The familiar decision analysis setting in accounting, such as a cost‐volume‐profit analysis setting, is modeled as arising in a “larger” context in which subsequent choice is also contemplated. We then ask when this “larger” context decomposes in a manner such that the immediate problem of interest can be modeled as an expected utility maximization problem that is completely divorced from the subsequent choice problem. Outcome and taste independence conditions are the key ingredients in being able to model the immediate problem as having no link whatever to its successor. Résumé. Le contexte dans lequel les techniques d'analyse de décision comme par exemple l'analyse coĉt‐volume‐profit est utilisée est un cadre d'un modèle élargi qui tient compte du choix subséquent. On se demande en quelles circonstances ce contexte élargi peut‐il se décomposer de façon telle que le problème d'intérêt immédiat en soit un de maximisation complètement séparé du problème du choix subséquent. Le résultat et des conditions d'indépendance de préférence sont les ingrédients clés pour modeler le problème immédiat comme n'ayant aucun lien avec son successeur.

ABSTRACT: Models that characterize Pareto-efficient sharing of joint venture profits or constrained Pareto-efficient sharing of income in principal-agent contracting problems have ignored tax considerations. We extend the theory by showing that the effect of taxes on optimal contracting (both in the face of and in the absence of moral hazard problems) is related to the effect of changes in risk attitudes towards lotteries over pre-tax income. For example, optimal contracts will reflect the tax-induced demand for insurance of a risk-neutral individual who faces a progressive income tax schedule; that is, the risk-neutral individual will not bear all the risk, and in the face of moral hazard on the act selection of a risk-neutral agent, demand for monitoring will be created where none existed in the absence of the progressive tax. We also show that Pareto-optimal risk-sharing contracts do not generally result in expected tax minimization, even when taxes are modeled as a deadweight loss to the system.

ABSTRACT: A simplified audit setting is used to illustrate the crucial nature of strategic interactions in audit planning and in assessing audit risk. Unlike single-person decisiontheoretic models which essentially represent games against nature, the model developed here allows a prospective audit to influence the behavior of the auditee. We reformulate the problem in a game-theoretic framework with rational players which (1) encompasses strategic factors for both the auditor and auditee, (2) is consistent with behavioral hypotheses regarding the effect of an audit, and (3) is consistent with certain audit phenomena such as randomized strategies. An illustration is provided which demonstrates several points. First, both the auditor and the auditee may frequently use a randomized strategy. Second, the auditor's strategy depends on the interaction between the accounting control system and the auditee's actions. In addition, the use of traditional single-person decision theory may frequently cause errors in estimating audit risk because it fails to consider audit influences on the auditee. Settings in which decision theory may serve as an adequate model simplification are also considered.

[A simplified audit setting is used to illustrate the crucial nature of strategic interactions in audit planning and in assessing audit risk. Unlike single-person decision-theoretic models which essentially represent games against nature, the model developed here allows a prospective audit to influence the behavior of the auditee. We reformulate the problem in a game-theoretic framework with rational players which (1) encompasses strategic factors for both the auditor and auditee, (2) is consistent with behavioral hypotheses regarding the effect of an audit, and (3) is consistent with certain audit phenomena such as randomized strategies. An illustration is provided which demonstrates several points. First, both the auditor and the auditee may frequently use a randomized strategy. Second, the auditor's strategy depends on the interaction between the accounting control system and the auditee's actions. In addition, the use of traditional single-person decision theory may frequently cause errors in estimating audit risk because it fails to consider audit influences on the auditee. Settings in which decision theory may serve as an adequate model simplification are also considered.]

[Models that characterize Pareto-efficient sharing of joint venture profits or constrained Pareto-efficient sharing of income in principal-agent contracting problems have ignored tax considerations. We extend the theory by showing that the effect of taxes on optimal contracting (both in the face of and in the absence of moral hazard problems) is related to the effect of changes in risk attitudes towards lotteries over pre-tax income. For example, optimal contracts will reflect the tax-induced demand for insurance of a risk-neutral individual who faces a progressive income tax schedule; that is, the risk-neutral individual will not bear all the risk, and in the face of moral hazard on the act selection of a risk-neutral agent, demand for monitoring will be created where none existed in the absence of the progressive tax. We also show that Pareto-optimal risk-sharing contracts do not generally result in expected tax minimization, even when taxes are modeled as a deadweight loss to the system.]