Published Papers

We study peer effects in the workplace. Specifically, we investigate whether, how, and why the productivity of a worker depends on the productivity of coworkers in the same team. Using high-frequency data on worker productivity from a large supermarket chain, we find strong evidence of positive productivity spillovers from the introduction of highly productive personnel into a shift. Worker effort is positively related to the productivity of workers who see him, but not workers who do not see him. Additionally, workers respond more to the presence of coworkers with whom they frequently interact. We conclude that social pressure can partially internalize free-riding externalities that are built into many workplaces.

In this paper, we bring together and advance a growing literature on a third feature: worker self-control. Individuals may not be able to work as hard as they would like. Some workplace arrangements may make self-control problems more severe, while others may ameliorate them.Below, we describe evidence from a field experiment broadly supportive of the selfcontrol perspective. We then argue that many work arrangements can be understood differently through this perspective. Specifically, we use self-control considerations to interpret the productivity increases and changes in work organization that accompany the shift from agrarian to industrialized production.

We study how peers impact worker productivity growth among salespeople in the cosmetics department of a department store. We first exploit a shift assignment policy that creates exogenous variation in salespersons' peers each week to identify and quantify sources of worker learning. We find that peer-based learning is more important than learning-by-doing for individuals, and there is no evidence of forgetting. Working with high-ability peers substantially increases the long-term productivity growth of new salespeople. We then examine possible mechanisms behind peer-based learning by exploiting the multiple colocated firms in our setting that sell products with different task difficulties and compensate their sales forces using either team-based or individual-based compensation systems. The variation in incentives to compete and cooperate within and across firm boundaries, combined with variation in sales difficulty for different product classes, allows us to suggest two mechanisms behind peer-based learning: observing successful sales techniques of peers and direct teaching. Our paper advocates the importance of learning from one another in the workplace and suggests that individual peer-based learning is a foundation of both organizational learning curves and knowledge spillovers across firms.

We estimate the magnitude of spillovers generated by 112 academic "superstars" who died prematurely and unexpectedly, thus providing an exogenous source of variation in the structure of their collaborators' coauthorship networks. Following the death of a superstar, we find that collaborators experience, on average, a lasting 5% to 8% decline in their quality-adjusted publication rates. By exploring interactions of the treatment effect with a variety of star, coauthor, and star/coauthor characteristics, we seek to adjudicate between plausible mechanisms that might explain this finding. Taken together, our results suggest that spillovers are circumscribed in idea space, but less so in physical or social space. In particular, superstar extinction reveals the boundaries of the scientific field to which the star contributes-the "invisible college".

I investigate the effect of faculty quality on PhD student outcomes. To address the endogeneity of faculty quality I use exogenous variation provided by the expulsion of mathematics professors in Nazi Germany. Faculty quality is a very important determinant of short- and long-run PhD student outcomes. A one-standard-deviation increase in faculty quality increases the probability of publishing the dissertation in a top journal by 13 percentage points, the probability of becoming a full professor by 10 percentage points, the probability of having positive lifetime citations by 16 percentage points, and the number of lifetime citations by 6.3.

Using longitudinal elementary school teacher and student data, we document that students have larger test score gains when their teachers experience improvements in the observable characteristics of their colleagues. Using within-school and within-teacher variation, we show that a teacher's students have larger achievement gains in math and reading when she has more effective colleagues (based on estimated value-added from an out-of-sample pre-period). Spillovers are strongest for less experienced teachers and persist over time, and historical peer quality explains away about 20 percent of the own-teacher effect, results that suggest peer learning.

We leverage a temporary block of the Chinese microblogging platform Sina Weibo due to political events to estimate the causal effect of online word-of-mouth content on product demand in the context of TV show viewership. Based on this source of exogenous variation, we estimate an elasticity of TV show ratings (market share in terms of viewership) with respect to the number of relevant comments (comments were disabled during the block) of 0.016. We find that more postshow microblogging activity increases demand, whereas comments posted prior to the show airing do not affect viewership. These patterns are inconsistent with informative or persuasive effects and suggest complementarity between TV consumption and anticipated postshow microblogging activity.

