Research

Abstract: How does hidden review manipulation reshape oligopoly market outcomes?  We model strategic reputation manipulation, including fake reviews and imprecise advertising, in price and quantity competitions. Each seller has private information about its product quality, which consumers need to infer before purchasing. Consumers observe noisy and seller-manipulated signals, such as product ratings, and distinguishing between authentic and manipulated reviews is imprecise, at best. We show that, when consumers rationally conjecture reputation manipulation, expected prices and quantities become invariant to a non-manipulation environment. Furthermore, due to the signaling role of manipulative behavior, expected consumer surplus increases. By contrast, when consumers naively believe observed signals are authentic, strategic substitutability emerges among sellers’ manipulative behavior. Then, equilibrium oligopoly market outcomes are distorted. Given the different informational roles of sellers' manipulating behavior, we demonstrate what kinds of market interventions improve or deteriorate consumer surplus depending on their rationality. 

Abstract: We compare Cournot and Bertrand competition in a differentiated duopoly in which firms privately observe product quality, invest in manipulative efforts to inflate their signals on qualities, and then compete in prices or quantities. Rational consumers anticipate equilibrium manipulation and discount signals accordingly. The following three main results were obtained: First, Bertrand competition induces higher quality sensitivity of manipulations because demand responds more strongly to reputation, whereas Cournot competition generates a higher expected level of manipulation because firms face weaker business stealing by rivals. Second, the classical consumer surplus ranking is preserved: Bertrand yields higher consumer surplus. In terms of the profits, we also show analytically that introducing a signal structure with manipulations by firms strictly reduces Cournot's profit advantage, although the ranking itself remains unchanged in the economically relevant parameter region, as numerical results confirm. Third, manipulative behaviors improve consumer welfare under rational beliefs by enhancing signal informativeness, as higher-quality firms manipulate more. These results indicate that the welfare-relevant dimension of manipulation is its quality sensitivity, not its expected level.

Abstract: In this paper, I theoretically analyze fake reviews on a platform market using models where a seller creates fake reviews through incentivized transactions, and its sales depend on its rating based on a review history. The platform can control the incentive for fake reviews by changing the parameters of the rating system, such as its filtering policy against suspicious reviews and weights for past reviews. At equilibrium, the number of fake reviews is increasing in the current quality and decreasing in the current reputation. Since fake reviews have a positive relationship with a product’s underlying quality, rational consumers might find a rating more informative when fake reviews exist, while credulous consumers suffer from a bias caused by boosted reputation. A stringent filtering policy can decrease the expected amount of fake reviews and the bias of credulous consumers, but at the same time, it can decrease the informativeness of reviews for rational consumers. In terms of the weight placed on the review history, rational consumers benefit from higher weights on past reviews than from optimal weights without fake reviews.

Abstract: This article theoretically analyzes the effects of the platform liability for the third-party products on incentives of the third-party vendors to make efforts to produce safe products. For this purpose, I suppose that the platform behaves as a mere marketplace and does not screen vendors or products sold there before the occurrence of a defect. Sellers have incentives to make an effort to build their own reputation. In this setting, the platform liability for third-party products has an unintended effect of reducing incentives for safe products. If the government sets a platform-liability rule for third-party products, the platform increases its monitoring efforts to detect defective products. However, the effectiveness of the liability rule depends on whether the sellers are building their reputation or not.

Abstract: We examined the effects of a rating system on sellers’ entry into and exit from a platform market characterized by monopolistic competition. We focused on the fact that the precision of a rating system affects the distribution of the products’ expected qualities in the marketplace. In the model, we assumed that the platform controls quality distribution in the monopolistic competitive marketplace.

Given the number of sellers (products), a more precise rating system designed by the platform always benefits consumers. By contrast, in a free-entry environment, a more precise rating system pushes out lower-quality sellers. This may decrease the consumer surplus. We provide some necessary and sufficient conditions for such a decrease in the consumer surplus.

In addition, in the absence of additional costs for improving the rating system, solutions of consumer surplus maximization and platform profit maximization coincide. That is, the platform favors consumers rather than sellers to gain profit. However, if a platform faces a convex cost function for improving the rating system, it underprovides the accuracy of the rating system.

Abstract: Assumptions of competitive structure are often crucial for marginal cost estimation and counterfactual predictions. This paper introduces tests for price competition among multi-product firms. The tests are based on the firm’s revealed preference (revealed profit function). In contrast to other approaches based on estimated demand functions such as conduct parameter estimation, the proposed tests do not require any instrumental variables, even though the models can accommodate structural error terms. In this paper, I employ a demand structure introduced by Nocke and Schutz (2018), the discrete/continuous choice model, which nests the multinomial logit demand and CES demand functions. Any price and quantity data can be rationalized by price competition under a discrete/continuous choice model and increasing marginal costs. Adding more assumptions to the demand functions, such as logit, CES, or the coevolving and log-concave property produces some falsifiable restrictions.

Abstract: This paper studies how the accuracy of online rating systems affects market competition and consumer welfare in a hybrid platform where the platform sells its own first-party (1P) products alongside third-party (3P) sellers. Unlike commission fees or search-ranking manipulation, rating accuracy does not directly steer consumers to 1P products. Still, platform’s incentives to improve rating accuracy are distorted by the existence of 1P products. The model captures accuracy as the dispersion of consumers’ beliefs on expected qualities across sellers, employs monopolistic competition with logit demand to summarize many small sellers into a single aggregate. The paper first shows that greater accuracy does not necessarily raise consumer surplus. It raises the chance of buying higher-quality products but also pushes lower-quality sellers out, reducing variety. While in a pure marketplace, profit and consumer-surplus maximization coincide without information costs, they diverge in a hybrid platform. The direction of divergence is characterized by the relative importance of 1P or 3P products and an effect of rating accuracy on the volume of marginal 3P sellers. Furthermore, when 1P products dominate, the platform’s incentive is almost reversed relative to consumers’ while they are largely aligned when 3P products dominate.

Abstract: We investigate which seller characteristics induce greater effort to enhance reputation via signal manipulation such as fake reviews on online marketplaces. In particular, we focus on (i) the absolute quality of the goods a seller provides, and (ii) the relative quality of those goods compared with the goods sold by another seller type in the market. In the experiment, we randomly assign one of a few quality-types to sellers. One type in a session is a relatively high-quality type, and another type in the same session is a relatively low-quality type. The absolute quality level determines the default signal distribution, and the realized signal determines the seller’s revenue. Each seller can manipulate their own signal distribution, but not that of the others, by paying a cost. We find that sellers assigned to the relatively low-quality type exert greater effort in the signal manipulation, and that this pattern becomes stronger as the quality disadvantage widens, even though the marginal monetary gains from the signal manipulation are identical across quality types. The results imply that the informativeness of product ratings on online marketplaces can be diminished by manipulations, such as fake reviews, as long as product qualities exhibit vertical differentiation.