A principles-based standard would rely on the management s professional judgement.

42 An Example: Revenue Recognition Has the majority of risks and rewards associated with products been transferred from firm to the buyer? A rules-based standard would rely on quantifiable evidence such as product shipment. A principles-based standard would rely on the management s professional judgement. In general, a standard could incorporate both quantifiable evidence and professional judgement.

43 An Example: Revenue Recognition Has the majority of risks and rewards associated with products been transferred from firm to the buyer? A rules-based standard would rely on quantifiable evidence such as product shipment. A principles-based standard would rely on the management s professional judgement. In general, a standard could incorporate both quantifiable evidence and professional judgement. Suppose manager obtains private evidence t about an anticipated product shipment:

44 An Example: Revenue Recognition Has the majority of risks and rewards associated with products been transferred from firm to the buyer? A rules-based standard would rely on quantifiable evidence such as product shipment. A principles-based standard would rely on the management s professional judgement. In general, a standard could incorporate both quantifiable evidence and professional judgement. Suppose manager obtains private evidence t about an anticipated product shipment: By engaging in channel stuffing, she can expedite delivery, i.e., convert t to t m.

45 An Example: Revenue Recognition Has the majority of risks and rewards associated with products been transferred from firm to the buyer? A rules-based standard would rely on quantifiable evidence such as product shipment. A principles-based standard would rely on the management s professional judgement. In general, a standard could incorporate both quantifiable evidence and professional judgement. Suppose manager obtains private evidence t about an anticipated product shipment: By engaging in channel stuffing, she can expedite delivery, i.e., convert t to t m. By abusing her discretion, she can also prematurely recognize revenue as long as δ φ.

46 An Example: Revenue Recognition Has the majority of risks and rewards associated with products been transferred from firm to the buyer? A rules-based standard would rely on quantifiable evidence such as product shipment. A principles-based standard would rely on the management s professional judgement. In general, a standard could incorporate both quantifiable evidence and professional judgement. Suppose manager obtains private evidence t about an anticipated product shipment: By engaging in channel stuffing, she can expedite delivery, i.e., convert t to t m. By abusing her discretion, she can also prematurely recognize revenue as long as δ φ. The distribution of φ depends on τ.

47 Reporting Standards S is defined as a mapping S(t m ) {b, g, p}

48 Reporting Standards S is defined as a mapping A Rules-Based Standard S(t m ) {b, g, p} S(t m ) = g if t m > T, b if t m T.

49 Reporting Standards S is defined as a mapping A Rules-Based Standard A Principles-Based Standard S(t m ) {b, g, p} S(t m ) = g if t m > T, b if t m T. S(t m ) = p t m

50 Reporting Standards S is defined as a mapping A Rules-Based Standard A Principles-Based Standard A Hybrid Standard S(t m ) {b, g, p} S(t m ) = g if t m > T, b if t m T. S(t m ) = p t m S(t m ) = g if t m > T 2, p if t m [T 1, T 2 ], b if t m < T 1.

51 Equilibrium Given S, t m, ω, and φ: manager chooses her optimal reporting strategy.

52 Equilibrium Given S, t m, ω, and φ: manager chooses her optimal reporting strategy. Given S, initial evidence t, and optimal reporting strategy, manager chooses her optimal evidence management strategy.

53 Equilibrium Given S, t m, ω, and φ: manager chooses her optimal reporting strategy. Given S, initial evidence t, and optimal reporting strategy, manager chooses her optimal evidence management strategy. Given manager s optimal evidence management and reporting strategies, standard setter chooses the optimal standard, S (t m ).

54 Equilibrium Given S, t m, ω, and φ: manager chooses her optimal reporting strategy. Given S, initial evidence t, and optimal reporting strategy, manager chooses her optimal evidence management strategy. Given manager s optimal evidence management and reporting strategies, standard setter chooses the optimal standard, S (t m ). All optimal decisions are consistent with each other in the sense of rational expectations.

55 Optimal Standard Optimal standard S (t m ) is fully characterized by a unique threshold T : S (t m ) = p if t m > T b if t m T where T will be characterized later.

56 Intuition for Shape of Optimal Standard For sufficiently positive evidence, requiring professional judgement eliminates false alarm errors. But, it also minimizes undue optimism errors for two reasons:

57 Intuition for Shape of Optimal Standard For sufficiently positive evidence, requiring professional judgement eliminates false alarm errors. But, it also minimizes undue optimism errors for two reasons: 1 MLRP implies there are few low types to begin with and those low types still need to exercise professional judgement.

58 Intuition for Shape of Optimal Standard For sufficiently positive evidence, requiring professional judgement eliminates false alarm errors. But, it also minimizes undue optimism errors for two reasons: 1 MLRP implies there are few low types to begin with and those low types still need to exercise professional judgement. 2 This, in turn, dampens the incentives for evidence management.

59 Intuition for Shape of Optimal Standard For sufficiently positive evidence, requiring professional judgement eliminates false alarm errors. But, it also minimizes undue optimism errors for two reasons: 1 MLRP implies there are few low types to begin with and those low types still need to exercise professional judgement. 2 This, in turn, dampens the incentives for evidence management. For sufficiently negative evidence, applying a strict rule (unfavorable treatment) minimizes undue optimism errors. While, false alarm errors are also low, they may not be trivial but...

