The OpportunityCard, ten years on

The proposal

In a 2015 Conversation article, I proposed an "OpportunityCard." A government or other trusted verifier would confirm that a customer was below an income threshold. The seller would receive a yes-or-no eligibility signal and could offer a posted discount.

This is a form of third-degree price discrimination (3DPD): a seller offers different prices to observable groups. Firms already use cruder groups such as age, student status, location, and purchase history. They also use second-degree price discrimination (2DPD): menus of quantities or qualities designed to make customers sort themselves.

The interaction matters. A seller that cannot observe willingness to pay may deliberately make the cheaper version worse, or attach time and hassle costs to a discount, so that high-value buyers do not take it. A verified income group may let the seller reduce some of that distortion. But it can also let the seller extract more surplus, exclude some customers, or alter the standard price. There is no general welfare result.

The original paper and model are in an Essex Economics discussion paper and the project archive. The old model's main mechanism remains relevant, but the claim to theoretical novelty does not: recent work now studies 2DPD and 3DPD together much more generally.

What the evidence currently says

Eligibility can be checked without sharing an income file

The UK's Department for Work and Pensions lets participating broadband providers verify benefit eligibility, with the customer's consent. The provider receives a limited confirmation rather than the claimant's benefit record. The EU Digital Identity Wallet framework is also designed around verified attributes and selective disclosure.

This addresses the paperwork problem. It does not settle governance, appeals, data retention, take-up, or stigma.

Related schemes show delivery, not a general welfare gain

US EBT cardholders can receive discounted commercial memberships and reduced museum admission. UK telecom providers offer social tariffs. Brazil automatically matches eligible households to a regulated electricity tariff. These programs differ in who pays, whether participation is voluntary for firms, and whether the product is regulated.

They show that eligibility-based prices can operate at scale. They do not tell us what would happen across ordinary retail markets. UK take-up is also low: Ofcom reports 532,000 social-tariff customers in June 2025, or 8.6% of a conservative proxy for eligible households.

The empirical and theoretical results are mixed

DellaVigna and Gentzkow find that major US retailers price nearly uniformly across areas and estimate that this can raise prices paid by poorer households relative to flexible local pricing. Dubé and Misra find that personalized pricing increased profit and lowered prices for more than 60% of customers in one field setting, while reducing aggregate consumer surplus. Fillmore's study of FAFSA-based college pricing finds redistribution toward low-income students on average, but harm to roughly one-third of them.

The theory is similarly conditional. Market segmentation can produce many different divisions of surplus. Voluntary disclosure can help consumers under some disclosure technologies and market structures. Once firms can vary quality as well as price, the relevant demand and cost elasticities become central.

A narrower research agenda

I would separate four questions that the earlier project sometimes ran together.

Rules I would impose

1. Discounts from a public standard price. The credential cannot be used to raise a person's price.
2. Eligibility, not income. The seller receives a time-limited yes-or-no token for a stated discount category.
3. No retention or secondary use. The signal cannot feed advertising, credit, product steering, or later pricing.
4. Easy receipt, optional disclosure. Eligibility should not require a fresh application at every firm, but presenting the credential remains the customer's choice.
5. Measure firm responses. A pilot must track the standard price, quality, availability, and effects on customers who do not qualify.
6. Pre-set stopping rules. End the trial if eligible customers are worse off, standard prices rise materially, privacy protections fail, or administration is too costly.

How I would proceed

1. Write a short technical note. State the model above, map it to the 2022–26 literature, and ask a few relevant theorists whether the proposed extensions are actually new. Do this before presenting the old model as a contribution.
2. Audit the UK and Brazil data. Produce a brief feasibility memo covering units, dates, denominators, provider or distributor identifiers, outcomes, and access restrictions. The result may be that neither setting identifies the effect cleanly.
3. Prepare a two-page policy note. Focus only on the distinction between surveillance-based markups and consented eligibility discounts. Include draft safe-harbor language and the reasons for a discount-only rule.
4. Seek one empirical partner. A regulator, provider, or merchant with transaction data is more valuable than a large public launch. Pre-register the analysis before any pilot.

Promotion should follow that work. I would first circulate the technical note privately to researchers in information design, industrial organization, public economics, and the closest empirical papers. I would send the policy note separately to consumer-protection and digital-identity groups. A public article or forum post would make more sense after those checks, with the page serving as the stable reference rather than the argument itself.