Michael Kurish

Does the Offline Environment Impact Online News Browsing: Evidence from Migration (Job Market Paper)

We address whether the offline news environment affects online news consumption choices, focusing on the nature of the relationship between online and offline partisan news sources - are they substitutes, complements, or independent? We build and estimate a model extending the standard discrete choice framework that allows us to identify this relationship via recovery of a single structural parameter. This parameter interprets the relationship between inside goods (online news) and the outside option (offline environmental news options). To estimate our model, we take advantage of a large novel data set of a subset of US users from a major web browser. We exploit the migration patterns of users as our source of identifying variation. In particular, we analyze consumer browsing before and after a move to identify our model. This exercise leads to evidence of independence between online and offline news goods at the population level, suggesting that, on average, consumers view these goods as neither substitutes nor complements. However, further investigation points to significant between-individual heterogeneity in the relationship between online and offline news, which we then show is partially explained by observable consumer characteristics via machine learning methods. We explore the implications of this heterogeneity and propose additional analysis to identify the potential mechanisms behind our findings.

Estimating the Impact of Social Media on Internet News Consumption

Considerable attention has been given to the impact of social media as a route of news discovery and consumption, especially around claims that it encourages the discovery of misleading or low quality news. We seek to discover the causal impact of social media on the quantity and mix of news consumed by readers. We exploit a January 2018 algorithm shift in the Facebook algorithm as a source of quasi-experimental variation. We match treatment users, who use Facebook, and control users, who do not, in order to identify the impact of the shift on browsing behavior. We identify different topics and news types using text analysis methods. We build a model of user browsing preferences to decompose the impact of social media on browsing for different topics and types of news. Our preliminary findings focus on opinion news, and indicate that social media use is associated with increased news browsing, and relatively more factual news versus less opinion-oriented content.