60 Intuition for Shape of Optimal Standard For sufficiently positive evidence, requiring professional judgement eliminates false alarm errors. But, it also minimizes undue optimism errors for two reasons: 1 MLRP implies there are few low types to begin with and those low types still need to exercise professional judgement. 2 This, in turn, dampens the incentives for evidence management. For sufficiently negative evidence, applying a strict rule (unfavorable treatment) minimizes undue optimism errors. While, false alarm errors are also low, they may not be trivial but... they can be controlled via T and optimized against undue optimism errors.

61 Implications If evidence is sufficiently favorable, rely on professional judgement. Otherwise, use a strict rule.

62 Implications If evidence is sufficiently favorable, rely on professional judgement. Otherwise, use a strict rule. Higher hurdle for favorable treatment relative to unfavorable treatment.

63 Equilibrium Reporting Decisions r (t m < T, ω, φ) = r (t m T, B, φ δ) = b, r (t m T, G, φ) = r (t m T, B, φ < δ) = g.

64 Equilibrium Reporting Decisions r (t m < T, ω, φ) = r (t m T, B, φ δ) = b, r (t m T, G, φ) = r (t m T, B, φ < δ) = g. Ex ante (i.e., before φ is realized) probability that the manager abuses discretion is K (φ ; τ) where φ < φ = δ.

65 Evidence Management For any t < T, expected incremental benefit is φ (t) = Pr(ω = G t)δ + Pr(ω = B t) (δ φ)dk (φ; τ) 0

66 Evidence Management For any t < T, expected incremental benefit is φ (t) = Pr(ω = G t)δ + Pr(ω = B t) (δ φ)dk (φ; τ) 0 while the incremental cost is C (T t)

67 Evidence Management For any t < T, expected incremental benefit is φ (t) = Pr(ω = G t)δ + Pr(ω = B t) (δ φ)dk (φ; τ) 0 while the incremental cost is C (T t) Evidence management strategy characterized by a unique threshold T (T ): ( T (T )) C (T T (T )) = 0

68 Evidence Management For any t < T, expected incremental benefit is φ (t) = Pr(ω = G t)δ + Pr(ω = B t) (δ φ)dk (φ; τ) 0 while the incremental cost is C (T t) Evidence management strategy characterized by a unique threshold T (T ): ( T (T )) C (T T (T )) = 0 The manager s optimal evidence management strategy is: m (t; T ) = T t if t ( ˆT (T ), T ) 0 otherwise.

69 Optimal Threshold At t = 1, standard setter chooses T to minimize L = q G L G ˆT (T ) where ˆT (T ) satisfies f G (x)dx + q B L B K (φ ; τ) f B (x)dx ˆT (T ) ( ˆT (T )) C (T ˆT (T )) = 0.

70 Optimal Threshold Optimal threshold T solves: ( ) (q G L G f G ( ˆT (T ) q B L B K (φ ˆT ; τ)f B (T ) ( ˆT (T )) = 0 T where ˆT (T ) T > 0.

71 Properties of Optimal Standard Optimal standard relies more on management s professional judgement:

72 Properties of Optimal Standard Optimal standard relies more on management s professional judgement: The more effective the enforcement of the standard: dt dτ < 0.

73 Properties of Optimal Standard Optimal standard relies more on management s professional judgement: The more effective the enforcement of the standard: dt dτ < 0. The less severe the conflict between insiders and outsiders: dt dδ > 0.

74 Properties of Optimal Standard Optimal standard relies more on management s professional judgement: The more effective the enforcement of the standard: dt dτ < 0. The less severe the conflict between insiders and outsiders: dt dδ > 0. The more costly evidence management is: dt dc < 0.

75 Properties of Optimal Standard Optimal standard relies more on management s professional judgement: The more effective the enforcement of the standard: dt dτ < 0. The less severe the conflict between insiders and outsiders: dt dδ > 0. The more costly evidence management is: dt dc < 0. The lower (higher) the cost of undue optimism error (false alarm error): dt dl B > 0 and dt dl G < 0.

76 Conclusion In a second best environment, we developed a model to analyze the trade-off between principles-based vs. rules-based elements.

77 Conclusion In a second best environment, we developed a model to analyze the trade-off between principles-based vs. rules-based elements. We show that optimal standard takes a simple form:

78 Conclusion In a second best environment, we developed a model to analyze the trade-off between principles-based vs. rules-based elements. We show that optimal standard takes a simple form: To get favorable treatment, need both professional judgement and favorable evidence. But for unfavorable evidence, apply a strict rule.

79 Conclusion In a second best environment, we developed a model to analyze the trade-off between principles-based vs. rules-based elements. We show that optimal standard takes a simple form: To get favorable treatment, need both professional judgement and favorable evidence. But for unfavorable evidence, apply a strict rule. We simplified the enforcement mechanism and captured it by the exogenous cost φ :

80 Conclusion In a second best environment, we developed a model to analyze the trade-off between principles-based vs. rules-based elements. We show that optimal standard takes a simple form: To get favorable treatment, need both professional judgement and favorable evidence. But for unfavorable evidence, apply a strict rule. We simplified the enforcement mechanism and captured it by the exogenous cost φ : φ could be endogenized as a transfer made from manager to an auditor to negotiate a favorable treatment.