Resumen de Essays on information acquisition
It is well-known that information plays an important role in an economy. In industrial economics, different information structures result in different conducts and performances in a given market. More often than not, the literature assumes that a market’s informational component is fixed and is taken as part of the exogenous environment.
However, the information structure could also be understood as an equilibrium composed of agents’ endogenous choices. A rational agent could treat information as a product, or rather, an input in making an economic decision. If we understand it in this way, then some basic economic questions immediately apply: is it worth paying for information? what is the quality of the information that we can get? As understanding agents’ choices in information acquisition is crucial for developing marketing strategies and for designing mechanisms to maximize social welfare, a central theme of all my chapters is to understand the incentive of information acquisition, and analyze the information transmission mechanisms in an economy.
Chapter 1 discusses a common economic decision in everyday life: should I buy this product? or, which product should I buy? It studies consumers’ social learning and decision making process from both theoretical and empirical angles. The learning process is as follows: when a product’s quality is unknown to consumers, consumers can first learn about the product through its salient and easy-to-digest characteristics, like popularity information, or how many times this product has been sold in the past; besides this, consumers can also continue with costly information acquisition, for example, they can spend time to read customer reviews, which are detailed and more informative.
This chapter addresses a key issue, that is, understanding the determinants of a consumer’s incentive to acquire extra information besides observing others’ choices.
The incentive can be affected by some exogenous factors, like the product’s value and price, and how urgent the need is. The incentive also hinges on a consumer’s perception of others’ information acquisition efforts. As consumers can first observe others’ choices, if they think those choices are made carefully with deep knowledge, they lose incentives to exert efforts to acquire information by themselves. In contrast, if they think their predecessors are myopic trend-followers, then they would spend more time to do their own research.
In this chapter, I first provide a tractable model that characterizes this kind of social learning process and highlights the endogenous information acquisition incentive. It combines two strands of literature: observational learning (Banerjee, 1992; Bikhchandani, Hirshleifer and Welch, 1992; Smith and Sørensen, 2000) and search (Wolinsky, 1986; Anderson and Renault, 1999) , but departs from their standard settings and is parsimonious to study their interactions.
To test my model, I use data from a Chinese online air purifier market. This market provides an ideal environment to study consumers’ social learning. First, the air pollution in China started to draw public attention since only a few years ago. It is a new market without much past consumption experience, and thus consumers rely mainly on social learning instead of own experience. Secondly, a mainstream air purifier costs around 500 dollars, which is about half of the average monthly wage in Beijing. Therefore, consumers take the decision seriously. A counter example could be an anti-pollution mask. Thirdly, air purifiers are differentiated mainly in the vertical dimension, unlike movie/book/music industry where consumers are heterogeneous in taste (horizontal dimension), this market provides a simple Bayesian learning environment.
Moreover, I exploit an unexpected pollution crisis in 2014 as a natural experiment that changed some factors affecting consumers’ incentives of learning.
On the first day of this pollution crisis, the Chinese government issued an alarm to the public which awakened public attention. I interpret this shock as increasing the urgencies of the need of an air purifier, and/or increasing the air purifier’s economic value. I find that, during this shock, consumers are more likely to buy the products that have high sales. This is because consumers in the crisis are more likely to make a quick purchase decision without searching for more detailed information. I also find that after the crisis, consumers are less likely to buy the products that have high sales.
This is because purchase decisions made during the shock are less reliable than before, hence, consumers after the shock have higher search incentives instead of relying on the sales record.
In Chapter 1, I only focus on information acquisition and transmission in the demand side. Though it is desirable to consider the supply side’s dynamic learning and pricing strategies for obtaining a general equilibrium, tractability is difficult to achieve because of interactions between the two sides. In Chapter 2, which is a joint work with Julio Crego, we consider information acquisition in the stock market. While learning has attracted the attention of many economists (Frick and Ishii, 2015; Young, 2009; Ellison and Fudenberg, 1993; Jensen, 1982), however, it has not been much explored in the stock market. This market provides an ideal setting for a joint analysis of demand and supply forces because of a special feature of it, that is, the price setter is a market maker who makes zero profit. This feature greatly reduces the complexity of pricing strategies and makes learning tractable even with multiple agents.
As anticipated, Chapter 2 works on the economic decision of buying or selling a stock. In the market microstructure literature, starting from Kyle (1985) and Glosten and Milgrom (1985), it is common to model three types of agent in the stock market: informed traders who are essentially insiders; noise traders who trade because of liquidity needs; and the market maker, who sets price quotes, behaves competitively and makes zero profits. We modify the classic setting of Easley and O’Hara (1992) by introducing another type of agent, less-informed traders, who are partially informed about the state of the asset but not the exact value. The learning incentive that we aim at characterizing would come from this type of agent because they could delay a trading opportunity for the purpose of observing others’ actions and obtaining complementary private information from insiders.
