No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Observational Learning and Demand for Search Goods
Abstract
We develop a model of herds in which consumers observe only the aggregate purchase history, not the complete ordered history of search actions. We show that the purchasing information changes the conditions under which herds can occur for both low- and high-quality products. Inferior products are certain to be ignored; high quality products may be ignored, but complete learning may also occur. We obtain closed form solutions for the probabilities of these events and conduct comparative statics. We test the model's predictions using data from an online music market created by Salganik, Dodds, and Watts (2006).
... Importantly, 2 These observations lead us to believe that the simultaneous examination of alternatives used in some previous models (Borghesi and Bouchaud, 2007) is an oversimplification. Similarly, modeling the process as a choice of a single good (Hendricks et al., 2012) is also oversimplifying since decisions might depend on alternatives examined previously. 3 An early model of this kind was put forward by Simon (1955a) as a boundedly rational alternative to omniscient utility-maximizing agents. ...
... Instead, the agents behave as naive statisticians, not questioning where their sample comes from (Fiedler and Juslin, 2006) but they still make rational decisions based on their sample (Denrell and Le Mens, 2007). This approach is in line with the observations from the Music Lab experiment (Krumme et al., 2012 Arthur, 1989; Banerjee, 1992; Hendricks et al., 2012) and illustrates how these can be operationalized in multi-alternative choice environments. However, it goes beyond these models as it illustrates that a simple cognitive mechanism — limited sequential search — captures the emerging properties of the market such as skewed popularity distributions , outcome unpredictability and the imperfect relation between perceived quality and popularity better than previous theoretical accounts. ...
... Alternatives with a high objective utility component stand a better chance of reaching these positions, which through social interactions may attract many more consumers over time. Hendricks et al. (2012) reached the same result with their model, but we show that it holds in much richer setting with within agent sequentiality. This setting led us to some novel insights on the inner workings of the market. ...
This dissertation comprises of three independent essays which introduce novel psychologically inspired process models and examine their implications for individual, collective or market behavior. The first essay studies multi-attribute choice as a guided process of search. It puts forward a theoretical framework which integrates work on search and stopping with partial information from economics with psychological subjective utility models from the field of judgment and decision making. The alternatives are searched in order of decreasing estimated utility, until the expected cost of search exceeds the relevant benefits; The essay presents the results of a performance comparison of three well-studied multi-attribute choice models.The second essay reports the results of two experiments designed to understand how people revise their judgments of factual questions after being exposed to the opinion and confidence levels of others. It introduces a tree model of judgment revision which is directly derived from the empirical observations. The model demonstrates how opinions in a group of interacting people can converge or polarize over repeated interactions. The third essay, studies collective behavior in markets for search products. The decision makers consider the alternatives in order of decreasing popularity and choose the first alternative with utility higher than a certain satisficing threshold. The popularity order is updated after each individual choice. The presented framework illustrates that such markets are characterized by rich-get-richer dynamics which lead to inequality in the market-share distribution and unpredictability in regard to the final outcome.
... Pollock et al. (2015) show that participating in blockbuster deals benefits well-known, high-status venture capital firms less than other venture capital firms. Salganik et al. (2006) showed that the long-term popularity of high-quality songs was less affected by the early popularity and attention than medium-quality songs were (Hendricks et al. 2012). In these settings, agents can succeed by having high quality and few resources or by having medium quality and high resources. ...
... We predict that songs that were not displayed prominently early on but nevertheless became popular are of higher quality than songs that were prominently displayed early on and became popular. We expect this effect based on past work showing that medium-quality songs benefit more from early attention than high-quality songs do (Salganik et al. 2006, Hendricks et al. 2012, suggesting β > 0. The setup in the Salganik study differs from our two-period model, because total song downloads is the result of the decisions of many participants. Still, simulations show that a multiperiod version of our model, where one participant makes a download choice in each period, predicts that (a) market share early on is positively correlated with quality, (b) final market share is positively correlated with quality, but (c) early market share has a negative impact on predicted quality in a regression that includes both early and final market share. ...
Success tends to increase and failure tends to decrease the chances of future success. We show that this impact of past outcomes can change how diagnostic success or failure are regarding the competence of an individual or a firm. Succeeding under adverse circumstances is especially impressive when initial failure reduces the chances of success more for low-quality agents than for high-quality agents. Succeeding after initial failure (being a “successful underdog”) can also indicate higher expected quality than succeeding twice if initial success increases the chances of success of all agents to a high level. In different circumstances, the outcome after success can be especially informative about quality, implying that failing after an initial success (a “one-hit wonder”) indicates lower quality than failing twice does. We find effects consistent with our model in data on Canadian professional hockey players and on data from the Music Laboratory experiment: Initial failure combined with eventual success is associated with high quality. The results help to clarify when failure should be attributed to the person in charge or to the situation, when underdogs and individuals who overcome adversity are especially impressive and when a naïve “more is better” heuristic for evaluating performance can be misleading.
This paper was accepted by Isabel Fernandez-Mateo, organizations.
Supplemental Material: The data files and online appendix are available at https://doi.org/10.1287/mnsc.2022.4630 .
... Also using the music industry as an application, Hendricks and Sorensen (2009a) show how consumer awareness and information affect purchasing 1 Accordingly, record labels (and artists) have close ties to radio stations (Vogel, 1986;Caves, 2000). behavior and the success of products; this may depend on gathering external information (Cai et al., 2009;Tucker and Zhang, 2011;Hendricks et al., 2012). Firms can influence the information available to consumers by their activities including advertising. ...
... First, our finding of the non-linear effect of competition on amount of advertising adds product-level evidence to the classic literature (Sutton, 1974;Reekie, 1975;Rees, 1975;Brush, 1976;Ehrlich and Fisher, 1982) which studied competition only at the industry level. Our findings provide further evidence that consumer awareness of products affects sales in certain industries (Cai et al., 2009;Hendricks and Sorensen, 2009a;Tucker and Zhang, 2011;Hendricks et al., 2012) and shows that firms do take this into account in their strategy. Moreover, it adds to the literature on competition and product introduction (Bayus et al., 1997;Boyd and Bresser, 2008;Ethiraj and Zhu, 2008;Lee et al., 2000); it also shows, how the introduction of products (singles) was affected by digitization. ...
In markets with thousands of products, firms cannot take it for granted that con-sumers are even aware of their articles’ existence. Advertising and actions to attract consumer attention are therefore integral components of a firm’s competitive toolbox. We study firms’ behavior in a perfect example for such a market: The music indus-try, in which consumers can choose from a plethora of albums and songs. We study a specific strategic instrument of firms, single releases, applying unique micro-level data. Arguing that the digitization of the industry via MP3, filesharing, and iTunes amounts to forced unbundling, the role of singles has changed from individual rev-enue generators (pre-digital era) to pure attention gatherers. In accordance with this driving hypothesis, we observe an inverse U-shaped relationship between competition intensity and the number of singles released in the digital era, while previously com-petition had a purely negative effect.
... Bose et al. [6] study how a monopolist can maximize revenue by jointly optimal dynamic pricing and information controls. Hendricks et al. [11] develop a model of herds where consumers observe only the aggregate purchasing history. For reputation games, e.g., Board and Meyer-ter Vehn [5] , the quality uncertainty is due to the seller's unobservable actions. ...
... For reputation games, e.g., Board and Meyer-ter Vehn [5] , the quality uncertainty is due to the seller's unobservable actions. Most of these models, except Hendricks et al. [11] , require the observations of complete history of predecessors' behavior, which is usually not applicable on the Internet platform. In our model, consumer reviews would be summarized in sufficient statistics. ...
