Publications

We investigate how inequality affects what is considered most necessary to purchase. Combining detailed consumption surveys in India with an instrumental variable approach, we first provide evidence that inequality is associated with a decrease in calorie consumption for poor households. This effect is driven by upward comparisons and associated with poor households spending more on luxuries, consistent with the relative deprivation theory. We then estimate a structural demand system to isolate the impact of inequality on needs from supply-side effects. We find that inequality increases the need of the poor for “little luxuries” (e.g., dairy products, energy use) at the expense of necessities (e.g., cereals). This shift in the hierarchy of needs accounts for more than two-thirds of the decline in the calorie consumption of the poor over the period.

This paper explores whether and how consumers respond to global social movements challenging systemic discrimination and stereotypes. We examine the impact of the #MeToo movement on the market for products with stereotypical markers of femininity. Our analysis of high-frequency stockout and price data from a leading global fashion retailer spans from January 2017 to December 2018 and covers 32 countries, or 89% of the OECD population. Using a triple-difference approach across time, countries, and products, we identify changes in product-level stockouts, corresponding to a drop in the purchase of stereotypically "feminine" items. Our findings suggest a sudden shift in demand, with no immediate adjustments in product assortments or pricing. We discuss potential driving mechanisms, as well as implications for product management and marketing in marketplaces where segmentation is gender based.

This paper examines how the construction of very large homes — or "McMansions" — in U.S. suburbs affects homeowners' satisfaction and housing behavior. Combining three decades of survey data with geolocated information on three million suburban houses, I find that homeowners exposed to newly constructed, large houses report lower satisfaction with their own homes, while their neighborhood satisfaction remains unaffected. This effect is contingent on the visual salience of McMansions, as indicated by their proximity to roads. Homeowners exposed to new-built McMansions are more likely to expand their own homes and take on more debt.

This article provides quasi-experimental evidence on the relationship between employee happiness and productivity in the field. We study the universe of call center sales workers at British Telecom (BT), one of the United Kingdom's largest private employers. We measure their happiness over a 6-month period using a novel weekly survey instrument, and link these reports with highly detailed administrative data on workplace behaviors and various measures of employee performance. Exploiting exogenous variation in employee happiness arising from weather shocks local to each of the 11 call centers, we document a strong causal effect of worker happiness on sales. This is driven by employees working more effectively on the intensive margin by making more calls per hour, adhering more closely to their workflow schedule, and converting more calls into sales when they are happier. In our setting, we find no effects on the extensive margin of happiness on various measures of high-frequency labor supply such as attendance and break-taking.

Advertising reflects and shapes culturally shared beliefs about gender, yet systematic evidence on gender portrayals in advertising and their relationship with consumers' own gender identities remains limited. Using the universe of TV ads aired in the United States between 2010 and 2020, this paper provides the first large-scale descriptive analysis of gender representation across products and brands. The analysis complements gender presence with a scalable, belief-based measure of gender-typing in advertising. Four key findings emerge: gender presence is broadly balanced, but portrayals remain strongly gender-typed, especially for men; representation is highly segmented across products, with men confined to male-typical depictions; this segmentation mirrors product-gender associations even without salient gender presence; and male-typical portrayals persist across both male-and female-dominated audiences. Linking advertising content to data on actual and perceived gender gaps in product consumption reveals that gender-typical representation aligns more closely with inaccurate beliefs about consumers' gender than with their actual gender. The results further evidence substantial heterogeneity across brands unexplained by the gender composition of their consumers, including systematic differences by founders' gender. These findings highlight the importance of studying representation beyond mere presence and motivate further research on the economic and societal consequences of inaccurate representation in advertising.

Drawing on 235,000 customer service calls and 24 million conversational turns from a leading Dutch telecommunications company, this paper brings a new empirical perspective to emotional labor—the management of one’s own and others’ emotions in workplace settings—and highlights its role in shaping consumer behavior and firm outcomes. We construct transcript-based measures of sentiment and four emotions (happiness, sadness, anger, worry) to study how emotional dynamics shape both productivity and consumer outcomes. Positive customer emotions are shown to reduce callback rates and increase sales and contract renewals within 24 hours after a customer call, raising long-term profitability. Crucially, these effects operate through workers’ ability to attenuate negative emotions during calls. The findings highlight emotional labor as a distinct, measurable input in service production, bridging economics of non-cognitive skills with marketing research on customer satisfaction and firm performance.

This project investigates whether gender stereotypes causally affect red meat consumption, which shows a persistent gap between men and women. We test two questions: whether stereotypical gender views explain the consumption gap, and how changeable stereotype-driven consumption is. The design uses two experiments: a gender identity prime that manipulates adherence to gender roles through essays emphasizing "agentic" traits (ambitious, assertive, decisive) versus "communal" traits (emotional, compassionate, affectionate), and an Implicit Association Test that reveals unconscious associations between gender and meat/vegetable consumption. We hypothesize that the gender identity prime increases the consumption gap, while the IAT awareness exercise decreases it. Outcomes are measured through direct methods (allocation tasks, consumption intentions) and indirect measures (a frozen pizza conjoint analysis), using approximately 4,500 American participants randomized into eight treatment arms.

Despite the importance of visual information for consumer decision-making, quantifying consumer visual interest for brands and products across markets and over time remains arduous. This paper introduces Excess Image Search (EIS) as a new behavioral, high-frequency, time- and space-varying, and easy-to-implement measure of visual interest. Using about 134.8 million records of online searches in the U.S. car market (55 brands, 659 car models), we show that a brand's or product's EIS strongly correlates with measures of visual salience, such as consumer ratings of brand visibility or product style and logo. We further examine how visible consumption affects EIS: COVID-19 lockdowns and adverse weather reduced EIS in densely populated areas, while rising used-car prices increased EIS the following week, particularly in economically unequal regions where visible consumption holds greater appeal. We replicate our main findings using apparel brand data, suggesting the measure's broader applicability across product categories.

We randomize the salience of income comparisons in two samples: nearly 1,250 top managers and executives drawn from the alumni of a leading international business school, and 4,800 United States residents. Treated respondents are directed to compare their financial standing to specific others. In three arms they choose the comparators themselves, from their family, colleagues, or former classmates. A fourth arm imposes an upward comparison to the world’s highest-paid CEOs. Among the business elite, every comparison arm shifts distributional preferences in the egalitarian direction: respondents reduce the income share they would allocate to the richest 1% of society by 18%, and allocate more of their own income to redistributive taxation over charitable giving. The evidence points to a fairness channel rather than an informational one. Treated executives attribute financial success more to luck, while their perceived position in the national income distribution does not move. Among representative US respondants, the same treatment produces the opposite pattern at the top of the distribution: high-income respondents revise their perceived rank upward, allocate more to the richest 1%, and shift their own money away from taxation. At overlapping points of the income distribution, the two populations respond in opposite directions (p = 0.002). These patterns suggest that salient comparisons move demand for redistribution through two distinct channels: new information for those who do not know where they stand, and fairness reassessment for those who do.