Mediation by Manipulating Attribution : Bayesian Persuasion in Ultimatum Games

How people attribute the others’ previous actions often influences their feeling of inequality and then their subsequent responses. In a bilateral ultimatum bargaining game with both inequality aversion and incomplete information, a greedy offer can be attributed to either the proposer’s high physical valuation or her selfishness, and the latter is more likely to trigger retaliation. In the framework of Bayesian persuasion, we examine the optimal strategy of attribution manipulation through verifiable information disclosure by an informed mediator in such games. A utility model is introduced to capture the influence of attribution on behavior, where the feeling of inequality is triggered by the belief that others’ actions violate the social norm attached to specific types. Three objective functions are considered: The settlement probability, the seller’s expected revenue and the buyer’s ex ante feeling of inequality. We show that the mediator generally benefit from information disclosure, and the optimal signal reveals the true state in a partial manner. Also, optimal signal has one of the three possible patterns, and the condition for each of them is solved by concavification