Consumer reviews have become pervasive for e-commerce in recent years, especially for electronic products. In this paper, we investigate the optimal pricing strategies for a platform selling electronic products when consumers sequentially learn about product quality from consumer reviews. We focus on the transient analysis to calibrate how information externalities across the time dimension would distort the seller’s optimal pricing strategies. Facing the “cold start” problem, the seller of high-quality products would choose lower prices to speed up the consumer learning process. Consequently, the optimal prices suffer from downward distortions that increase in product quality in this reputation-riding regime.
In the extensions, we propose a tractable and flexible framework to support both operational and strategic decision-making processes. We explore the value of persuasive advertisement, and the results suggest that consumer reviews and marketing efforts are strategic substitutes. In terms of quality control, we find that it would be optimal to invest in quality in the early stages, but stop at a certain time threshold, resulting in a reputation-building and reputation-spending pattern. Finally, we extend the framework to study a duopoly pricing problem. We show that the high-quality seller could strategically accommodate the low-quality seller in the early stages, and wages a price war at later stages.
... Recent reanalyses of the experimental data suggest that participants made their listening (i.e., sampling) decisions based on popularity. But their evaluations of the songs were not driven by social influence: there was no effect of popularity once sampling was controlled for (Hendricks et al. 2012, Krumme et al. 2012). ...
... Because our model does not assume that information about the majority opinion directly influences quality estimates, it may be more suitable for settings in which such an assumption is not warranted. First, individuals may have different tastes or skills and may not believe that others know best (e.g., Banerjee et al. 2013, Hendricks et al. 2012, Munshi 2004. In situations where tastes and skills differ between people, it makes little sense for an individual to make an inference about how good an alternative is for her on the basis of popularity. ...
We demonstrate that a sampling-based mechanism can offer an alternative explanation for belief synchronization in social groups and the persistence of collective illusions. Our model assumes that people are more likely to sample popular alternatives than unpopular alternatives. We show that this mechanism is sufficient to explain belief synchronization: a strong majority of opinions will likely emerge in favor of one alternative. The reason is that the group is unlikely to move away from a state in which one alternative is very unpopular. If by chance most people come to dislike alternative A, they are all unlikely to sample it again and their opinions of A remain the same. When A is in fact the best alternative, a collective illusion has emerged because people mistakenly believe that a suboptimal alternative is the best. Our model implies that such a collective illusion is persistent. The model thus offers an existence proof that a collective illusion can occur even in settings where people do not infer that popular alternatives are better. The model also casts new light on the effect of online recommendation systems on attitude homogenization and the effect of majority voting on beliefs and attitudes.
This paper was accepted by Yuval Rottenstreich, judgment and decision making.
... Social influence enters as an availability parameter that controls for the probability that an item will be sampled given its position on the screen. Finally, Hendricks, Sorensen, and Wiseman (2012), leveraging the insights from informational cascade models (Banerjee, 1992;Bikhchandani, Hirshleifer, & Welch, 1992), developed a model in which agents with heterogeneous preferences follow a simple search model to decide whether to sample and then buy each alternative. By breaking down the process into two steps -a sampling and a buying step -the last two modeling accounts have made significant progress. ...
... We assumed that the agents would act as they do with a bestseller list and would start by examining the alternative that has been selected by most agents and move down to less popular alternatives. Conceptually, our model incorporates some of the notions of existing social influence models such as sequentiality among agents, threshold rules, and heterogeneity in preferences (Granovetter 4 & Soong, 1986;Arthur, 1989;Banerjee, 1992;Hendricks et al., 2012) and illustrates how these can be operationalized in multi-alternative choice environments. However, it goes beyond these models as it illustrates that a simple cognitive mechanism -limited sequential search -captures the patterns of collective behavior, such as skewed popularity distributions, outcome unpredictability, and the imperfect relation between perceived quality and popularity, better than previous theoretical accounts. ...
A few “hit” cultural products tend to dominate consumer attention. Yet, it is notoriously hard to predict ex ante which products will become “hits”. What are the decision-making processes that lead to these patterns of collective behavior? We advance a novel process model in which agents with diverse yet correlated preferences search the alternatives in order of popularity and choose the first alternative with utility higher than a certain satisficing threshold. The model goes beyond existing accounts of the popularity dynamics in that (i) it suggests a cognitive process through which social influence plays out in these markets, (ii) it allows us to study how inequality and unpredictability in the market change as a function of the diversity of preferences in the consumer population and the satisficing threshold, (iii) it is amenable to welfare analysis, and (iv) it facilitates comparisons with scenarios without social influence. In agent-based simulations we found that social influence led to an increase in inequality and unpredictability in the market, especially when agents employed a low satisficing threshold. In addition, we found that social influence led to a larger increase in the average consumer welfare when there was at least some diversity of preferences among consumers.
... Our model belongs more generally to the literature on observational learning, in which individuals sequentially choose an action after seeing predecessors' choices (Banerjee, 1992;Çelen and Kariv, 2004;Garcia and Shelegia, 2018;Guarino and Jehiel, 2013;Hendricks et al., 2012). Within this literature, it contributes to a subset of papers looking at the effect of information -in our case on previous contributions -on contributions to a public good (Figuieres et al., 2012;Tajika, 2020). ...
We generalize the model of Gallice and Monzón (2019) to incorporate a public goods game with groups, position uncertainty, and observational learning. Contributions are simultaneous within groups, but groups play sequentially based on their observation of an incomplete sample of past contributions. We show that full cooperation between and within groups is possible with self-interested players on a fixed horizon. Position uncertainty implies the existence of an equilibrium where groups of players conditionally cooperate in the hope of influencing further groups. Conditional cooperation implies that each group member is pivotal, so that efficient simultaneous provision within groups is an equilibrium.
... When observational learning is based only on popularity, consumers do not know the rationales behind others' purchase decisions. Hendricks et al. (2012) suggest that observational learning only driven by popularity may lead consumers to ignore superior products and purchase inferior products. ...
Purpose
To investigate the psychological mechanism of observational learning in the online retailing context, the purpose of this paper is to show how the psychological distance between consumers and products affects modes of observational learning.
Design/methodology/approach
Five experimental studies are conducted to test the hypotheses.
Findings
The findings show that which modes of observational learning are adopted by consumers is affected by consumers’ psychological distance. Specifically, when the psychological distance between consumers and products is proximal, consumers tend to adopt the termed adequate observational learning mode by considering the interaction of information about popularity and the breadth of appeal of a product to make purchase. However, when the psychological distance is distal, consumers would consider information of popularity and breadth of appeal separately without considering the interaction, termed as inadequate observational learning mode. The observed relationship between psychological distance and observational learning mode could be explained by the construal level.
Research limitations/implications
This research advances the observational learning and psychological distance literature by investigating the psychological mechanism behind observational learning modes. Limitations include the use of scenario-based experiments to test the hypotheses, investigation of a single product attribute (i.e. breadth of appeal) and assessment of popularity information by sales volume alone.
Practical implications
The current research provides a deeper understanding of consumer observational learning modes, which can help online retailers to develop effective product strategies and marketing tactics and, finally, achieve stronger competitive positions.
Originality/value
The present research contributes to the literature by examining the psychological mechanism involved in observational learning. This research distinguishes adequate and inadequate observational learning modes from the perspective of psychological distance.
... 3 In contrast to our model, agents know their private types and there is no information choice: private signals are given exogenously. Similar to our framework, in Hendricks et al. (2012), the payoff of agents depends on the sum of a common and an idiosyncratic component. In contrast to our framework, agents observe only the aggregate purchase history and decide whether to learn their payoff perfectly or not at all (thus, the question of how much to learn about which component does not arise). ...
... However, customers may not always care about the nature of products. Hendricks et al. discovered that, in an on-line music market, consumers rely on the purchase history of others and are unwilling to incur searching costs to avail of higher quality products [12]. As a result, a poor purchasing record may lead consumers to erroneously ignore high quality products. ...
