UC Berkeley

This dissertation consists of three essays on economic development in the context of South Asia. In the first essay, I study how power outages impact large scale manufacturing firms. First, I use monthly electricity billing data from large scale manufacturing firms in Pakistan to characterize how these firms respond to interruptions in electricity supply. I show that firms differentially adjust on-grid capacity utilization based on their sensitivities to increases in fuel costs. Next, I incorporate the empirical results into a dynamic model of utilization adjustment and self-generation in response to power outages and estimate structural cost parameters. Finally, I use my estimates to determine the magnitude of cost increases that firms incur. My results suggest that between January 2010 to March 2012, no firm in my sample had a cost increase above 1%. During quarters of high numbers of outages, the cost increases for most firms is below 3%. In my sample, firms in the chemical and textile industries are impacted by power outages more than firms, of similar size, in other industries.

In the second essay, I examine the impact of power outages on the likelihood of manu- facturing firm exit. Again, I use proprietary electricity billing data for manufacturing firms in Pakistan to determine the month of exit and merge this with monthly outage data during the period 2009-2012. I find that power outages disproportionately increase the exit proba- bility of small firms (both in the formal and informal sector). For each 100 hours of power outages, the probability of small firm exit increases by over 6%. This result is consistent with existing evidence on the impact of power disruption on various measures of firm health as well as research on the cost impact of outages on large scale manufacturing and suggests that policies which target small firms are more likely to alleviate the detrimental effects of power disruption.

In the third essay, I study how religious holidays impact the incidence of violence between Hindus and Muslims in India and I analyze how violence impacts vote share. By exploiting variation in the timing of religious holidays - because Hindu and Muslim holidays follow Sanskrit and Lunar calendars, respectively - I show that religious holidays and the interaction of Hindu and Muslim holidays positively affect the incidence of violence. Next, I consider the usefulness of violence by studying the electoral effects of violence on vote share. I find that violence before an election is associated with electoral gains.