The Argument over Fair Prices and Privacy

Suppose we could establish every technical fact about a pricing system. We knew its inputs, decisions and effects. We would still have to decide which uses of advantage are acceptable.

That is a moral and political question informed by evidence. It cannot be delegated entirely to a profit model. Nor should moral certainty excuse poor evidence about what a business has actually done.

My position is that useful personalization should make exchange easier without leaving people unable to understand or challenge the terms. More computing power gives us more ways to organize a market. It does not relieve us of choosing what kind of market we want.

Fairness has more than one object

One idea of fairness is equal prices for equivalent offers. Another is affordable access for people who would otherwise be excluded. A third is a fair procedure: people understand the criteria, can correct mistaken information and have meaningful alternatives. A fourth concerns privacy itself, including uses of information people should not have to accept to obtain an ordinary service.

These goals can conflict. A transparent concession can produce unequal prices while expanding access. A universal price can treat buyers alike while excluding people who cannot pay it. A personalized discount can help a buyer financially while relying on information they would prefer not to share.

We do not resolve those conflicts by calling every difference discrimination or every discount inclusion. We need the criterion, the relevant comparison and the distribution of consequences. We also need an account of who decided the rule.

The economic review by Alessandro Acquisti, Curtis Taylor and Liad Wagman emphasizes that both sharing and protecting information can have benefits and costs. It also distinguishes the value people place on privacy itself from the consequences of revealing data. That distinction helps explain why a lower price cannot, on its own, demonstrate that a person received a better bargain in every respect. The Economics of Privacy

What the small model leaves open

In our synthetic market, information can support a range of divisions between seller profit and consumer surplus. The diagram below illustrates the established theoretical bounds for that particular monopoly setting. It is not a map of actual retailers.

The picture cannot tell us which point a market will choose. The choice depends on who controls information, what sellers can do with it and which rules they face. A consumer-friendly point being possible is not enough. Somebody has to make it feasible to reach and worthwhile to maintain.

There is also no competitor in the example. That omission is deliberate: it lets us inspect the arithmetic. It prevents us from using the result as a conclusion about competitive markets.

Andrew Rhodes and Jidong Zhou’s 2024 theoretical paper studies several competing firms. In their model, the effect of personalization depends on market coverage, meaning how widely the market serves potential buyers. Their analysis can yield consumer benefits or harms under different conditions; asymmetric access to data can also matter. This is a reason to examine competition and data distribution together, rather than announce that more rivals or more information always produces one result. The study is theory, not an estimate of the average household bill. Published paper

My practical inference is limited but important: an evaluation of pricing rules should describe the alternatives buyers can actually use. Counting businesses is a start. Understanding whether a person can switch at reasonable cost is closer to the question we care about.

A public bargain worth testing

I would judge proposals against four requirements.

First, buyers should be able to identify and compare the complete offer. That means the relevant terms and unavoidable costs should be available at the point of decision. A claimed saving should name its comparison. A personal promotion should not require the buyer to guess what they would otherwise have paid.

Second, organizations should answer for material uses of personal information. The degree of detail can be proportionate, but the explanation should identify what categories of information can change the offer and how a buyer can challenge an error. Records should support confidential independent investigation where public disclosure would expose another person’s information or genuine trade secrets.

Third, benefits should be assessed across people, not only across transactions. A programme reporting many discounted purchases should also examine who was excluded, who paid more and whether important losses were concentrated. The appropriate measures will differ by market. The requirement is to make the choice of measures visible and defend it.

Fourth, policies should preserve meaningful ways to refuse. An opt-out that makes an essential service practically inaccessible is a different arrangement from a choice among close substitutes. That is why a privacy policy and a competition policy cannot always be treated as separate conversations.

These are proposals, not a list of existing universal legal duties. They also need testing. A disclosure may be ignored, misunderstood or costly to provide. A price rule may reduce one harmful practice while encouraging a different one. A prohibition may preserve privacy while removing a discount that some households depended on. We should investigate those effects and improve the design rather than assume a good intention settles the result.

The burden of explanation should increase with the power to make consequential decisions about others. I think that is a defensible principle for a digital economy. It places responsibility with the organizations able to change the system, while preserving room for genuinely useful services.

What would change my judgment?

Evidence that a clearly described practice expands access, improves outcomes and respects informed refusal would strengthen the case for permitting it. Evidence that a rule predictably targets people with poor alternatives, conceals material conditions or concentrates losses would strengthen the case for restricting it. The relevant test is the practice and its consequences, not whether its maker uses the word innovation.

The evidence reviewed here does not support a national estimate of the cost of personalized pricing to an average household. It does support a serious research agenda, a distinction between mechanisms and an insistence that claims of consumer benefit have an adequate comparison.

The completed analysis in this brief is narrower: an independently implemented synthetic model makes several possible outcomes and fragile assumptions visible. It does not measure a retailer, estimate a demand curve from customers or demonstrate a novel theorem. That boundary is part of the value of the work. Readers can inspect exactly where the numbers came from.

At the counter

Put the terms where I can read them.
Leave a little time to choose.
If the offer knows my habits,
let me know what habits do.

I am here to buy a useful thing.
You are here to earn your pay.
We can make that bargain plainly.
We can build the rules that way.

There is room for ambition here. We can build tools that help people find suitable goods, reduce waste and obtain services they value. We can also require those tools to be accountable. A future in which prices use more information should not be a future in which buyers understand less.

That is the direction I would choose: better evidence, clearer offers and a real share of the gains for the people whose lives make the data.