... Second, my article changes the object of learning as compared to adverse-selection search models with observational learning. In models such as Kircher and Postlewaite (2008), Hendricks et al. (2012) and Garcia and Shelegia (2015), a buyer learns about a single seller's option from others' behaviour, while in my model he learns about the aggregate state of the market. 3 Third and most importantly, my model adds payoff externalities, which generate intricate learning features, as compared to other adverse-selection search models and financial market models with learning. ...
I study stationary cut‐off‐strategy equilibria of a dynamic market model where buyers sample sellers sequentially from an unknown distribution. Buyers learn about the distribution from the sampled sellers and a ‘trade signal’. The trade signal reveals whether a randomly chosen seller traded yesterday. Observing a trade (as opposed to no trade) is good news about the distribution. Buyers who observe a trade use a higher cut‐off than buyers who observe no trade, despite buyers’ learning from sampled sellers that puts a countervailing pressure on the cut‐offs. The trade signal may reduce market efficiency, while an appropriate exogenous signal increases efficiency.
... However, customers may not always care about the nature of products. Hendricks et al. discovered that, in an on-line music market, consumers rely on the purchase history of others and are unwilling to incur searching costs to avail of higher quality products [12]. As a result, a poor purchasing record may lead consumers to erroneously ignore high quality products. ...
Product configurators have been acknowledged as an important enabling toolkit to realise mass customisation. During the configuration process, customers are presented with an operational list of predefined attributes and their variants. They can choose between different variants in accordance with their preferences. The combination of their choices constitutes their desired specification and thus the customised product. This configure-to-order process has been implemented in various industries. Currently, some on-line configurators provide extra information on attribute variants to improve the information available to customers, such as popularity or bestseller ratings. This information is derived from the purchase behaviours of other customers. However, it is not clear how such information affects consumer decision making. Drawing on the notion of observational learning, it could be posited that customers tend to follow the choices made by others, particularly when they lack adequate prior knowledge about products. However, one of the advantages of customisation is that it enables customers to harness and reveal their own identities by configuring and purchasing a bespoke unique product that is tailored to their own desires. Thus, customers may not be willing to select the popular variants during configuration process. To solve this paradox, we empirically find that observational learning is significant in the product configuration process. Customers' abilities and motivations to process information mediate this observational learning effect.
... Kircher and Postlewaite show that equilibria may arise where high-quality firms offer a discount to those consumers who search more actively, and those who search less actively follow their advice. Hendricks, Sorensen, and Wiseman (2012) introduce a model of observational learning with multiple types in the spirit of Smith and Sørensen (2000), where each consumer has to decide between acquiring a costly signal about the quality of a single good, buying it immediately, or not buying. They focus on the long-run dynamics of sales for high-and low-quality products and the possibility of "bad" herds arising. ...
This article studies observational learning in a consumer search environment. Consumers observe the purchasing decision of a predecessor with similar preferences. Consumers rationally emulate by initiating their search at the firm from which their predecessor purchased, free-riding on search effort, and reacting less to price changes. Prices are nonmonotone in search costs and may be as low as marginal costs. We discuss several extensions and show that the effect of emulation on prices is stronger when (i) the number of firms increases, (ii) consumers' first visits are more elastic with respect to market shares, and (iii) prices are adjusted more frequently.
... First, our finding of the non-linear effect of competition on the amount of advertising adds product-level evidence to the classic literature ( Sutton, 1974;Reekie, 1975;Rees, 1975;Brush, 1976;Ehrlich and Fisher, 1982 ) which studied competition only at the industry level. Our findings provide further evidence that consumer awareness of products affects sales in certain industries ( Cai et al., 2009;Hendricks and Sorensen, 2009;Tucker and Zhang, 2011;Hendricks et al., 2012 ) and shows that firms do take this into account in their strategy. Perhaps most importantly, we contribute to the literature on the economics of digitization ( Greenstein et al., 2010 ), by showing that digitization, through forced unbundling, has made the music market move faster. ...
In markets with long tails and thousands of products, like recorded music, consumers cannot possibly be aware of every product. We analyze how record labels use single releases as a strategic instrument to attract consumer attention in a competitive environment. In particular, we study how the advent of digitization has changed firm strategy. In accordance with predictions from a simple theoretical model, we show that record labels release more singles with shorter intervals in between when facing greater competitive pressure. We show that this effect is stronger in the digital age. With individual songs becoming readily available (forced unbundling), the attention generation motive becomes predominant and single releases more closely resemble a form of advertising.
... As Salganik et al. (2006, p. 855) put it, in the treatment worlds, "the best songs never do very badly, and the worst songs never do extremely well, but almost any other result is possible." Hendricks et al. (2012) argue that, from a social welfare perspective, this unpredictability of outcomes is the most important aspect of observational learning. In markets with large numbers of products, where observational learning mechanisms appear to be most influential, the key question is whether observational learning leads consumers to ignore superior products or waste time and resources learning about inferior products. ...
Innovations in information technology have increased the prevalence of markets with large numbers of products. Each week, for example, an average of over 800 books are published, and an average of over 1,100 iOS apps are released in Apple's App Store. This review summarizes existing research about how consumers learn about new products in such markets, focusing on mechanisms like bestseller lists and user-generated product reviews. In addition to reviewing research findings about how these mechanisms directly influence sales, this article also discusses how these mechanisms affect broader market outcomes, such as the shape of the success distribution and the likelihood that good products are discovered.
... nnus, Salganik et al. [2006] isolent l'impact des compteurs de téléchargement sur les choix musicaux des individus et montrent que le succès d'un titre dépend fortement des informations fournies sur sa popularité. Dans un marché avecDeux modèles théoriques ont été proposés pour comprendre le mécanisme sous-jacent à ce processus d'influence sociale.Hendricks et al. [2012] supposent que les individus ont tendance à rechercher des produits déjà achetés par d'autres consommateurs afin de diminuer le coût et le temps de leur recherche. Leurs résultats suggèrent que, sur le long terme, si des biens de ...
An Economic Analysis of the Long Tail Phenomenon in Cultural Markets
Since the 1990s, digital technologies have transformed the value chain in cultural industries by introducing new ways of creating, distributing and promoting information goods. The concept of the Long Tail (Anderson [2004]) states that the development of Internet and e-commerce should lead to an increase in market share of less popular products (i.e., niche products), whose sales are expected to generate profits at least comparable to that of bestsellers. In this paper, we propose an economic analysis of the Long Tail. We first present the mixed empirical results of the literature on the magnitude of the Long Tail phenomenon in cultural markets, and then analyze the economic mechanisms that can contribute to its development. We also discuss the implications of the Long Tail for firms’ strategies in cultural industries.
... Another branch of the literature introduces a costly and strategic choice into the decision process -each agent can pay to acquire an informative signal, or to "search", i.e., sample an available option and know its value. Notable works in this area include Hendricks, Sorensen and Wiseman [20], Mueller-Frank and Pai [25] and Ali [2]. My paper differs from this stream of the literature in two aspects. ...
I study the problem of social learning in a model where agents move sequentially. Each agent receives a private signal about the underlying state of the world, observes the past actions in a neighborhood of individuals, and chooses her action attempting to match the true state. Earlier research in this field emphasizes that herding behavior occurs with a positive probability in certain special cases; recent studies show that asymptotic learning is achievable under a more general observation structure. In particular, with unbounded private beliefs, asymptotic learning occurs if and only if agents observe a close predecessor, i.e., the action of a close predecessor reveals the true state in the limit. However, a prevailing assumption in these studies is that the observation structure in the society is exogenous. In contrast to most of the previous literature, I assume in this paper that observation is endogenous and costly. More specifically, each agent must pay a specific cost to make any observation and can strategically choose the set of actions to observe privately. I introduce the notion of maximal learning (relative to cost) as a natural extension of asymptotic learning: society achieves maximal learning when agents can learn the true state with probability 1 in the limit after paying the cost of observation. I show that observing only a close predecessor is no longer sufficient for learning the true state with unbounded private beliefs and positive costs. Instead, maximal learning occurs if and only if the size of the observations extends to infinity. I provide interesting comparative statics as to how various settings in the model affect the learning probability. For instance, the probability to learn the true state may be higher under positive costs than under zero cost; in addition, the probability to learn the true state may be higher under weaker private signals.
