Under Review

with Anocha Aribarg

Center for Brand Leadership Best Brand Paper Award, Haring Symposium, 2026

Abstract

Firms increasingly use sustainability claims and marketing campaigns to promote sustainable consumption. Designing effective messages requires understanding why consumers value sustainability, because different motives require different persuasive appeals. This paper develops an empirical framework to recover and quantify the primary latent benefits driving sustainable choices: Do Good (environmental impact), Feel Good (warm glow), and Look Good (social image). We design an incentive-compatible conjoint study that elicits consumers' choices and perceived benefits. We then estimate a hierarchical Bayesian choice model linking sustainability features, latent benefits, and choice, while allowing benefit importance to correlate and vary across consumers. Applied to coffee capsules, the model shows that Feel Good dominates for recyclable and compostable packaging, whose sustainability value depends on individuals' disposal effort, whereas Do Good dominates for a low-carbon footprint, whose sustainability value is achieved through supply-chain activities with little consumer involvement. Look Good matters little on average but becomes more important among consumers highly sensitive to social norms. A follow-up randomized experiment confirms that advertisements emphasizing consumers' relevant latent benefits raise willingness to pay more than advertisements mentioning sustainability features alone. Our framework extends to other contexts, helping firms and policymakers identify motives behind sustainable choices and design marketing strategies accordingly.

with Sherry He

Media attention: Harvard Business Review (2025)

AbstractSSRN

Companies are increasingly prioritizing environmental, social, and governance strategies, with sustainability labeling emerging as a prominent approach to inform consumers about a product's sustainability status. Despite the widespread adoption of sustainability labels, whether and why they affect product sales remains unclear. This paper quantifies the causal impact of sustainability labeling by leveraging a large-scale dataset from Amazon. Employing a difference-in-differences framework, we find that sustainability labeling leads to a 12.5% decrease in sales rank (indicating an increase in sales units) of labeled products within eight weeks. Furthermore, we show that the primary mechanism is the information effect – sustainability labeling influences consumers' decisions by conveying information that aligns with their preferences – rather than the salience effect (i.e., improved product listing salience owing to label salience or search filter availability) or platform endorsement (i.e., improved organic search ranking). Moreover, consumers are more responsive to easily accessible information on sustainability features than to hard-to-access details about certification processes, such as stringency and transparency. These findings suggest that consumers passively search for and purchase sustainable products. These insights carry implications for both firm-level sustainability strategies and platform-level information architecture.

with Anna Tuchman and Youngeun Lee

Abstract

This paper investigates product reformulation – the adjustment of ingredients or composition of ingredients in existing products – in the food and beverage industries. Using comprehensive data on product ingredients, nutrition facts, and store sales, we document the prevalence of reformulation, characterize the tactics firms employ, assess the impact of reformulation on product quality and differentiation, and evaluate consumer demand for reformulated products. We find that reformulation is a widespread and persistent practice, with reformulated products accounting for 26% of volume sales across categories and over time. Firms use a diverse range of tactics, including dilution and adjusting compositional flexibility, leading to moderate shifts in nutrient density, but minimal change in the average nutritional quality within a category. We find that less healthy products tend to experience improvements in their nutritional profiles while healthier products are reformulated to be less healthy, suggesting a convergence toward average nutrient values within a category following reformulation. Finally, we find limited evidence of meaningful demand response, indicating that reformulated versions may serve as close substitutes for their originals, even when nutrient density deteriorates without corresponding price changes. Such minimal responses suggest that reformulation may modify product quality in a way that is difficult for consumers to detect.

with Sherry He