The Economic Value of Natural Language Communication: Theory and Experiments

Siyu Wang

Advisor: Daniel E Houser, PhD, Department of Economics

Committee Members: Thomas Stratmann, David Eil, Omar Al-Ubaydli, Cesar Martinelli

Vernon Smith Hall (formerly Metropolitan Building), #5043
April 15, 2016, 10:00 AM to 07:00 AM


Language is a powerful and complex human tool facilitating social and economic decisions. That the rich structure of language has evidently survived an evolutionary process has led some economists to argue for the importance of understanding its influence on human decision-making. Recently, an extensive experimental literature reveals, in relation to a well-designed predetermined language space, efficient economic outcomes emerge more readily when players can communicate using natural language. This dissertation investigates why natural language communication promotes efficient economic outcomes. In doing so, I bridge theory and experiments in a way that sheds light on the rich structure of language.

In Chapter I, I build a theoretical framework which predicts that, in order to coordinate, people both communicate with and respond to two key features of language: intentions and attitudes, where attitude indicates the strength of a message sender’s desire to have her intention followed. I demonstrate that three types of equilibria coexist in this environment: agreement equilibrium, negotiation equilibrium and communication failure equilibrium. I show that negotiation equilibrium significantly improves the probability of coordination as compared to the classical intention signaling model. 

In Chapter II, I use controlled laboratory experimentation in both the United States and China to test the predictions of the model detailed in Chapter I. I find (i) natural language messages do include both signaled intentions and attitudes; (ii) people respond both to intentions and attitudes when making decisions; and (iii) the use of attitude significantly improves coordination. Moreover, while males and females recognize and respond to intentions and attitudes equally well, we find females are more likely to send demanding signals, while males are more likely to send messages focused on intentions.

In Chapter III, I investigate why “naïve advice” often promotes social learning better than observing a complete history of past actions and corresponding outcomes. For example, Schotter (2003) shows that advice is often found to be more effective in shaping the decisions that people make and tends to push those decisions in the direction of the predictions of rational theory. I ask, why does advice impact decisions differently than observations? Inspired by Manski (2004), I hypothesize that advice is different because it provides information about past generations’ expectations regarding counterfactual outcomes. To test this I design an experiment using inter-generational games. In the experiment, players can learn from others by observing the history of decisions and outcomes and/or receiving written advice. I expect the advice we collect to contain not only chosen actions and corresponding outcomes, but also counterfactual related to expected outcomes of the actions not chosen.