... First, our finding of the nonlinear effect of competition on the amount of advertising adds product-level evidence to the classic literature (Sutton, 1974;Reekie, 1975;Rees, 1975;Brush, 1976;Ehrlich and Fisher, 1982) which studied competition only at the industry level. Our findings provide further evidence that consumer awareness of products affects sales in certain industries (Cai et al., 2009;Hendricks and Sorensen, 2009;Tucker and Zhang, 2011;Hendricks et al., 2012) and shows that firms do take this into account in their strategy. Perhaps most importantly, we contribute to the literature on the economics of digitization (Greenstein et al., 2010), by showing that digitization, through forced unbundling, has made the music market move faster. ...
In markets with long tails and thousands of products, like recorded music, consumers cannot possibly be aware of every product. We analyze how record labels use single releases as a strategic instrument to attract consumer attention in a competitive environment. In particular, we study how the advent of digitization has changed firm strategy. In accordance with predictions from a simple theoretical model, we show that record labels release more singles with shorter intervals in between when facing greater competitive pressure. We show that this effect is stronger in the digital age. With individual songs becoming readily available (forced unbundling), the attention generation motive becomes predominant and single releases more closely resemble a form of advertising.
... Information is of special importance in markets with experience goods, where product quality cannot be completely assessed prior to consumption. Consumers collect external information from popularity rankings (Tucker and Zhang, 2011;Hendricks et al., 2012), recommendations (Oestreicher-Singer and Sundararajan, 2012;Dewan and Ramaprasad, 2012) and from related products whose quality attributes are already known (Hendricks and Sorensen, 2009). Firms advertise to inform consumers about product quality. ...
Sampling poses an interesting problem in markets with experience goods. Free samples reveal product quality and help consumers to make informed purchase decisions (promotional effect). However, sampling may also induce consumers to substitute purchases with free consumption (displacement effect). We study this trade-off in the market for digital music where consumers can sample the quality of songs by watching free music videos online. Identification comes from a natural experiment in Germany, where virtually all videos that contain music are blocked on a popular video platform due to a legal dispute with representatives of the rights-holders. We show that promotional and displacement effects cancel out in the sales performance of individual songs, whereas online music videos trigger sales of albums.
Consumers commonly rely on electronic word‐of‐mouth (eWOM) to inform their purchase decisions. The purpose of this study is to investigate when and why repeat purchase information in eWOM influenced customers' purchase behavior, using the social learning theory as a framework. Three experiments were conducted to test the hypotheses. The results showed that repeat purchase information in eWOM positively affected consumers' purchase intention through perceived diagnosticity of eWOM and perceived value. Additionally, the effect varied from product type. For utilitarian products, the repeat purchase information in eWOM had a significant impact on consumers' perceived diagnosticity and purchase intention, while for hedonic products, the impact was not significant. The current study deepens the understanding of social learning theory in eWOM content by providing a novel look at repeat purchase information in eWOM. These findings offer important implications for both research and practical applications of eWOM, particularly in term of effectively utilizing repeat purchase information of eWOM.
Research Summary
This article studies how the entry of an imitative product influences the demand for the original, in markets with a large number of products where consumers are not aware of most products. We suggest that the release of an imitative product triggers two countervailing forces: a discovery effect that increases awareness and demand for the original, and a substitution effect decreasing that demand. When the imitation is horizontally differentiated, the substitution effect is weaker, and the discovery effect leads to increased demand for the original, particularly so when the original is less well‐known. However, when the imitation is vertically differentiated, the discovery effect does not benefit the original, and the demand decreases given the substitution effect. We test our theory in the context of 3D‐printable products.
Managerial Summary
Imitative products typically harm the performance of the original product. This is in part the rationale for deterring imitation by acquiring intellectual property (IP) rights. However, as we demonstrate, the release of an imitative product may increase discovery of the original product, and therefore demand for it. In markets with a large number of products, products typically compete for attention, and an imitative product may serve as a channel through which attention spills over from the imitation to the original. This effect is more pronounced when the original product is less well‐known. This suggests that, under certain conditions, firms may benefit from pursuing a more open IP strategy that encourages imitation. Examples of marketplaces where imitation may increase demand for the original include those for music, e‐books, software, and mobile apps.
We model a dynamic public good contribution game, where players are (naturally) formed into groups. The groups are exogenously placed in a sequence, with limited information available to players about their groups' position in the sequence. Contribution decisions are made by players simultaneously and independently, and the groups' total contribution is made sequentially. We try to capture both inter and intra-group behaviors and analyze different situations where players observe partial history about total contributions of their predecessor groups. Given this framework, we show that even when players observe a history of defection (no contribution), a cooperative outcome is achievable. This is particularly interesting in the situation when players observe only their immediate predecessor groups' contribution, where we observe that players play an important role in motivating others to contribute.
Consumer learning prior to and after the purchase, with endogenous pricing, leads to greater search prior to the purchase for more durable products.
We analyze a social learning model where the agents' utility depends on a common component and an idiosyncratic component. Each agent splits a learning budget between the two components. We show that information about the common component is fully aggregated if and only if agents do not have to sacrifice learning about their idiosyncratic component in order to learn marginally about the common component. If agents vary in how much they value their idiosyncratic component, then the order of agents can strictly impact how much information is aggregated.
We review the theory of information cascades and social learning. Our goal is to describe in a relatively integrated and accessible way the more important themes, insights and applications of the literature as it has developed over the last thirty years. We also highlight open questions and promising directions for further theoretical and empirical exploration.
We review the theory of information cascades and social learning. Our goal is to describe in a relatively integrated and accessible way the more important themes, insights and applications of the literature as it has developed over the last thirty years. We also highlight open questions and promising directions for further theoretical and empirical exploration.
The unprecedented growth of social network users in the last decade has resulted in significant increases in the availability of individual-specific information such as holiday pictures, mobile check-ins at restaurants, and information on everyday purchases. Consumers shopping through social network channels are increasingly using this information in making their purchase decisions. We find that social ties impact the magnitude of observational learning. In the case of strangers, the effect of learning is stronger for vertically differentiated products than for horizontally differentiated products; whereas in the case of friends, the effect of learning for vertically differentiated products is similar to that for horizontally differentiated products. Moreover, the type of product impacts the magnitude of observational learning. For horizontally differentiated products, the effect of learning from friends is stronger than that from strangers; whereas for vertically differentiated products, the effect of learning from friends is similar to that from strangers. These findings provide motivation for online retailers to generate alternative strategies for increasing product sales through social networks. For example, online retailers offering horizontally differentiated products have strong incentives to cooperate with social media platforms (e.g., Instagram and Pinterest) in encouraging customers to share their purchase information.
We consider a sequential investment process that is characteristic of crowdfunding platforms, among other contexts. Investors wish to avoid the cost of information acquisition and thus prefer to rely on information acquired by previous investors. This may lead to a phenomenon similar to an information cascade. We characterize the optimal policy that balances between the incentive to acquire information and the optimal investment decision. The policy is based on time-varying transparency levels such that it may be worthwhile to conceal some information in some periods.
This paper was accepted by Joshua Gans, business strategy.
