Inconsistent behaviour and beliefs in a framed ultimatum game: evidence from Khorog, Tajikistan

We are the first to combine harassment bribery and social norms in a laboratory experiment with 296 participants implemented in Khorog (Tajikistan), where bribery (bakshish) is widespread. We investigate the (in)consistency among different aspects of social norms such as
actual behavior, empirical expectations, normative expectations, and the conditionality of behaviour upon expectations, using the example of harassment bribery, where "officials" have the possibility to harass a bribe from "citizens" in order to provide a prize they are entitled to. The experiment shows that the vast majority of officials demand bribes and citizens agree to pay them. Empirical expectations of officials about mean bribes are consistent with actual behaviour, however participants underestimate the actual prevalence of corrupt behaviour. On average participants consider that it is socially appropriate to demand and pay relatively small bribes. However, participants consider larger bribe amounts demanded and paid to be less socially appropriate. The experiment provided evidence of contagion, but not conformism effects of harassment behaviour.
Our experiment demonstrates that empirical information affects normative expectations, but asymmetrically.
Empirical information about the behaviour of officials does not impact the social appropriateness of bribe harassment and bribe acceptance, while empirical information about citizens paying bribes makes harassing and accepting bribes more socially appropriate.
The results from this study will help anti-corruption agencies, particularly in identifying which outcomes to publicize in order to improve policy effectiveness.

本实验首次将贿赂、骚扰及社会规范结合于一项实验室研究之中，该研究以塔吉克斯坦的霍罗格（Khorog）为实施地，参与人数共计296名，其中贿赂（bakshish）现象普遍存在。本研究旨在探讨社会规范不同方面之间的（不）一致性，如实际行为、实证预期、规范预期以及行为对预期的依赖性条件，以骚扰贿赂为例，其中“官员”拥有骚扰“公民”以索贿并为其提供应得奖励的可能性。实验结果显示，绝大多数官员要求贿赂，而公民则同意支付。官员对平均贿赂数额的实证预期与实际行为保持一致，然而参与者低估了腐败行为的实际普遍性。平均而言，参与者认为要求并支付相对较小的贿赂在社会上是可以接受的。然而，参与者认为要求并支付较大的贿赂则在社会上显得不那么合适。实验提供了感染效应的证据，但并未发现一致性效应。我们的实验表明，实证信息影响规范预期，但这种影响是非对称的。关于官员行为的实证信息并未影响贿赂骚扰和接受的社会适当性，而关于公民支付贿赂的实证信息则使得骚扰和接受贿赂在社会上显得更为合适。本研究的结果将有助于反腐败机构，特别是在确定哪些成果应公开以提升政策有效性方面。