05 Nov 2023

privacy economics sources

In one generation, we have gone from buying Computer Shopper for the ads to a whole ad blocking software development scene. What happened? The advertising that people are blocking now is different from what advertising mostly used to be, and most of the ways it’s different get lumped together under privacy.

Alessandro Acquisti writes, in The Economics of Privacy at a Crossroads, I argue that as economists we have, by and large, adopted a reductionist view of privacy that overlooks the richness and nuance of the contemporary debate around privacy.

Part of the problem, it seems to me, is the idea that people are balancing privacy harms on one side against supposed benefits of personalized advertising on the other side. But the harms are documented, while the benefits are…just kind of assumed. (usually) Meanwhile, a substantial body of literature supports the argument that personalization has built-in costs of its own. So privacy regulations and technology that limit personalization could be failing to get credit for some important economic wins. If you look at the math, a personalized message carries less information than a message sent without knowledge of the recipient.

Gardete and Bart “We find that when the sender’s motives are transparent to the receiver, communication can only be influential if the sender is not well informed about the receiver’s preferences. The sender prefers an interior level of information quality, while the receiver prefers complete privacy unless disclosure is necessary to induce communication.” Tailored Cheap Talk

This makes sense to me. If you hear an entrepreneur make two claims about their new product, one in a meeting with you alone, and one as an answer to a question on a trade show panel, you’re more likely to believe the second one. In that case, the entrepreneur didn’t know who was in the audience: existing customers, experts in the field, competitors, maybe even the government.

So it seems like moving the same ad from Computer Shopper style, seen by a whole community of practice, to cold call style, seen by specific groups only, would tend to reduce its credibility. But why does a widely shown ad carry information in the first place?

Kihlstrom and Riordan A great deal of advertising appears to convey no direct credible information about product qualities. Nevertheless such advertising may indirectly signal quality if there exist market mechanisms that produce a positive relationship between product quality and advertising expenditures. Advertising as a Signal

Ambler and Hollier High perceived advertising expense enhances an advertisement’s persuasiveness significantly, but largely indirectly, by strengthening perceptions of brand quality. The Waste in Advertising Is the Part That Works

Davis, Kay, and Star It is not so much the claims made by advertisers that are helpful but the fact that they are willing to spend extravagant amounts of money. Is advertising rational?

Now, here is where the value destruction from personalization comes in. What if an ad medium supports enough personalization that it becomes too hard for consumers to detect any economic signal from a legit ad that appears in the medium?

Hertzendorf This result, however, is sensitive to the assumption that consumers can perfectly observe the firm’s advertising expenditure. This assumption is somewhat unreasonable in light of the fact that much advertising takes place over various electronic media to which not everyone is ‘tuned in.’…Furthermore, the noise complicates the process of customer inference. This enables a low-quality firm to take advantage of consumer ignorance by partially mimicking the strategy of the high-quality firm. I’m Not a High-Quality Firm-But I Play One on TV

That paper covers the situation where viewers have a multitude of TV channels to choose from, but the signal loss effect is at its worst in highly personalizable online ad media. In a healthy ad medium, the value of ad-supported content plus the signaling value of the advertising itself is greater than the annoyance of advertising plus the risks of privacy threats. As signal goes to zero, the ad has a harder time paying its way, even if annoyances and risks are somehow reduced.

What’s even worse for users is that personalization provides extra, bonus value to deceptive advertisers. For a legit advertiser, the value of showing an ad to the wrong person is zero. You wasted a fraction of a cent. Deceptive advertisers, though, obtain value from avoiding showing their ads to people who would try to hold them accountable in some way. For example, a seller of designer knock-offs can avoid employees of the legit manufacturer, their ad agency, and their country’s law enforcement agencies. Sellers of quack medicine and investment scams can avoid regulators and journalists. The same personalization capacity can be worth more to deceptive sellers than to honest ones. (More examples of how online ad media facilitate deception, from the FBI, FTC, and NBER.)

Lin et al. We find that for users currently using an ad-blocker, uninstalling them leads to a deterioration in their online experiences and lower satisfaction with recent purchases. For users that were not using Ad-Blockers, installing one led to fewer reported regrets with purchases, an improvement in subjective well-being, and a less positive view of online advertising. The Welfare Effects of Ad Blocking

The signaling problem and the deceptive sellers problem might help explain why privacy tool developers are choosing to block Google’s in-browser ad personalization. So why are so many intermediary companies so fired up about tracking people when it reduces the value to the buyer and seller ends? Maybe because overgrown open-source companies dominate advertising now? More: the Linux device driver hacker’s guide to giant Internet monopoly dominance Whether based on actual user research, user requests, or intuition, privacy developers are anticpating that users are just as likely to want to avoid personalization when it’s free of individualized tracking as they are to avoid personalization based on an individual identifier. Yes, there are group discrimination risks that are still an unsolved problem, but right now it looks like privacy developers are betting that users will want to avoid not just individual tracking, but all personalization. The signaling literature might provide some hints why.

#