This paper incorporates costly information into a model of observational learning. Individuals would like to avoid the cost of buying information and free-ride on the public history. The paper characterizes when learning is nevertheless complete. Necessary and sufficient conditions for complete learning follow from an elementary principle: a player purchases information only if it can influence her action. With a “coarse” action space, learning is complete if and only if for every cost c>0, a positive measure of types can acquire, at cost less than c, an experiment that can overturn the public history. With a “rich” action space, learning is complete if and only if for every cost c>0, a positive measure of types can acquire any informative signal at cost weakly less than c. The results are applied to financial markets to evaluate when markets are informationally efficient despite information being costly.
Word of mouth (WOM) and observational learning (OL), as two types of social interaction, have been generally discussed by researchers. And both have a significant impact on consumer purchasing decisions. Online shopping has also begun in Indonesia, but consumers are not accustomed to shopping online often. During the 19th Pandemic many consumers did online shopping from various online websites. This study aims to examine the effect of Online Social Interaction: Observational Learning and Word of Mouth Against Consumer Online Purchase Decision during the pandemic covid 19. Researched as many as 250 respondents using online samples, using the convenience method. The results are expected to be able to test the proposed hypothesis about the interrelationships between variables. The results of testing using e views found that word of mouth has a significant effect on online purchase decisions, observational learning has a significant effect on online purchase decisions, there is an influence of word of mouth and observational learning on online purchase decisions simultaneously. The word of mouth is a factor that influences online purchase decisions with positive word of mouth, which will increase consumer online purchase decisions. Observational learning is a factor in making a purchase decision.
Using data on hospitals’ purchases across a large number of important product categories, we find that access to information on purchasing by peer hospitals leads to reductions in the prices hospitals negotiate for supplies. These effects are concentrated among hospitals previously paying relatively high prices for brands purchased in large volumes. Evidence from coronary stents suggests that transparency allows hospitals to resolve asymmetric information problems, but savings are limited in part by the stickiness of contracts in business-to-business settings. Savings are largest for physician preference items, where high-price, high-quantity hospital-brand combinations average 3.9% savings, versus 1.6% for commodities.
There is now a very large literature on dynamic learning models in marketing. Learning dynamics can be broadly defined as encompassing any process whereby the prior history of a consumer or market affects current utility evaluations (e.g., social learning, search, correlated learning, information spillover, etc.). In the present chapter, we focus on discussing this rapidly growing literature that deals with this broader view of learning dynamics.
I study stationary cut-off-strategy equilibria of a dynamic market model where buyers sample sellers sequentially from an unknown distribution. Buyers learn about the distribution from the sampled sellers and a ‘trade signal’. The trade signal reveals whether a randomly chosen seller traded yesterday. Observing a trade (as opposed to no trade) is good news about the distribution. Buyers who observe a trade use a higher cut-off than buyers who observe no trade, despite buyers’ learning from sampled sellers that puts a countervailing pressure on the cut-offs. The trade signal may reduce market efficiency, while an appropriate exogenous signal increases efficiency.
The increasing pervasiveness of social networks allows users to share purchase behaviors with their online friends. In the present study, we examine optimal pricing strategies of a monopolistic firm using an analytical model that accounts for behavioral observational learning in social networks. We show that a seller could potentially control the information available to future customers and induce behavioral observational learning by using an information-revealing pricing strategy. This result suggests that offering introductory discounts is not always an effective method to boost purchases in social networks. It could prevent the behavioral observational learning that would increase future customers’ willingness to pay. This article is protected by copyright. All rights reserved.
I study the problem of social learning in a model where agents move sequentially. Each agent receives a private signal about the underlying state of the world, observes the past actions in a neighborhood of individuals, and chooses her action attempting to match the true state. In contrast to the most existing literature that assumes an exogenous observation structure, observation in this paper is endogenous. More specifically, each agent must pay a cost to make any observation and can strategically choose the set of actions to observe. I show that when private beliefs are strong relative to cost, observation becomes fully informative if and only if the size of the observed actions extends to infinity. In addition, costly observation may lead to better learning than free observation, and the order of acquiring signal and observation significantly affects the learning pattern.
This article provides an empirical examination of observational learning. Using data from an online market for music, I find that observational learning benefits consumers, producers of high-quality music, and the online platform. I also study the role of pricing as a friction to the learning process by comparing outcomes under demand-based pricing to counterfactual pricing schemes. I find that employing a fixed price (the industry standard) can hamper learning by reducing the incentive to experiment, resulting in less consumer surplus, but more expected revenue for the platform.
We study a sequential social learning model where agents privately acquire information by costly search. Search costs of agents are private, and are independently and identically distributed. We show that asymptotic learning occurs if and only if search costs are not bounded away from zero. We explicitly characterize equilibria for the case of two actions, and show that the probability of late moving agents taking the suboptimal action vanishes at a linear rate. Social welfare converges to the social optimum as the discount rate converges to one if and only if search costs are not bounded away from zero.
In recent years, there has been stellar growth of location-based/enabled social networks in which people can "check in" to physical venues they are visiting and share with friends. In this paper, we hypothesize that the "check-ins" made by friends help users learn the potential payoff of visiting a venue. We argue that this learning-in-a-network process differs from the classic observational learning model in a subtle yet important way: Rather than from anonymous others, the agents learn from their network friends, about whose tastes in experience goods the agents are better informed. The empirical analyses are conducted on a unique data set in which we observe both the explicit interpersonal relationships and their ensuing check-ins. The key result is that the proportion of checked-in friends is not positively associated with the likelihood of a new visit, rejecting the prediction of the conventional observational learning model. Drawing on the literature in sociology and computer science, we show that weighting the friends' check-ins by a parsimonious proximity measure can yield a more intuitive result than the plain proportion does. Repeated check-ins by friends are found to have a pronounced effect. Our empirical result calls for the revisiting of observational learning in a social network setting. It also suggests that practitioners should incorporate network proximity when designing social recommendation products and conducting promotional campaigns in a social network.
The emergence of the Internet brought changes to traditional Word-of-Mouth Communication (WOM). Consumers are able to share and exchange opinions with anyone in the world, while WOM has customarily resulted from face-to-face relationships. Online portals allow individuals to purchase books, movie tickets, and music CDs as well as hospitality services. Those consumers made purchase decisions based on other customers' opinions on the Internet). Consumers feel safer subscribing to opinions stated by the masses because people try to imitate others' behaviors when there is not enough information. This phenomenon is called herding behavior. Many social media Web sites offering hotel rooms and travel amenities provide online platforms which consumers communicate with others who have information or previous experiences about the hotels. In this paper, the relationship between quantity of consumer reviews and online consumer purchasing intention in the hotel industry is tested.
Many markets have historically been dominated by a small number of best-selling products. The Pareto principle, also known as the 80/20 rule, describes this common pattern of sales concentration. However, information technology in general and Internet markets in particular have the potential to substantially increase the collective share of niche products, thereby creating a longer tail in the distribution of sales. This paper investigates the Internet's “long tail” phenomenon. By analyzing data collected from a multichannel retailer, it provides empirical evidence that the Internet channel exhibits a significantly less concentrated sales distribution when compared with traditional channels. Previous explanations for this result have focused on differences in product availability between channels. However, we demonstrate that the result survives even when the Internet and traditional channels share exactly the same product availability and prices. Instead, we find that consumers' usage of Internet search and discovery tools, such as recommendation engines, are associated with an increase the share of niche products. We conclude that the Internet's long tail is not solely due to the increase in product selection but may also partly reflect lower search costs on the Internet. If the relationships we uncover persist, the underlying trends in technology portend an ongoing shift in the distribution of product sales.
This paper was accepted by Ramayya Krishnan, information systems.
