Working Papers
- Job Market Paper:
- How do Influencers Influence? Evidence from Engineers Without Borders' Online Threaded Discussions - work in progress
- Empirical Investigation of Dynamic Networks - work in progress
- Diffusion of Peer to Peer Fundaraising Campaigns - work in progress
- A Theory of Twitter - joint with Anke Kessler
- Optimal Crime Networks - Theory and Lessons for Policy - joint with Stephen Easton and Alexander Karaivanov
On the Impact of Social Networks on Charitable Behaviour: Theory and Evidence
I study the direct and spillover effects of social interactions using in a network of volunteers from Engineers Without Borders (EWB), Canada. I model social interactions as a network game in which agents simultaneously choose their effort levels, taking the network and their friends' efforts as given. The effects from social interactions are introduced through two separate channels: a strategic interaction term which affects the marginal benefit from supplying effort and a direct spillover term affecting the level of an agent's payoff. I construct three different categories of online and offline networks and estimate the model using instrumental variables and system GMM. The identification strategy relies on the yearly variation in the location of the EWB national conference and new members' participation levels in this event each year. The estimates demonstrate different patterns for engagement versus fundraising activities. Large significant levels of strategic complementarities are always present in fundraising activities regardless of the definition of links, however, in engagement activities, strategic complementarities are only significant in online networks. Additionally, engagement activities exhibit positive significant levels of direct spillovers for all networks. In contrast, in fundraising campaigns, the direct spillover effect is only significant in large offline networks.
We develop a theoretical network-based model of the social network Twitter, formulating individual interaction as a dynamic game in which heterogenous agents choose a ‘niche’ (a subset of the type space) to tweet in, and whom to follow. Agents consume tweets close to their own types, and seek to maximize the number of their followers. Starting from any initial niche with an arbitrary length, we show that the dynamic Markov process converges to a niche with a finite maximum length, and this niche contains agent’s own type. We also show that information does not diffuse as widely as one might expect: although many agents are directly or indirectly connected to each other, the news does not travel too far since agents strategically choose what news to tweet or retweet in accordance with their niche, i.e., they strategically filter information. We also discuss the stable networks that the dynamic process converges to in equilibrium and show that the star network is never stable if agents are similar enough, and the only stable network is the bidirectional full network. In contrast, when agents are further away in the type space, the star network is the only stable network.
We construct a social network model of criminal activity. Agents’ payoffs depend on the number and the structure of their connections with each other and are determined in a Nash equilibrium of a crime activity supply game. Unlike much of the literature which takes the network structure as given, we study optimal networks, defined as the networks that maximize the sum of agents’ payoffs. We characterize the Nash equilibrium in crime activity and use our theoretical results to identify the optimal network for given cost and benefit parameters using an algorithm that searches over all possible non-isomorphic graphs of given size. We also analyze, via simulations, the effects of different anti-crime policies (both expected and unexpected) on the optimal crime network structure and the overall crime level – removing agents, removing links, and/or varying the probability of apprehension.
Presentations
- Econometric Society World Congress - Montreal, August 2015
- Canadian Economic Association - Vancouver, May 2014
- Canadian Economic Association - Montreal, May 2013
- Stat In Paris, Statistics and Econometrics of Networks - Paris, Nov 2013
Other Conferences & Research Opportunities
- Visiting PhD Student at Stanford University (with Matthew Jackson) - Palo Alto, Dec - Feb 2014
- Canadian Economic Association - Toronto, May 2015
- 2014 Network Science Conference - Berkeley, June 2014
- Canadian Econometric Study Group - Vancouver, Oct 2014
- North American Summer Meeting of the Econometric Society, Chicago, June 2012