Research

Spatial media competition under multi-homing link  

Media outlets, l and r, spatially compete to report a state. A fixed reader share of l (r) is primary (who read its report as first choice), while another fixed share read l after r (or vice versa), acting as secondary or multi-homing readers. I propose a perception index which shows that prior exposure to r (l) lead multi-homers to perceive l's (r's) report more moderately than its primary readers. If outlets' profit depend on partisan interests, as well as satisfying its readers whose utility depends on report perception, how do outlets strategically design their reports by investing (to acquire) and spinning (to reveal) information?  The perception difference between primary and secondary readers relaxes either outlet's intensive-extensive margin trade-off, prevalent in any spatial competition. This allows an outlet to piggy-back on their opponent's strategy, thereby generating more spin, but not always with higher investment, and at the expense of reader dissatisfaction. Comparative statics show how change in multi-homing not only lead outlets to minimally or maximally differentiate their report, but also the rate at which such differentiation occurs. 

Personalised Incentives: Segregation in a platform with networks* (with Ratul Das Chaudhury) (submitted) link 

An online platform often lacks complete information about user-level preferences over networked goods. Can the platform instead utilise information about user-level social networks to maximise profit? We develop a binary choice framework where an online platform generates ad-revenue from a free basic plan (B) and subscription revenue from a fee-based exclusive plan (E). Generic content is costlessly provided on B, but the platform endogenously provides costly personalised incentives on E by leveraging user-level network information. In equilibrium, influential users receive greater incentives, which attract other susceptible users to join E. Comparative statics show how changes in ad-revenue, subscription fee, and users’ bias towards B or E impact platform profit monotonically or non-monotonically, depending on how the change interacts with user-level network statistics like the average influence of users, or the distribution of influence across users. We further analyze whether an increase in strength of network and platform competition increases profit.

Land-redistribution and coercive violence (with Rohan Gudibande and Vatsalya Srivastava) link 

In a post land reform environment, violence and redistribution are substitutes for an opportunistic incumbent attempting industrialization. We setup a theoretical model to explore this relationship for individual and group level violence across two types of policy driven post land-reform industrialization (PLRI) --- small and large-scale, in the Indian state of West Bengal. We find strong evidence for our theoretical prediction of an inverse relationship between land redistribution and both types of violence during the period of small scale industrialization after an industrial policy was announced in 1994-95 by the incumbent. This relationship however breaks down for individual level violence during attempts at large-scale industrialization between 2006-2011, even though both types of violence increase in this phase.