Interpersonal communications have long been recognized as an influential source of information for consumers. Internet-based media have facilitated information exchange among firms and consumers, as well as observability and measurement of such exchanges. However, much of the research addressing online communication focuses on ratings collected from online forums. In this paper, we look beyond ratings to a more comprehensive view of online communications. We consider the sales effect of the volume of positive, negative, and neutral online communications captured by Web crawler technology and classified by automated sentiment analysis. Our modeling approach captures two key features of our data, dynamics and endogeneity. In terms of dynamics, we model daily measures of online communications about a firm and its products as contributing to a latent demand-generating stock variable. To account for the endogeneity, we extend the latent instrumental variable technique to account for dynamic endogenous regressors. Our results demonstrate a significant effect of positive, negative, and neutral online communications on daily sales performance. Failure to account for endogeneity results in a severe attenuation of the estimated effects. From a managerial perspective, we demonstrate the importance of accounting for communication valence as well as the impact of shocks to positive, negative, and neutral online communications.

We study how opinion leadership and social contagion within social networks affect the adoption of a new product. In contrast to earlier studies, we find evidence of contagion operating over network ties, even after controlling for marketing effort and arbitrary systemwide changes. More importantly, we also find that the amount of contagion is moderated by both the recipients' perception of their opinion leadership and the sources' volume of product usage. The other key finding is that sociometric and self-reported measures of leadership are weakly correlated and associated with different kinds of adoption-related behaviors, which suggests that they probably capture different constructs. We discuss the implications of these novel findings for diffusion theory and research and for marketing practice.

Using box-office data for all movies released between 1982 and 2000, I quantify how much the consumption decisions of individuals depend on information they receive from their peers, when quality is ex ante uncertain. In the presence of social learning, we should see different box-office sales dynamics depending on whether opening weekend demand is higher or lower than expected. I use a unique feature of the movie industry to identify ex ante demand expectations: the number of screens dedicated to a movie in its opening weekend reflects the sales expectations held by profit-maximizing theatre owners. Several pieces of evidence are consistent with social learning. First, sales of movies with positive surprise and negative surprise in opening weekend demand diverge over time. If a movie has better than expected appeal and therefore experiences larger than expected sales in Week 1, consumers in Week 2 update upward their expectations of quality, further increasing Week 2 sales. Second, this divergence is small for movies for which consumers have strong priors and large for movies for which consumers have weak priors. Third, the effect of a surprise is stronger for audiences with large social networks. Finally, consumers do not respond to surprises in first-week sales that are orthogonal to movie quality, like weather shocks. Overall, social learning appears to be an important determinant of sales in the movie industry, accounting for 32% of sales for the typical movie with positive surprise. This implies the existence of a large "social multiplier" such that the elasticity of aggregate demand to movie quality is larger than the elasticity of individual demand.

The authors quantify the impact of social interactions and peer effects in the context of physicians' prescription choices. Using detailed individual-level prescription data, along with self-reported social network information, the authors document that physician prescription behavior is significantly influenced by the behavior of research-active specialists, or "opinion leaders," in the physician's reference group. The authors leverage a natural experiment in the category: New guidelines released about the therapeutic nature of the focal drug generated conditions in which physicians were more likely to be influenced by the behavior of specialist physicians in their network. The authors (1) find important, statistically significant peer effects that are robust across model specifications; (2) document asymmetries in response to marketing activity across nominators and opinion leaders; (3) measure the incremental value to firms of directing targeted sales force activity to these opinion leaders; and (4) present estimates of the social multiplier of detailing in this category.

Word of mouth affects consumer behavior, but how does the language used in word of mouth shape that impact? Might certain types of consumers be more likely to use certain types of language, affecting whose words have more influence? Five studies, including textual analysis of more than 1,000 online reviews, demonstrate that compared to more implicit endorsements (e.g., "I liked it," "I enjoyed it"), explicit endorsements (e.g., "I recommend it") are more persuasive and increase purchase intent. This occurs because explicit endorsers are perceived to like the product more and have more expertise. Looking at the endorsement language consumers actually use, however, shows that while consumer knowledge does affect endorsement style, its effect actually works in the opposite direction. Because novices are less aware that others have heterogeneous product preferences, they are more likely to use explicit endorsements. Consequently, the endorsement styles novices and experts tend to use may lead to greater persuasion by novices. These findings highlight the important role that language, and endorsement styles in particular, plays in shaping the effects of word of mouth.