Working Papers

Alone, Together: Product Discovery Through Consumer Ratings [PDF]

Consumer ratings have become a prevalent driver of choice. I develop a model of social learning in which ratings can inform consumers about both product quality and their idiosyncratic taste for them. Depending on consumers’ prior knowledge, I show that ratings relatively advantage lower quality and more polarizing products. The reason lies in the stronger positive consumer self-selection these products generate: to buy them despite their deficiencies, their buyers must have a strong taste for them. Relatedly, consumer ratings should not be used to infer product design: what is polarizing ex-ante needs not be so among its buyers. I test these predictions using Goodreads book ratings data, and find strong evidence for them. Goodreads appears to serve mostly a matching purpose: tracking the behaviour of its users over time reveals an increasing degree of specialization as they gather experience on the platform: they rate books with a lower average and number of ratings, while focusing on fewer genres. Thus, they become less similar to their average peer. Taken together, the findings suggest that consumer ratings contribute to both the long tail and, relatedly, consumption segregation. For managers, this illustrates, counterintuitively, the reputational benefits of polarizing products, particularly early in a firm’s lifecycle, but only when paired with the ability to match with the right consumers.

The Good, the Bad and the Picky: Consumer Heterogeneity and the Reversal of Movie Ratings (with Ryan Stevens) [PDF]

We explore the consequences of consumer heterogeneity on product ratings. Consumers differ in their experience, which has two effects. First, experience is instrumental to choice: experts purchase (and thus review) better products than non-experts. Second, because of their superior choices, experts endogenously form higher expectations, and thus post more stringent ratings, for any given quality. Combined, these two forces imply that the better the product, the higher the standard it is held to, the more stringent its rating. Thus, relative ratings are biased: low quality products enjoy unfairly high ratings compared to their superior alternatives. When this bias gets large, reputation needs not be increasing in quality. The bias needs not disappear, and can worsen, over time: because it is mostly non-experts who rely on product reviews, products which received unfairly high ones will attract more of them, reinforcing their advantage. We test our theory by scraping data from a well known movie ratings website. We find strong evidence for both of our hypotheses, and that this bias is quantitatively important. We then debias the ratings, and find that the new ones better correlate with the opinions of external critics.

Range Effects in Multi-Attribute Choice: An Experiment (with Daniel Csába and Evan K. Friedman) [PDF]

Several behavioral theories suggest that, when choosing between multi-attribute goods, choices are context-dependent. Two theories provide such predictions explicitly in terms of attribute ranges. According to the theory of Focusing (Kószegi and Szeidl [2012]), attributes with larger ranges receive more attention. On the other hand, Relative thinking (Bushong et al. [2015]) posits that fixed differences look smaller when the range is large. It is as if attributes with larger ranges are over- and under-weighted, respectively. Since the two theories make opposing predictions, it is important to understand which features of the environment affect their relative prevalence. We conduct an experiment designed to test for both of these opposing range effects in different environments. Using choice under risk, we use a two-by-two design defined by high or low stakes and high or low dimensionality (as measured by the number of attributes). In the aggregate, we find evidence of focusing in low-dimensional treatments. Classifying subjects into focusers and relative thinkers, we find that focusers are associated with quicker response times and that types are more stable when the stakes are high.

Work in Progress

Is Ignorance Bliss? Information Sharing and Consumer Welfare

This paper theoretically studies the impact of consumer ratings on both welfare and the division of surplus in a two-sided market. Three features of the model are key. First, firms are strategic: they i) design their products anticipating the presence of consumer ratings, and ii) dynamically adjust their prices as information about them is revealed (as is the case on platforms like eBay, TaskRabbit and AirBnB, to name a few). Consumers can be either strategic or naïve in how they learn from others; ratings can be informative about both products' quality and consumers-products fit. I find that consumers can benefit from “knowing less”: while the optimal amount of information contained in product ratings is always positive, excessively precise ratings can harm consumers. This is because, on top of allowing consumers to make better purchases, more precise information increases high quality firms' market power, empowering them to charge substantially higher prices. Relatedly, consumer welfare is not increasing in the share of rational types. Third, aggregate firms’ welfare is also non-monotonic in the precision of ratings, while positive (negative) monotonicity can be established for the highest (lowest) quality firms. 

Attention Wanes: A Large-Scale Field Experiment on Online Reminders (with Gentry Johnson)

In this paper, we analyze the effects of online reminders on consumer behavior. We do so by partnering with an Italian startup to run large-scale field experiments involving over 10,000 reminders a day, over a span of six months. Methodologically, we employ novel machine learning methods adapted from Johnson et al. (2019) to aid in our experimental design. Specifically, we utilize both deep learning and recursive partitioning procedures (CART) to find and organize, respectively, latent spaces for stratification. These methods collapse pre-treatment outcomes and high-dimensional time-invariant covariates into a manageable low-dimensional space, and then construct optimal blocks over this space. Two issues are of special interest: first, heterogeneous treatment effects, with reminders incentivizing some users while demotivating others. Second, we disentangle potential mechanisms through which reminders might operate, and particularly memory and salience. The combination of our large pool of potential treated and control units and our novel design allow us to neatly separate these two channels. Our findings are of interest to managers, as they illustrate the optimal design of online communication strategies, as well as their dependence on customers' characteristics.

Measuring the Consequences of Reputation Inflation: Evidence from a Field Experiment (with Apostolos Filippas and John J. Horton)