نوع مقاله : مقاله پژوهشی
عنوان مقاله English
نویسندگان English
Mechanism design, the engineering branch of game theory, rests on a foundational premise: arrange the payoffs correctly and the rational agent will select the action the designer intends. That premise is the explicit operating logic of Iran's new-generation petroleum contracts, and more generally of decision architecture in large state-linked organizations. This paper addresses a strategic question: when an engineered incentive not only fails to work but drives behavior in the direction opposite to its intent, where in the design chain has the failure occurred? Rather than adding a further catalogue of behavioral biases, the paper locates the failure in a constraint that standard mechanism design leaves implicit.
The study is a conceptual–integrative paper employing abductive reasoning: it moves from a set of observations to the most parsimonious generative structure that would render them expected, and then converts that structure into falsifiable propositions. The anomalies come from semi-structured interviews with 10 industry experts, and which were translated into an experiment design with 68 graduate participants (34 employer–contractor dyads) playing nine incentivized, industry-framed economic games with real monetary payoffs at the Behavioral Economics Laboratory, Faculty of Management, University of Tehran. The full statistical treatment of each anomaly is reported in companion papers; here the evidence functions as the explanandum, not as a finding claimed by this paper. Because abduction establishes only the best available explanation, a dedicated section evaluates four rival accounts - comprehension failure, experimenter demand, risk preference, and measurement ceiling - and shows that none reproduces the full observed pattern.
Three anomalies that appear distinct are argued to be one phenomenon. The alignment paradox: the 5% stake raised delegation from 51. 5% to 91%, precisely against the strategy it had made dominant, and the switching was strongly asymmetric (94% versus 11%) , a directional signature that random misunderstanding cannot produce. The quality ceiling: 94. 1% invested in high quality irrespective of contract type, with 85. 3% within-subject consistency. The asymmetric cooperation pattern: dictator offers averaged 37. 3% against a 28. 3% benchmark with zero-offers entirely absent, yet ultimatum offers (35. 2%) did not exceed their own benchmark, generosity appeared precisely where strategy did not compel it, which is a prediction of the framework rather than an exception to it. The proposed framework, normative override, posits a two-stage choice architecture in which Stage 2, the consequentialist payoff comparison the standard model assumes, is reached only when Stag-1 norm retrieval is weak or contested and the decision-maker has both the cognitive capacity and the psychological safety to get there. Four propositions follow: norm primacy, perverse alignment, cognitive gating, and hierarchy/psychological-safety moderation.
The central theoretical implication is stated as a constraint. In norm-governed populations, incentive compatibility and individual rationality are not jointly sufficient for implementation; a mechanism must additionally satisfy what we term the norm-compatibility constraint: the designer's target action must lie within the norm-permissible action set. Where it does not, no transfer implements it, and raising the transfer can push the target action further outside that set — which is the alignment paradox. The standard model is the special case in which the full action set coincides with normative -compatible set, so the claim delimits rather than refutes it. The practical ordering of intervention inverts: legitimize the action first, then tune the payoffs. Five boundary tests are specified, one testable on data already in hand.
کلیدواژهها English
