Explore the vast range of titles available.

This paper investigates the economic value of online reviews for consumers and restaurants. We use a data set from Dianping.com , a leading Chinese website providing user-generated reviews, to study how consumers learn, from reading online reviews, the quality and cost of restaurant dining. We propose a learning model with three novel features: (1) different reviews offer different informational value to different types of consumers; (2) consumers learn their own preferences, and not the distribution of preferences among the entire population, for multiple product attributes; and (3) consumers update not only the expectation but also the variance of their preferences. Based on estimation results, we conduct a series of counterfactual experiments and find that the value from Dianping is about 7 CNY for each user, and about 8.6 CNY from each user for the reviewed restaurants in this study. The majority of the value comes from reviews on restaurant quality, and contextual comments are more valuable than numerical ratings in reviews.

In this paper, we develop a structural model of household behavior in an environment where there is uncertainty about brand attributes and both prices and advertising signal brand quality. Four quality signaling mechanisms are at work: (1) price signals quality, (2) advertising frequency signals quality, (3) advertising content provides direct (but noisy) information about quality, and (4) use experience provides direct (but noisy) information about quality. We estimate our proposed model using scanner panel data on ketchup. If price is important as a signal of brand quality, then frequent price promotion may have the unintended consequence of reducing brand equity. We use our estimated model to measure the importance of such effects. Our results imply that price is an important quality-signaling mechanism and that frequent price cuts can have significant adverse effects on brand equity. The role of advertising frequency in signaling quality is also significant, but it is less quantitatively important than price. In the printed version of Marketing Science , Vol. 27, No. 6, Erdem et al. (2008) was mistakenly identified as a Research Note. It is a regular article and has been corrected here and in the online table of contents.

We develop a structural dynamic demand model that examines how brand preferences evolve when consumers are uncertain about product quality and their needs change periodically. We allow for strategic sampling behavior of consumers under quality uncertainty and allow for strategic sampling to increase periodically as consumers’ needs change periodically. We differ from previous work on forward-looking consumer Bayesian learning by allowing for 1) spill-over learning effects across different versions of products or products in different product categories that share a brand name and 2) duration-dependence in utility for a specific version of a product or product class to capture systematic periodic changes in consumer utility and migration of consumers across product versions or classes. We also assess the evolution of price elasticities in markets where there is consumer quality uncertainty that diminishes over time as consumers get more experienced. We estimate our model using scanner data for the disposable diapers category and discuss the consumer behavior and managerial implications of our estimation and policy simulation results.

We consider the consumer problem under uncertainty when the consumer can choose the quantity of a risk-free good and the lottery, or distribution, of a risky good from a set of distributions. These goods are imperfect substitutes in the consumer preferences, with additive preferences a special case. We develop sufficient conditions for the choice of the risky good to be monotone with respect to income, exploring different notions of monotonicity. The sufficient conditions are ordinal, independent of concavity, and do not require differentiability or continuity. Cardinal conditions and conditions from the single good case are not necessary and are not always sufficient. The sufficient conditions are formulated in appropriate value lattices. The framework is flexible and adaptable to handle different uncertainty applications. Examples demonstrate the sufficient conditions and different applications where available lotteries may be finite in number, may have discrete support, or may form a chain or a lattice.

We develop a model in which reward-based crowdfunding enables firms to obtain a reliable proof of concept early in their production cycle: they learn about total demand from a limited sample of target consumers preordering a new product. Learning from the crowdfunding sample creates a valuable real option because firms invest only if updated expectations about total demand are sufficiently high. This is particularly valuable for firms facing a high degree of uncertainty about consumer preferences, such as developers of innovative consumer products. Learning also enables firms to overcome moral hazard. The higher the funds raised, the lower the firms’ incentives to divert them, provided third-party platforms limit the sample size by restricting campaign length. Although the probability of campaign success decreases with sample size, the expected funds raised are maximized at an intermediate sample size. Our results are consistent with stylized facts and lead to new empirical implications. This paper was accepted by Gustavo Manso, finance.

