Hide and Seek: Costly Consumer Privacy in a Market with Repeat Purchases
The authors, Conitzer, Taylor and Wagman model the possible relationship between firm profit and consumer benefit when consumers choose to be anonymous. The authors consider a B2C scenario, where all consumers are anonymous when the first price is offered. Consumers can then decide to maintain their privacy or be identified by the firm. Firms then have to make decisions about which price to offer to which consumers, private or identifiable, on the second round.
The authors enumerate a set of assumptions, build a model, study that model, and then make a number of assertions about the relationship between consumer choice and firm profits.
They assume that firms use information about past purchases to price discriminate. Their model shows that firms can drive higher profits even when consumers choose to maintain their anonymity. Conversely, consumers can be better off when anonymity is costly. This is the game of hide of seek between consumers and firms.
Consumers can chose a few ways to hide from a firm. In one sense, a repeat customer may simply refuse to buy again, in anticipation of an introductory offer for new customers. Indeed, promises of free cash, better rates, and goodies like free ipods or tablets for new or switching customers are fairly common. Hiding may also include an explicit opt-out such that a firm would not be able to identify whether or not a customer was a repeat customer or not.
Consumers may choose to hide from a firm, anticipating a predatory price upon their second purchase. In response, the firm may offer excellent introductory prices, which benefits consumers overall.
As the cost of anonymity rises, fewer customers will opt for it. As a result, the firm can ramp up its price discrimination, maximizing its profits as a result.
The authors conclude that the economic considerations of privacy may not be as obvious as they appear.
The paper is appropriately titled "Hide and Seek”. Consumers adjust their behavior according to how a firm adjusts its price strategy. Firms seek to use data to gain asymmetrical advantage, and, some consumers seek to use the company’s pricing behavior to maximize their own advantage.
While the authors did not present data to either validate their model or reject it, there’s an important intersection of two themes that’s worth pointing out.
We’re in an area of big consolidation. Fewer firms compete in the traditional Big Data B2C industries: banks, airlines, telecom, and search. These industries have consolidated markedly in the past decade through mergers, failures, acquisitions, and, in rare cases, offering a superior product.
The authors demonstrate what can happen in a monopoly. However, many duopolies or oligopolies have been known to behave, collectively, in a monopolistic manner. The monopoly assumption, while expedient for the sake of simplicity and to make the authors’ math workable, isn’t off base.
The second trend is that of the rise of privacy concerns and the re-emergence of consumer protection as a policy set. Most of the discussion, from 2010 to 2012, has been centered on ad targeting and cookies. Some firms have become very skilled at augmenting their targeting with features about people based on where they are currently and where they have been previously. Another class of firm unmasks consumers further by inferring age, gender, income and interests. All of these variables are used to predict whether or not you will click on one of the 20,000 display ads you see each year (and which one).
Yet, ad targeting didn’t cause the original concern about HTTP cookies and tracking generally.
It was price.
In 2000, Amazon.com did a simple A/B test on a DVD. A buyer visited a product page and the price displayed was $26.24. The buyer deleted his cookies and the price fell to $22.74. That’s a difference of $3.50.
This incident caused the original outrage. It’s what triggered the original, mass, clear-your-cookies habit. If you use web analytics, you continue to be affected by it to this day.
Amazon used the least discriminating variable possible: randomness. Everybody had an equal opportunity to be discriminated against in terms of price. It wasn’t based on previous buying behavior. It wasn’t based on age, sex, location or any of the advanced attributes that are known about people. It wasn’t based on browser, operating system, or domain. It was random price discrimination for the sake of identifying a demand curve.
It’s an accepted economic law that if a firm can charge different customers different prices based on where they lay along that demand curve, a firm increases its profit. You can see this law in action with airlines (class, loyalty points, time, competition, loads), telecom (plans), banks (credit worthiness) and especially insurance (risk). When firms charge different prices to different people, firms win. This is the very root of price optimization.
That doesn’t mean that all consumers are worse off. Some are better off. Some are worse off. The value that a consumer places on privacy may have a fairly large effect on the price offered.
What happens when CRM systems are linked to ad targeting systems? What happens to the demand curve then?
Most interestingly, the authors introduce a cost function to anonymity. Might there be a future where firms seek to maximize their profit by selectively offering anonymity at a cost? Would consumers pay for previously free utilities if it meant anonymity?
The authors model suggests that it’s far from clear who might be better off.
I recommend that interested members of the DAA read the introduction and conclusion of this paper.