Essays on Productivity within Firms

Abstract

This dissertation studies how workplace environments and organizational structure shape worker outcomes within firms. Across three chapters, it examines how environmental quality, cohort-based entry, and the rival nature of workplace information affect productivity, social interaction, and learning in distinct organizational settings in India. The first chapter studies the short-term effects of improved indoor air quality on worker productivity, cognitive performance, and physical health. Using a day-level randomized experiment at a call center in the National Capital Region of India, I show that High- Efficiency Particulate Air (HEPA) purifiers reduced indoor PM2.5 concentrations by 51.5 percent over 110 study days. Despite this substantial improvement in air quality, I detect no contemporaneous effects on work output, cognitive test performance, or physiological health indicators. The estimates are precise enough to rule out small-to-modest effects, suggesting that immediate productivity and health benefits in this white-collar, high-pollution setting are limited. The second chapter examines how cohort-based entry shapes the informal organization of work. Using detailed records on workers’ movements on and off the work floor in an insurance claims processing firm, I use coordinated breaks between coworkers to measure social proximity over the course of their tenure. Workers who join the firm at the same time coordinate breaks at much higher rates than workers who do not. This gap is largest early on, but it remains substantial well into a worker’s tenure. Over time, cohorts fragment along narrower lines, while cross-cohort interaction emerges mainly among coworkers on the same team and of the same sex. This suggests that a routine feature of hiring can leave a durable imprint on workplace social structure. The third chapter studies how knowledge diffuses inside firms when it is rival, meaning that one worker’s use of the information reduces its private value to others. Using high- frequency operational data from a call center where sales workers compete for leads from a common pool, I infer workers’ knowledge of lead desirability from persistent changes in their choice behavior and identify learning events over the course of their tenure. These events are temporally clustered and substantially more concentrated within entry cohorts than across them. Taken together with evidence that workers in the same entry cohort interact more in this setting, these patterns are consistent with selective sharing of rival information among more socially connected coworkers.