«Information, Privacy, and the Internet: An Economic Perspective Susan Athey Stanford Graduate School of Business1 Contents Introduction 5 1. ...»
I am currently working with Markus Mobius on follow-on research to understand the role of social media in determining the kinds of news people read. In preliminary findings, we show that people read different types of news through social media. For example, we find that within a user, articles read through social media tend to take a more caring tone, and to show more individual perspectives on the news. We also find that readers who read liberal political news tend to skew much more liberal in their reading from social media than in their other reading, and similarly conservative readers read more conservative articles in social media than their other reading. Overall, these preliminary findings support the premise that users are influenced by their context and the information that is presented to them, and also that we should expect to see greater impact of data and personalization on news consumption in the future, especially as social media becomes more important as a source of news.
4 The Impact of User Multi-Homing on Publishers We have presented evidence that internet search and news aggregators can greatly affect the way consumers view information on the internet. News aggregators lead to more diverse browsing patterns. When an audience views content from a variety of different websites, it becomes more difficult for advertisers to control how many times the show ads to the same user. In 2010, ComScore, a U.S. firm that tracks web browsing for a panel of users, provided me with a custom analysis of 30 very large, cross-media online ad campaigns. Most advertisers attempt to reach a given user between 3 and 7-8 times with the same advertisement.
Advertising is considered “wasted” if a user sees the ad too few or too many times.
The chart above illustrates the percentage of impressions shown to consumers where there was waste. Clearly, the technology used by the advertisers to attempt to manage wasted impressions did not fully accomplish the objective.
The inability of advertisers to avoid wasting ad impressions has a depressing effect on advertising prices, both because it affects efficiency and because it affects the way prices are determined.
Athey, Calvano, and Gans (2013) analyze a model with partially multi-homing consumers, showing in such a world, high-value advertisers buy ads on multiple outlets in order to achieve maximum “reach,” accepting that some impressions will be wasted.
Lower-value advertisers, whose value determine the market prices of ad space, tend to focus on one or a small number of outlets, as they find that once waste from duplicated impressions is accounted for, the price of ad space is greater than the benefit to them. In contrast, when they focus on a single outlet, there is no waste, so the impressions that they buy are worthwhile. The paper further shows that a decrease in consumer loyalty (an increase in switching) generally leads to lower advertising prices, and that outlets that have a broad “reach” can command higher advertising prices, creating a force in favor of consolidation.
Problems of fragmented viewing patterns in online news are exacerbated as consumers shift to doing more and more of their internet activity on mobile devices. Due to the increased use of apps to access the internet instead of web browsing, it becomes more difficult for advertisers to track users across multiple outlets and devices. Indeed, the ecosystem has been described as “fractured, complex, and hugely important” (Baye, 2013).
Further, “In mobile advertising, the rules of the road change with different combinations of device, wireless operator, and operating system. And there are few shared protocols or standards: Mobile lacks the technical consensus that enables ad targeting, delivery, and measurement to work fairly seamlessly across the desktop world.” Consistent with this, mobile so far attracts only 3% of the ad dollars, even though it is responsible for 12% of utilization time (Danova, 2014).
There are a number of possible concerns about online advertising on mobile devices, as well. Mobile devices are much more tightly controlled than, for example, personal computers running Microsoft Windows have been historically, or browsers in the PC environment. Applications must meet guidelines to be allowed into “app stores,” for example. It is conceivable that a mobile platform could, through policies or technology, eliminate the possibility for third party ad platforms to operate on a given mobile device platform. This could eliminate competition and allow the platform’s own advertising platform to charge high fees. Similarly, regulation that prevented all data sharing across apps could disadvantage small, independent app providers, since they would have no way to know enough about consumers to serve relevant ads, and thus no way to monetize their apps. As a result, niche apps might not be invented, or might not get as much funding and investment.
Let us now review a simple framework for evaluating the costs and benefits of privacy regulation. See Acquisiti (2010) for a more complete discussion of costs and benefits, as well as Goldfarb and Tucker (2011).
To start, consider the goals and potential benefits of privacy regulation. Any thorough discussion should begin by analyzing the different reasons we care about privacy. There are a few main categories, many of which have been introduced in this paper
Direct economic harm to a consumer from the use of the data, e.g. a user being denied insurance, employment, credit, etc. on the basis of the data, or experiencing reputational harm or legal difficulties.