We study learning in a setting where agents receive independent noisy signals about the true value of a variable and then communicate in a network. They naïvely update beliefs by repeatedly taking weighted averages of neighbors' opinions. We show that all opinions in a large society converge to the truth if and only if the influence of the most influential agent vanishes as the society grows. We also identify obstructions to this, including prominent groups, and provide structural conditions on the network ensuring efficient learning. Whether agents converge to the truth is unrelated to how quickly consensus is approached. (JEL D83, D85, Z13)
We report results from a randomized natural field experiment conducted in a restaurant dining setting to distinguish the observational learning effect from the saliency effect. We find that, when customers are given ranking information of the five most popular dishes, the demand for those dishes increases by 13 to 20 percent. We do not find a significant saliency effect. We also find modest evidence that the observational learning effects are stronger among infrequent customers, and that dining satisfaction is increased when customers are presented with the information of the top five dishes, but not when presented with only names of some sample dishes. (JEL C93, D83)
This paper studies agents who consider the experiences of their neighbors in deciding which of two technologies to use. The authors analyze two learning environments, one in which the same technology is optimal for all players and another in which each technology is better for some of them. In both environments, players use exogenously specified rules of thumb that ignore historical data but may incorporate a tendency to use the more popular technology. In some cases, these naive rules can lead to fairly efficient decisions in the long run but adjustment can be slow when a superior technology is first introduced. Copyright 1993 by University of Chicago Press.
An informational cascade occurs when it is optimal for an individual, having observed the actions of those ahead of him, to follow the behavior of the preceding individual with regard to his own information. The authors argue that localized conformity of behavior and the fragility of mass behaviors can be explained by informational cascades. Copyright 1992 by University of Chicago Press.
We propose a boundedly rational model of opinion formation in which individuals are subject to persuasion bias; that is, they fail to account for possible repetition in the information they receive. We show that persuasion bias implies the phenomenon of social influence, whereby one's influence on group opinions depends not only on accuracy, but also on how well-connected one is in the social network that determines communication. Persuasion bias also implies the phenomenon of unidimensional opinions; that is, individuals' opinions over a multidimensional set of issues converge to a single "left-right" spectrum. We explore the implications of our model in several natural settings, including political science and marketing, and we obtain a number of novel empirical implications. © 2001 the President and Fellows of Harvard College and the Massachusetts Institute of Technology
This paper explores how Bayes-rational individuals learn sequentially from the discrete actions of others. Unlike earlier informational herding papers, we admit heterogeneous preferences. Not only may type-specific "herds" eventually arise, but a new robust possibility emerges: confounded learning. Beliefs may converge to a limit point where history offers no decisive lessons for anyone, and each type's actions forever nontrivially split between two actions. To verify that our identified limit outcomes do arise, we exploit the Markov-martingale character of beliefs. Learning dynamics are stochastically stable near a fixed point in many Bayesian learning models like this one.
The internet has made it easier for customers to find and buy a wide variety of products. This may lead to a "long tail" effect as more customers buy low-volume products. However, the internet has also made it easier for customers to find out which products are most popular. This could lead to a "steep tail" effect as customers flock towards the most popular products. Using data from a field experiment with a website that lists wedding service vendors, we find empirical evidence that a steep tail exists. The most popular vendors become more popular when customers can easily observe previous customers' click-through behavior. Then, we ask whether this steep tail effect "complements" the long tail, by attracting customers who would otherwise have chosen nothing, or "competes with" the long tail, by shifting customers from less popular vendors to popular ones. We find evidence of a complementary effect, where the steep tail indicates new interest in the most popular vendors from outside, with negligible cannibalization of interest for less popular vendors. The findings suggest that popularity information can serve as a powerful marketing tool that facilitates product category growth. They also explain the prevalence of firm practices to highlight bestsellers.
This paper reports an experimental test of how individuals learn from the behavior of others. By using techniques only available in the laboratory, we elicit subjects' beliefs. This allows us to distinguish informational cascades from herd behavior. By adding a setup with continuous signal and discrete action, we enrich the ball-andurn observational learning experiments paradigm of Lisa R. Anderson and Charles Holt (1997). We attempt to understand subjects' behavior by estimating a model that allows for the possibility of errors in earlier decisions.
When a series of individuals with private information announce public predictions initial conformity can create an "information cascade" in which later predictions match the early announcements. This paper reports an experiment in which private signals are draws from an unobserved urn. Subjects make predictions in sequence and are paid if they correctly guess which of two urns was used for the draws. If initial decisions coincide, then it is rational for subsequent decision makers to follow the established pattern, regardless of their private information. Rational cascades formed in most periods in which such an imbalance occurred.
This article examines how a decision maker who is only partially aware of his temptations learns about them over time. In facing temptations, individuals use their experience to forecast future self-control problems and choose the appropriate level of commitment. I demonstrate that rational learning can be perpetually partial and need not result in full sophistication. The main result of this article characterizes necessary and sufficient conditions for learning to converge to full sophistication. I apply this result to a consumption-savings environment in which a decision maker is tempted by present bias and establish a learning-theoretic justification for assuming sophistication in this setting.
“An individual who finds himself continuously repudiating his past plans may learn to distrust his future behavior, and may do something about it.” — Strotz (1955)
This paper explores rational social learning in which everyone only sees un- ordered samples from the action history. When the past is not "over-sampled" — not forever affected by any one individual — there is complete learning if private signal strengths are not bounded. When sampling is "recursive", the fractions taking each action converge if the recent past is also not over-sampled. This limiting fraction is one — a correct "purifying herd" — with unbounded beliefs and unit sample sizes. With bounded beliefs, incorrect purifying herds can occur with larger sample sizes. The extra noise in this urn model induces a path-dependence that twists standard insights — for instance, beliefs need never approach a cascade set in a purifying herd. The paper shows that a more informed sample need not beget a more in- formed observer, and as a result, welfare may fall in the sample size.
Observational learning is typically examined when agents have precise information about their position in the sequence of play. We present a model in which agents are uncertain about their positions. Agents are allowed to have arbitrary ex-ante beliefs about their positions: they may observe their position perfectly, imperfectly, or not at all. Agents sample the decisions of past individuals and receive a private signal about the state of the world. We show that social learning is robust to position uncertainty. Under any sampling rule satisfying a stationarity assumption, learning is complete if signal strength is unbounded. In cases with bounded signal strength, we show that agents achieve what we define as constrained efficient learning: individuals do at least as well as the most informed agent would do in isolation.
This paper examines how a decisionmaker who is only partially aware of his tempta-tions learns about them over time. In facing temptations over time, individuals use their experiences to forecast future self-control problems and choose the appropriate level of com-mitment. I demonstrate that rational learning can be perpetually partial and need not result in full sophistication. The main result of this paper characterizes necessary and sufficient conditions for learning to converge to full sophistication. This result has implications for commonly studied self-control environments such as savings and addiction.
Summary. When there are competing technologies or products with unknown payoffs an important question is which technology will prevail
and whether technologies with different payoffs can coexist in the long run. In this paper, we use a social learning model
with local interactions to study this question. We show that the adoption of technologies as well as the prospects of conformism/diversity
depend crucially on the nature of interaction between individuals and the heterogeneity of preferences in a society.
We consider a simple variant of the canonical model of social learning and show that in many situations it is optimal for an agent to abandon her own information and follow the minority rather than the majority. This possibility depends on two economically meaningful requirements: agents are differentially informed and observe only the number of agents having chosen each option, such as consumers observing only market shares. We show that minority wisdom arises when information is sufficiently heterogeneous and the well informed are not overly abundant, yet the conditions necessary are not overly restrictive. In fact, it is possible for the minority to be wise even when the minority consists of a lone dissenter and a majority of citizens are well informed.