We find that the equilibrium exists where an insider will act at the earliest possible time to profit like an early bird catching the worms, while under some conditions a less-informed trader prefers to wait until there are previous orders for them to learn from. The less-informed trader is a second mouse getting the cheese thanks to learning from predecessors. Moreover, we show that the greater the information asymmetry between the less-informed trader and the market maker, the more cheese the former can get.
Chapter 3 studies the news industry where a relevant economic decision for the demand side is whether or not to pay a subscription fee to read news. Different from the other two chapters, this chapter introduces a third party, a news aggregator such as Google News, who offers the platform for readers to easily acquire useful information on the quality of news that they hesitate to pay to read. With the presence of this third party in market, an important question arises: "what is the effect that a news aggregator has on newspapers? ".
Since 2007, when a Belgian court ruled that Google did not have the right to display the lead paragraph from French-language Belgian news sources, there have been successive disputes against Google. On January 31st 2013, Google agreed to create a 60me digital publishing innovation fund to help support the French press. In 2014, the Spanish Cabinet approved a draft reform of the Intellectual Property Law, which includes a so-called “Google tax” on search engines for showing snippets. As a result, Google shut down its Google News service in Spain on 16 December, 2014 before the law came into effect.
Those disputes arise from the fact that a news aggregator is like a double-edged sword. It has two effects on newspapers: “business-stealing effect”and “marketexpansion effect”. For a better understanding, here I quote two statements to illustrate.
For the “business-stealing effect”, Carlos Fernando Lindenberg Neto, the president of Brazil’s National Association of Newspapers, says: “Staying with Google News was not helping us grow our digital audiences; on the contrary, by providing the first few lines of our stories to Internet users, the service reduces the chances that they will look at the entire story in our websites." However Google’s Public Policy Director, Marcel Leonardi claims that Google News expands the market and channels a billion clicks to news sites around the world, and compares requiring Google to pay is like: “Taxing a taxi driver for taking tourists to eat at a particular restaurant." Some empirical studies have documented how the presence of a news aggregator affects news consumption through newspapers. Athey and Mobius (2012) provide a case analysis of France Google News’ introduction of “localization”in 2009. They find that users who adopted the localization feature increased subsequently their usage of Google News and also consumed more local news from local outlets in the short run. In the long run, the reliance of users on Google News homepage increased.
Consistent with this empirical finding, the theoretical model of Jeon and Nasr (2016) explains that news aggregators’ market-expansion effect is due to the fact that news aggregators make it easier for consumers to access various high-quality contents; and the business-stealing effect comes from the shift of reader’s anchor sites from news sites to news aggregators. They find that the presence of aggregators may lead to newspapers’ specialization on news coverage, an increase in the quality of newspapers, and an increase in consumer surplus.
Complementary to Jeon and Nasr (2016), Chapter 3 illustrates these two effects from two novel perspectives. In terms of the market-expansion effect, this model shows that the presence of news aggregators can eliminate the information asymmetry between news readers and newspapers through ranking and snippets, which leads to more readers willing to pay for news. In terms of the business-stealing effect, this model emphasizes that the informative snippet is a substitution of the original content, hence, some light readers would end up reading snippets on news aggregators instead of clicking on the links to read original news on the newspaper websites.
Two empirical papers, Chiou and Tucker (2017) and Dellarocas et al. (2015), have also discussed the substitution effect of snippets. Chiou and Tucker (2017) find that extracts of content do not serve as a complete substitute for the full content. Instead, news aggregators induce users to navigate further by featuring multiple sources.
While by conducting a field experiment on a Swiss mobile news aggregator, Dellarocas et al. (2015) find a substitution relationship between the length of the text snippet and the probability that users visit the news site. Given that recent oppositions to news aggregators often come from the controversy over displaying snippets, it is of current interest to model how the design of snippets affect newspapers and consumers.
The model of Chapter 3 features a news aggregator’ decision on the snippet length, and could be used to evaluate different tax policies on news aggregators.
The plan of the rest of this thesis is as follows: In Chapter 1 I introduce the paper, To Glance or to Peruse: Observational and Active Learning from Pesr Consumers, which focuses on consumer social learning. Chapter 2 includes the paper entitled Early Birds and Second Mice in the Stock Market. In Chapter 3, I present the paper on news aggregators with the title Should Google Profit like a Taxi Driver?. Some concluding comments are made at the end of this thesis.