Direct economic harm to a consumer from a security breach involving the data, e.g. a stolen identity, public embarrassment, harm to career or reputation Direct economic benefits from the use of data, e.g.
personalized services, more relevant ads Indirect economic benefits from the use of data, e.g. the existence of web sites, political news and investigative journalism, and ad-funded services Indirect economic benefits from the security of data against other nations and from surveillance to protect a nation, e.g.
national security Consumer intrinsic value (“feelings” or “utility”) about personal considerations such as being treated fairly, having their personal rights respected, not being tricked, and sharing in value creation from something they feel is theirs Consumer intrinsic value (“feelings” or “utility”) about the values incorporated in the economic and political systems in which they live There are a number of ways in which markets don’t function well when it comes to privacy. First, many consumers are not informed about privacy and do not understand the risks and benefits of alternative privacy policies. This is due to a combination of factors, but one appears to be the complexity of the policies and the lack of standardization in the industry. Another key factor is that it is objectively difficult for even experts to learn how governments and other entities make use of data, as well as how it might be used if a company suffered a security breach. Thus, the true implications of privacy policies are extremely difficult to understand. Indeed, many of the case studies where firms have been forced to change their privacy policies have come about when the news media covers new revelations from experts, insiders, advocacy groups, governments or competitors. As a recent example, Google publicly “shamed” Comcast, a major U.S. cable provider and internet service provider, into encrypting email exchanged Gmail, by publishing a list of which email providers use encryption (Constantin, 2014).
Without this public announcement and subsequent media coverage, it would have been difficult for typical users to ascertain that their provider was not following industry best practices.
Acquisti et al (2011) reinforces the finding that privacy preferences are context-dependent. Research by the World Economic Forum (2013) also argued that privacy preferences are highly contextspecific and vary across countries. One piece of context they identified as important is whether or not consumers perceive that they get something of value in exchange for their data, preferably something related to the use of the data. This finding has been echoed by other industry observers as well (Smith, 2013).
Although policy makers face a similar challenge, the incentives of policy makers differ, and they may be more likely to account for consumer welfare even when it is difficult to measure.
Fifth and related, consumers’ preferences are very difficult to measure even when they relate to products and policies in their current environment, because the way that questions are framed and the supplementary information that is provided around choices can make a big difference. Acquisti (2013) establishes using a field experiment that users respond differently when they have privacy and are offered money to give it up, than when they don’t have privacy but are asked to pay to gain it. In this environment, it is difficult to believe that firms will be able to discern “true” preferences for privacy and respond to them; this favors a more direct approach to regulation that either makes policy alternatives simpler and clearer to consumers, or that directly regulates certain aspects of privacy.
Sixth, consumers may also change their feelings about the risks of a large firm retaining their data after news about government subpoenas or U.S. National Security Administration surveillance.
Again, the signals firms would get from consumers about their preferences would be misleading in this case, since the revelation occurs after decisions have been made. Goldfarb and Tucker (2012) and Marthews and Tucker (2012) provide some evidence that views do indeed evolve over time in response “shocks” to information. Indeed, Marthews and Tucker (2012) show that users change their search behavior, reducing their queries on politically sensitive terms, after media reports about government surveillance. In such an environment, it is difficult to know how to put a dollar value on benefits to privacy protection to trade off against harm to long-term welfare, innovation, and so on.
Government policy may struggle with the same problems faced by the private market. Still, there may be a role for regulation in an environment where the market gives almost no weight to a potentially relevant concern (like the effects of government surveillance on the costs and benefits of long-term data retention about individual users, for example).
Seventh, regulation may correct for the free rider problem faced by consumers—no individual has the incentive (nor the expertise) to audit major technology firms with which they interact. It may also provide expert opinion about what is important.
Now consider some potential harms from privacy regulation. First and foremost, as discussed above, privacy regulation that interferes with the effectiveness of online advertising makes it hard for new ventures to attract an initial user base, as well as hard for new ventures to monetize their content by showing ads and sharing the revenue from them. In short, decreases in efficiency in online advertising lead to decreases in innovation and the creation of content. For example, Miller and Tucker (2014) show that state level protections for genetic testing have a mixed impact. Providing users with informed consent leads to the users not purchasing the product.