Hit songs, books, and movies are many times more successful than average, suggesting that “the best” alternatives are qualitatively
different from “the rest”; yet experts routinely fail to predict which products will succeed. We investigated this paradox
experimentally, by creating an artificial “music market” in which 14,341 participants downloaded previously unknown songs
either with or without knowledge of previous participants' choices. Increasing the strength of social influence increased
both inequality and unpredictability of success. Success was also only partly determined by quality: The best songs rarely
did poorly, and the worst rarely did well, but any other result was possible.
This paper studies the role of product discovery in the demand for recorded music. We show that releasing a new album causes a substantial and permanent increase in sales of the artist's old albums-especially if the new release is a hit. Patterns in these "backward spillovers" suggest that they result from consumers discovering the artist upon hearing the new release. To explore the implications of consumers' incomplete information, we estimate a simple, learning-based model of market demand. Our results imply that the distribution of sales is substantially more skewed than it would be if consumers were more fully informed. (c) 2009 by The University of Chicago. All rights reserved..
When there are competing technologies or products with unknown payoffs which are adopted over time within a society, an important question is whether conformism or diversity will prevail. We use a learning model with local interactions to study this question. We show that the structure of information flows within a society helps to determine whether conformism or diversity obtains. We find that if information is public then society conforms to a single technology in the long run. On the other hand, if society consists of smaller groups of individuals and interaction within groups is more intense as compared to interaction across the groups, then two technologies can coexist and diversity obtains, in the long run. Our analysis involves a novel application of the Law of the Iterated Logarithm.
We study a market model in which competing firms use costly marketing devices to influence the set of alternatives that boundedly rational consumers perceive as relevant. Consumers in our model apply well-defined preferences to a consideration set, which is a subset of the feasible set and subject to manipulation by firms. We examine the implications of this behavioral model on otherwise competitive markets. In our model, the market equilibrium outcome is not competitive, yet firms earn competitive payoffs because the strategic use of costly marketing devices wears off the collusive impact of consumers' bounded rationality. Equilibrium behavior satisfies an effective marketing property: if a consumer considers a firm only because of a marketing device it employs, he ends up buying from that firm.
We extend the Bikhchandani, Hirshleifer and Welch (1992) informational cascade framework to allow for asymmetric signal accuracy. Simulations demonstrate that even small departures from symmetry may lead to non-monotonic effects of signal accuracy on the likelihood of an inefficient cascade.
The authors present a theory of rational behavior in which individuals maximize a set of stable preferences over goods with unknown addictive power. The theory is based on three fundamental postulates: consumption of the addictive good is not equally harmful to all, individuals possess subjective beliefs concerning this harm, and beliefs are optimally updated with information gained through consumption. Although individual actions are optimal and dynamically consistent, addicts regret their past consumption decisions and their initial assessment of the potential harm of the good. Addict-prone individuals who believe 'it could not happen to them' are most likely to be drawn into a harmful addiction. Copyright 1995 by University of Chicago Press.
This paper studies the way that word-of-mouth communication aggregates the information of individual agents. The authors find that the structure of the communication process determines whether all agents end up making identical choices, with less communication making this conformity more likely. Despite the players' naive decision rules and the stochastic decision environment, word-of-mouth communication may lead all players to adopt the action that is on average superior. These socially efficient outcomes tend to occur when each agent samples only a few others. Copyright 1995, the President and Fellows of Harvard College and the Massachusetts Institute of Technology.
The author analyzes a sequential decision model in which each decisionmaker looks at the decisions made by previous decisionmakers in taking her own decision. This is rational for her because these other decisionmakers may have some information that is important for her. The author then shows that the decision rules that are chosen by optimizing individuals will be characterized by herd behavior; i.e., people will be doing what others are doing rather than using their information. The author then shows that the resulting equilibrium is inefficient. Copyright 1992, the President and Fellows of Harvard College and the Massachusetts Institute of Technology.
We consider an environment where individuals sequentially choose among several actions. The payoff to an individual depends on her action choice, the state of the world, and an idiosyncratic, privately observed preference shock. Under weak conditions, as the number of individuals increases, the sequence of choices always reveals the state of the world. This contrasts with the familiar result for pure common-value environments where the state is never learned, resulting in herds or informational cascades. The medium run dynamics to convergence can be very complex and non-monotone: posterior beliefs may be concentrated on a wrong state for a long time, shifting suddenly to the correct state. Copyright Springer-Verlag Berlin/Heidelberg 2006
Short-lived agents want to predict a random variable and have to decide how much effort to devote to collect private information and consequently how much to rely on public information. The latter is just a noisy average of past predictions. It is shown that costly information acquisition prevents an unbounded accumulation of public information if (and only if) the marginal cost to acquire information is positive at zero . When public precision at period n, , tends to infinity with n but the rate of convergence of public information to is slowed down with respect to the exogenous information case. At the market outcome agents acquire too little private information. This happens either with respect to a (decentralized) first best benchmark or, for n large, with respect to a (decentralized) second best benchmark. For high discount factors the limit point of market public precision always falls short of the welfare benchmarks whenever . In the extreme, as the discount factor tends to one public precision tends to infinity in the welfare-optimal programs while it remains bounded at the market solution. Otherwise, if public precision accumulates in an unbounded way both at the first and second best solutions. More public information may hurt at either the market or second best solutions.
The creation of online consumer communities to provide product reviews and advice has been touted as an important, albeit somewhat expensive component of Internet retail strategies. In this paper, we characterize reviewer behavior at two popular Internet sites and examine the effect of consumer reviews on firms' sales. We use publicly available data from the two leading online booksellers, Amazon.com and BarnesandNoble.com, to construct measures of each firm's sales of individual books. We also gather extensive consumer review data at the two sites. First, we characterize the reviewer behavior on the two sites such as the distribution of the number of ratings and the valence and length of ratings, as well as ratings across different subject categories. Second, we measure the effect of individual reviews on the relative shares of books across the two sites. We argue that our methodology of comparing the sales and reviews of a given book across Internet retailers allows us to improve on the existing literature by better capturing a causal relationship between word of mouth (reviews) and sales since we are able to difference out factors that affect the sales and word of mouth of both retailers, such as the book's quality. We examine the incremental sales effects of having reviews for a particular book versus not having reviews and also the differential sales effects of positive and negative reviews. Our large database of books also allows us to control for other important confounding factors such as differences across the sites in prices and shipping times.
This paper describes a range of methods which have been proposed to study interactions in economic and social contexts. By interactions, we refer to interdependences between individual decisions which are not mediated by markets. These types of models have been employed to understand phenomena ranging from the effect of neighborhoods on the life prospects of children to the evolution of political party platforms. We provide a general choice-based framework for modeling such interactions which subsumes a number of specific models which have been studied. This framework illustrates the relationship between interactions-based models and models in statistical mechanics. Our analysis is then extended to the econometrics of these models, with an emphasis on the identification of group-level influences on individual behavior. Finally, we review some of the empirical work on interactions which has appeared in the social science literature.
We study learning and influence in a setting where agents communicate according to an arbitrary social network and naïvely update their beliefs by repeatedly taking weighted averages of their neighbors’ opinions. A focus is on conditions under which beliefs of all agents in large societies converge to the truth, despite their naïve updating. We show that this happens if and only if the influence of the most influential agent in the society is vanishing as the society grows. Using simple examples, we identify two main obstructions which can prevent this. By ruling out these obstructions, we provide general structural conditions on the social network that are sufficient for convergence to truth. In addition, we show how social influence changes when some agents redistribute their trust, and we provide a complete characterization of the social networks for which there is a convergence of beliefs. Finally, we survey some recent structural results on the speed of convergence and relate these to issues of segregation, polarization and propaganda.
This paper examines the reflection problem that arises when a researcher observing the distribution of behaviour in a population
tries to infer whether the average behaviour in some group influences the behaviour of the individuals that comprise the group.