Although advance selling and probabilistic selling differ in both motivation and implementation, we argue that they share a common characteristic-both offer consumers a choice involving buyer uncertainty. We develop a formal model to examine the general economics of purchase options that involve buyer uncertainty, explore the differences in buyer uncertainty created via these two strategies, and derive conditions under which one dominates the other. We show that the seller can address unobservable buyer heterogeneity by inducing sales involving buyer uncertainty via two different mechanisms: (1) homogenizing heterogeneous consumers and (2) separating heterogeneous consumers. Offering advance sales encourages customers to purchase while they are uncertain about their consumption states (more homogeneous), but offering probabilistic goods encourages customers to reveal their heterogeneity via self-selecting whether or not to purchase the uncertain product. The relative attractiveness of these two selling strategies depends on the degree of two types of buyer heterogeneity: (1) Max_Value-Heterogeneity , which is the variation in consumers' valuations for their preferred good, and (2) Strength-Heterogeneity , which is the variation in the strength of consumers' preferences. Neither strategy is advantageous unless the market exhibits sufficient Max_Value-Heterogeneity . However, whereas Strength-Heterogeneity can destroy the profit advantage of advance selling, a mid-range of Strength-Heterogeneity is necessary for probabilistic selling to be advantageous.

Traditional models of insurance choice are predicated on fully informed and rational consumers protecting themselves from exposure to financial risk. In practice, choosing an insurance plan is a complicated decision often made without full information. In this paper we combine new administrative data on health plan choices and claims with unique survey data on consumer information to identify risk preferences, information frictions, and hassle costs. Our additional friction measures are important predictors of choices and meaningfully impact risk preference estimates. We study the implications of counterfactual insurance allocations to illustrate the importance of distinguishing between these micro-foundations for welfare analysis.

The standard revealed preference argument relies on an implicit assumption that a decision maker considers all feasible alternatives. The marketing and psychology literatures, however, provide well-established evidence that consumers do not consider all brands in a given market before making a purchase (Limited Attention). In this paper, we illustrate how one can deduce both the decision maker's preference and the alternatives to which she pays attention and inattention from the observed behavior. We illustrate how seemingly compelling welfare judgments without specifying the underlying choice procedure are misleading. Further, we provide a choice theoretical foundation for maximizing a single preference relation under limited attention.

We propose a structural model to study the effect of online product reviews on consumer purchases of experiential products. Such purchases are characterized by limited repeat purchase behavior of the same product item (such as a book title) but significant past usage experience with other products of the same type (such as books of the same genre). To cope with the uncertainty in quality of the product item, we posit that consumers may learn from their experience with the same type of product and others' experiences with the product item. We model the review credibility as the precision with which product reviews reflect the consumer's own product evaluation. The higher the precision, the more credible the information obtained from product reviews for the consumer, and the larger the effect of reviews on the consumer's choice probabilities. We extend the Bayesian learning framework to model consumer learning on both product quality and review credibility. We apply the model to a panel data set of 1,919 book purchases by 243 consumers. We find that consumers learn more from online reviews of book titles than from their own experience with other books of the same genre. In the counterfactual analysis, we illustrate the profit impact of product reviews and how it varies with the number of reviews. We also study the phenomenon of fake reviews. We find that fake reviews increase consumer uncertainty. The effects of more positive reviews and more numerous reviews on consumer choice are smaller on online retailing platforms that have fake product reviews.

It is ultimately an empirical question whether risk preferences are stable over time. The evidence comes from diverse strands of literature, covering the stability of risk preferences in panel data over shorter periods of time, life-cycle dynamics in risk preferences, the possibly long-lasting effects of exogenous shocks on risk preferences as well as temporary variations in risk preferences. Individual risk preferences appear to be persistent and moderately stable over time, but their degree of stability is too low to be reconciled with the assumption of perfect stability in neoclassical economic theory. We offer an alternative conceptual framework for preference stability that builds on research regarding the stability of personality traits in psychology. The definition of stability used in psychology implies high levels of rank-order stability across individuals and not that the individual will maintain the same level of a trait over time. Preference parameters are considered as distributions with a mean that is significantly but less than perfectly stable, plus some systematic variance. This framework accommodates evidence on systematic changes in risk preferences over the life cycle, due to exogenous shocks such as economic crises or natural catastrophes, and due to temporary changes in self-control resources, emotions, or stress. We note that research on the stability of (risk) preferences is conceptually at the heart of microeconomics and systematic changes in risk preferences have vital real-world consequences.