讲座题目：CHEAP TALK DELEGATION EXPERIMENTS

　　时间：2017年1月10日上午10：00-11：30

　　地点：B127

　　报告人：张雪，博士候选人，The University of Adelaide

　　摘要：We study experimentally how informative cheap talk is in a delegation game where the interests of the delegator and the delegatees are not aligned and information is asymmetric. Further we ask how much efficiency gain can be achieved in delegation by cheap talk. This paper contributes to the cheap talk literature by a novel delegation scenario to study how the form of cheap talk can affect information transmission. We design a one-shot tailored three-person delegation game based on a real effort task. Two players can simultaneously send a costless message about their performance along with their avatar to a delegator. If the task is delegated, a bonus is transferred from the delegator to the chosen player. The realistic performance context and the avatar element permit us to study the social dimension inquiry like gender gap and reciprocity. To explore the role of cheap talk, we vary the form of message space. We start by a structured massage space which consists of integers to represent how good the performances are (i.e. Precise Message Treatment). Next we introduce noise by partitioning the massage space into intervals (i.e. Fussy Message Treatment). Lastly we allow free text communication to allow the size of the massage space to be effectively unbounded (i.e. Free Communication Treatment). In line with the literature of lie aversion, truthful reports are observed in messages across all treatments. Our data provide further evidence to explain the lie aversion by other-regarding preference in FMT. Surprisingly we find that delegators disregard messages in PMT but respond to favourable messages by increasing chance of delegation in FMT and FCT. Compared to no delegation scenario our cheap talk delegation options can improve the social welfare to a certain degree. The most efficient outcome is obtained in FMT where players can express freely how competent they are and delegators pick up the right cue.