It is found that inference is not possible unless the researcher has prior information specifying the compisition of reference
groups. If this information is available, the prospects for inference depend critically on the population relationship between
the variables defining reference groups and those directly affecting outcomes. Inference is difficult to implossible if these
variables are functionally dependent or are statistically independent. The prospects are better if the variables defining
reference groups and those directly affecting outcomes are moderately related in the population.
To test a model of rational addiction, the authors examine whether lower past and future prices for cigarettes raise current cigarette consumption. The empirical results tend to support the implication of addictive behavior that cross-price effects are negative and that long-run responses exceed short-run responses. Since the long-run price elasticity of demand is almost twice as large as the short-run price elasticity, the long-run increase in tax revenue from an increase in the federal excise tax on cigarettes is considerably smaller than the short-run increase Copyright 1994 by American Economic Association.
This paper considers a smooth and noisy version of the statistical prediction model studied in the herding/informational cascades literature and compares market and optimal learning. The latter is characterized by defining a decentralized welfare benchmark as the solution to an infinite horizonteamproblem. Market behavior involves herding, in the sense that agents put too little weight on their private information for any given precision of public information, and yields underinvestment in the production of public information. However, both market and optimal learning involveslowlearning. Examples of the model include learning by doing, reaching consensus, and consumer learning about quality.Journal of Economic LiteratureClassification Numbers: D82, D83.
This paper explores how Bayes-rational individuals learn sequentially from the discrete actions of others. Unlike earlier informational herding papers, we admit heterogeneous preferences. Not only may type-specific "herds" eventually arise, but a new robust possibility emerges: "confounded learning". Beliefs may converge to a limit point where history offers no decisive lessons for anyone, and each type's actions forever nontrivially split between two actions. To verify that our identified limit outcomes do arise, we exploit the Markov-martingale character of beliefs. Learning dynamics are stochastically stable near a fixed point in many Bayesian learning models like this one.
This paper uses detailed weekly data on sales of hardcover fiction books to evaluate the impact of the New York Times bestseller list on sales and product variety. In order to circumvent the obvious problem of simultaneity of sales and bestseller status, the analysis exploits time lags and accidental omissions in the construction of the list. The empirical results indicate that appearing on the list leads to a modest increase in sales for the average book, and that the effect is more dramatic for bestsellers by debut authors. The paper discusses how the additional concentration of demand on top-selling books could lead to a reduction in the privately optimal number of books to publish. However, the data suggest the opposite is true: the market expansion effect of bestseller lists appears to dominate any business stealing from non-bestselling titles.
This paper considers a problem of optimal learning by experimentation by a single decision maker. Most of the analysis is
concerned with the characterisation of limit beliefs and actions. We take a two-stage approach to this problem: first, understand
the case where the agent's payoff function is deterministic; then, address the additional issues arising when noise is present.
Our analysis indicates that local properties of the payoff function (such as smoothness) are crucial in determining whether
the agent eventually attains the true maximum payoff or not. The paper also makes a limited attempt at characterising optimal
experimentation strategies.
We show that far from capturing a formally new phenomenon, informational herding is really a special case of single-person experimentation -- and 'bad herds' the typical failure of complete learning. We then analyze the analogous team equilibrium, where individuals maximize the present discounted welfare of posterity. To do so, we generalize Gittins indices to our non-bandit learning problem, and thereby characterize when contrarian behaviour arises: (i) While herds are still constrained efficient, they arise for a strictly smaller belief set. (ii) A log-concave log-likelihood ratio density robustly ensures that individuals should lean more against their myopic preference for an action the more popular it becomes.
Players with observable, continuous payoff types choose actions sequentially. Players update beliefs about the payoff-relevant state of nature using private signals and information about previous signals in the history of types and actions. Heterogeneity in preferences ensures complete learning.
This paper uses detailed weekly data on sales of hardcover fiction books to evaluate the impact of the New York Times bestseller list on sales and product variety. In order to circumvent the obvious problem of simultaneity of sales and bestseller status, the analysis exploits time lags and accidental omissions in the construction of the list. The empirical results indicate that appearing on the list leads to a modest increase in sales for the average book, and that the effect is more dramatic for bestsellers by debut authors. The paper discusses how the additional concentration of demand on top-selling books could lead to a reduction in the privately optimal number of books to publish. However, the data suggest the opposite is true: the market expansion effect of bestseller lists appears to dominate any business stealing from non-bestselling titles. Copyright 2007 Blackwell Publishing Ltd..
Corruption in the public sector erodes tax compliance and leads to higher tax evasion. Moreover, corrupt public officials abuse their public power to extort bribes from the private agents. In both types of interaction with the public sector, the private agents are bound to face uncertainty with respect to their disposable incomes. To analyse effects of this uncertainty, a stochastic dynamic growth model with the public sector is examined. It is shown that deterministic excessive red tape and corruption deteriorate the growth potential through income redistribution and public sector inefficiencies. Most importantly, it is demonstrated that the increase in corruption via higher uncertainty exerts adverse effects on capital accumulation, thus leading to lower growth rates.
This paper analyzes a model of rational word-of-mouth learning, in which successive generations of agents make once-and-for-all choices between two alternatives. Before making a decision, each new agent samples N old ones and asks them which choice they used and how satisfied they were with it. If (a) the sampling rule is “unbiased” in the sense that the samples are representative of the overall population, (b) each player samples two or more others, and (c) there is any information at all in the payoff observations, then in the long run every agent will choose the same thing. If in addition the payoff observation is sufficiently informative, the long-run outcome is efficient. We also investigate a range of biased sampling rules, such as those that over-represent popular or successful choices, and determine which ones favor global convergence towards efficiency.
Short-lived agents want to predict a random variable θ and have to decide how much effort to devote to collect private information and consequently how much to rely on public information. The latter is just a noisy average of past predictions. It is shown that costly information acquisition prevents an unbounded accumulation of public information if (and only if) the marginal cost to acquire information is positive at zero (C′(0) > 0). When C′(0) - 0 public precision at period n, τn, tends to infinity with n but the rate of convergence of public information to θ is slowed down with respect to the exogenous information case. At the market outcome agents acquire too little private information. This happens either with respect to a (decentralized) first best benchmark or, for n large, with respect to a (decentralized) second best benchmark. For high discount factors the limit point of market public precision always falls short of the welfare benchmarks whenever C′(0) > 0. In the extreme, as the discount factor tends to one public precision tends to infinity in the welfare-optimal programs while it remains bounded at the market solution. Otherwise, if C′(0) = 0 public precision accumulates in an unbounded way both at the first and second best solutions. More public information may hurt at either the market or second best solutions.
Error Persistence, and Experimental versus Observational Learning
- L Smith
Smith, L. " Error Persistence, and Experimental versus Observational Learning, " 1991,
working paper.
Success and Failure in Cultural Markets
- M Salganik
Salganik, M., " Success and Failure in Cultural Markets, " Ph.D. thesis, 2007, Department
of Sociology, Columbia University, New York.
Biased Social Learning Cowles Foundation Discussion Paper 1738
- H Herrera
- J Horner
Herrera, H. and Horner, J. " Biased Social Learning, " Cowles Foundation Discussion Paper
1738, November, 2009.
- D Acemoglu
- M Dahleh
- I Lobel
- A Ozdaglar
Acemoglu, D., Dahleh, M., Lobel, I. and Ozdaglar, A. " Heterogeneity and Social Networks, " 2008, slides for MIT Theory talk.
- P M Demarzo
- D Vayanos
- J Zwiebel
Demarzo, P.M., Vayanos, D., and Zwiebel, J. " Persuasion Bias, Social Influence, and
Unidimensional Opinions, " Quarterly Journal of Economics, 118, 2003, pp. 909-968.