Politics of Consumption
Problem: How do political views shape everyday life? Specifically, we study the relationship between political views and what kinds of consumer goods people buy, as well as testing whether highly vivid cultural events shape individual's consumption basket.
Sources: Voting data for more than 6500 polling stations across 5 parliamentary elections, receipt-level data from a major supermarket chain, web-scraped nutrition data for more than 100k products, and data on more than 700 cultural events.
Methods: We use voronoi polygons to estimate different regions’ ideological preferences and their corresponding consumption patterns. We then use various kinds of difference-in-differences analyses to estimate the causal effect of cultural events, some of which have political valence and some of which do not.
Challenges: Extremely large datasets (approx 400+ GB) pose their own analytical challenges. We needed a strong testable theory before diving into the dataset to avoid making too much of spurious relationships. Technical challenges also include extrapolating point-level data (e.g. spending at a particular store or voting at a polling station) to form area-wide estimates.
Findings: Consuming good produced locally has a robust positive correlation with voting for more right-wing parties. Consumption of locally produced goods also increases in the wake of nationalistically charged cultural events for certain types of consumers. Political views clearly shape day-to-day consumer behavior.