中文摘要
一名开发商试图说服任期较短、以福利最大化为目标的官僚将一份合同授予她。继任者的决策遵循一种官僚规范：决策必须依赖由其前任记录在案或直接呈交给他的证据。因此，当第一位官员未能记录证据时，贝叶斯推断受到限制。第一位官员可以利用这一点，诱导信息量更高的实验。在说服的静态价值对官僚机构为零、对开发商严格为正的参数设定下，动态博弈中出现两种可能：要么该官员立即凭借更多信息作出决策，从而提高社会福利；要么出现延迟，此时延迟给官僚机构带来的成本抵消了更优信息带来的收益。无论哪种情形，开发商都比在静态说服下处境更差。若推断不受限制，一个直观的完美贝叶斯均衡（PBE）会复制静态结果，因此官僚规范可能提高社会福利，且绝不会降低社会福利。
Abstract
A developer seeks to persuade short-tenured, welfare-maximizing bureaucrats to award a contract to her. The successor's decisions follow a bureaucratic norm: decisions must rely on evidence either recorded by his predecessor or presented directly to him. Thus, Bayesian inference is restricted when the first official fails to record evidence. The first official can exploit this and induce more informative experiments. In parameter settings where the static values of persuasion are zero for the bureaucracy and strictly positive for the developer, two possibilities arise in the dynamic game: either the official decides immediately with more information, increasing social welfare, or there is delay, where the delay cost to the bureaucracy offsets the benefits of better information. In either case, the developer is worse off than in static persuasion. With unrestricted inference, an intuitive PBE replicates the static outcome, so the bureaucratic norm may increase social welfare and never reduces it.