Efficiency and other-regarding preferences in information and job-referral networks

Thesis

In this thesis I study how networks are formed and I analyse the strategies that well-connected individuals adopt in public good games on a network.

In chapter one I study an artefactual field experiment in rural India which tests whether farmers can create efficient networks in a repeated link formation game, and whether group categorisation increases the frequency of in-group links and reduces network efficiency. I find that the efficiency of the networks formed in the experiment is significantly lower than the efficiency which could be achieved under selfish, rational play. When information about group membership is disclosed, in-group links are chosen more frequently, while the efficiency of network structure is not significantly affected.

Using a job-referral network experiment in an urban area of Ethiopia, I investigate in chapter two whether individuals create new links with the least connected players in the network. In a first treatment, competition for job-referrals makes it in the player’s interest to link with the least connected partners. In this treatment, links to the least connected players are significantly more likely than links to better connected individuals. In a second treatment, connections only affect the welfare of the new partner. Choosing the least connected player minimises inequality and maximises aggregate efficiency. This may motivate other-regarding players. In this treatment, however, links to least connected partners are not significantly more likely than links to other players.

In chapter three I explore the characteristics that individuals value in the people they approach for advice. Using cross-sectional data on cocoa farmers in Ghanaian villages and a matched lottery experiment, I find an association between the difference in the aversion to risk of two farmers and the probability that one farmer is interested in the advice of the other farmer.

In chapter four I study a one-shot public good game in rural India between farmers connected by a star network. Contributions by the centre of the star have a larger impact on aggregate payoffs than contributions by the spoke players. I use the strategy method to study whether the centre of the star contributes more than the average of the spokes. In selected sessions, I disclose participants’ expectations about the choices of the centre of star. I find that the centre player contributes just as much as the average of the spokes, and that he is influenced by the expectations that other players hold about his decisions.

Authors: Caria, AS

• DOI: 10.5287/ora-kobqpqkyj
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: lab-in-the-field experiment; social networks; other-regarding preferences; public goods
• Subjects: Economics--Psychological aspects; Economic development