Essays in networks, conflict and contests

Bose, Tamoghna (2025). Essays in networks, conflict and contests. University of Birmingham. Ph.D.

Bose2025PhD.pdf
Text - Accepted Version
Available under License All rights reserved.
Download (5MB)

Abstract

My thesis has three chapters which look at: (i) the centralised network design in the presence of infiltration, (ii) an experiment, analysing the impact of voluntary disclosure of information in a model of asymmetric contests under incomplete information and, (iii) optimal contracts between a politician and agent in deciding resource allocation in a scenario of conflict. The summaries of each chapter are as follows:

Chapter 1: Terrorist organisations face a dilemma: increased connectivity can enhance their benefits but also increase the risk of infiltration. In this paper, we examine how infiltration by law enforcement authorities affects the structure of terrorist networks. We model the network structure in the presence of infiltration and solve for an efficient network. The efficient network is determined by the probability of an individual terrorist being captured through direct and indirect connections. If capturing an individual terrorist is possible through both direct and indirect connections with equal probability, the efficient structure will form of maximally connected sub-structures i.e., `components'. If the probability of an individual terrorist getting caught through direct connections is higher than indirect connections then non-maximally connected sub-structures may arise. The efficient network can have either symmetric or asymmetric components, depending on the probability of infiltration. For some parameter values the efficient network is composed of only symmetric components.

Chapter 2: In many contest scenarios, such as space races, R\&D, and military conflicts, participant's abilities are publicly known, while their costs remain private. This paper experimentally investigates the incentives for voluntarily disclosing private information about marginal costs in an asymmetric one-shot contest. Our between-subjects experimental design with two players includes a baseline with incomplete information with no disclosure, and two treatments with incomplete information with two-sided disclosure and one-sided disclosure. Participants risk preferences were elicited during the experiment to better understand their decision-making processes. We report the findings of our experimental study on bidding behaviour (average investment levels), probabilities of winning, and expected payoffs in ex-ante asymmetric contests. We explore whether participants consistently choose to reveal their private information when given the option. Additionally, we examine how investment levels, winning probabilities, and expected payoffs vary by ability and disclosure decisions. Our results show that participants do not always reveal their private information when given the opportunity to do so, and this decision depends on their unit cost.

Chapter 3: We analyse a principal-agent model where a politically elected leader of a group (citizens) negotiates division of resources with another group (insurgents). Citizens have preferences over resources that have economic and ideological value but keeping more resources leads to more conflict with the insurgent group. Leaders differ in their ability which affects both the cost of conflict as well as the value of the resources net of costs of conflict. We illustrate when leaders of different abilities and ideologies choose to cede more or less of the resources to the insurgent group. The first best contract from the citizens point of view is characterised. We find that in the first best contract, the amount of resources ceded to the insurgent group can be non-monotonic in ability of the citizen's leader. We provide illustrations of such non monotonic contracts. We further show what the optimal choice is by the leader when he is re-elected vs replaced in the first best contract.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):