Summary

  • The AFL-CIO Tech Institute’s September 9, 2026 report bundles three distinguishable pricing mechanisms — personalized pricing based on identified customer data, contextual dynamic pricing based on environmental signals, and instantaneous price-change infrastructure — under the single category “surveillance pricing,” according to the report’s authors.
  • Maryland, Connecticut, and New Jersey enacted surveillance-pricing bans in April, June, and July 2026, establishing the state-level precedent the AFL-CIO now seeks to extend into a federal ban.
  • The report projects 44,223 to 191,633 grocery job losses and $1.6 billion to $6.9 billion in annual lost wages without disclosing the methodology behind either estimate.
  • An Instacart pilot ended in December after an investigation by Consumer Reports and Groundwork Collaborative reportedly found grocery prices differing by as much as 23 percent between customers.
  • A May United Food and Commercial Workers poll reports 67 percent support for banning digital price tags and surveillance pricing, with 26 percent opposed.
  • The published record contains no retailer or manufacturer response to the report’s framing, leaving the evidentiary symmetry check incomplete.

The AFL-CIO Tech Institute’s report, “Priced Out, Pushed Out: Electronic Shelf Labels Raise Prices and Shrink Paychecks,” released Tuesday, argues that universal adoption of electronic shelf labels in U.S. grocery stores would cost 44,223 to 191,633 jobs and $1.6 billion to $6.9 billion in lost wages annually. The report’s central analytical claim, as stated by its authors, is that the digital tags connect store shelves to “the same algorithmic pricing software that online retailers are already using,” and that this infrastructure would “make rapid algorithmic price changes easier to implement at scale.” The report arrives alongside a legislative record — Maryland, Connecticut, and New Jersey have each enacted surveillance-pricing bans this year, the first state-level bans in the country — and a May union-commissioned poll reporting 67 percent public support for banning digital price tags and surveillance pricing. The report’s significance lies less in its job-loss projections, whose methodology is not disclosed, than in the conceptual construction that organizes the debate: the bundling of three distinct pricing practices into one regulated category.

The report’s core claim and its construction of “surveillance pricing”

The report, according to its authors, treats electronic shelf labels as the physical delivery mechanism for a pricing capability already standard in online retail. Sunny Glottmann, policy and programs manager at the AFL-CIO Tech Institute and a co-author of the report, described the labels as connected to “the same algorithmic pricing software that online retailers are already using,” and stated that “electronic shelf labels create the infrastructure that would make rapid algorithmic price changes easier to implement at scale, and this raises concerns for consumers that are already struggling with grocery costs.” Lauren McFerran, the institute’s executive director and a former chairman of the National Labor Relations Board, said the labels enable “instantaneous price changes,” and described the technology as “both … enabling this kind of ‘how do we extract the absolute most money out of grocery shoppers that is humanly possible’ and ‘how do we squeeze our workers as much as is humanly possible.’”

The report’s construction of “surveillance pricing” — a term its authors use and are exporting into the broader debate — spans three mechanisms that operate differently and carry different evidentiary records. The first is personalized pricing, in which retailers set prices based on identified customer information; the December Instacart pilot, which ended after an investigation by Consumer Reports and Groundwork Collaborative reportedly found grocery prices differing by as much as 23 percent between customers, provides documented precedent for price discrimination of this kind. The second is contextual dynamic pricing, in which prices respond to environmental or supply signals; the report’s cited example — raising allergy medication prices when pollen counts are high — turns on an environmental data input, not on identified customer information, and the substrate does not specify whether this is a documented practice or a hypothetical. The third is instantaneous price-change infrastructure, a capability claim about the technology itself rather than about any particular data use.

The report’s framing binds these three mechanisms into a single category and, in doing so, constrains the available policy responses to a binary between prohibition and permission. Unbundling them would open a wider set of options: restrictions on identified-customer-data use for pricing, disclosure requirements for pricing algorithms and their contextual inputs, or prohibition of the technology itself. The report’s authors, however, treat the delivery mechanism and the concept as inseparable — pull the labels, the report argues, and shelf-level personalized pricing becomes operationally harder. Whether the engineered concept can be tightened to forbid personalized pricing while preserving time-based dynamic pricing is a revision the three state bans will eventually test in litigation or amendment, according to the substrate’s terms.

The framing’s political utility is documented in the same record that supplies its evidentiary basis. The May United Food and Commercial Workers poll reports 67 percent support for banning digital price tags and surveillance pricing, with 26 percent opposed; 68 percent of respondents believe surveillance pricing will increase grocery costs, and 65 percent say digital price tags will do the same. The bundled category, on this reading, produces a wider coalition than narrower definitions might assemble, because it gathers consumers concerned about personalized data use, consumers concerned about dynamic pricing, and labor organizations concerned about shelf-stocking job displacement under one label.

Who benefits, who bears cost

Four party positions surface in the published record, each with distinct interests. The labor coalition — the AFL-CIO Tech Institute and the United Food and Commercial Workers International Union — holds a substantive interest in preserving the grocery jobs the report projects ESL adoption would displace, and a procedural interest in shaping the regulatory environment before deployment scales. Ademola Oyefeso, vice-president of UFCW International, articulated the identity-and-recognition component of that interest, invoking “the trust stores have built with people” and the expectation that two shoppers pay the same in-store price; he characterized the labor trajectory as workers moving “from full-time to part-time, from part-time to no time,” and described electronic shelf labels as “the embodiment of surge pricing and surveillance pricing,” equating grocery shopping to “airline ticket shopping.” The coalition’s alternatives to a federal ban include continued state-level legislative wins, the poll numbers, and the December Instacart pilot as documented precedent.

Grocery retailers hold a substantive interest in reducing labor cost in a low-margin sector and in deploying dynamic-pricing infrastructure that has become standard in online retail, and a procedural interest in avoiding a state-by-state patchwork that increases compliance cost. Their available alternatives include slower voluntary deployment, voluntary codes of conduct distinguishing dynamic from individualized pricing, and engagement on federal preemption. Retailer counter-arguments and operational data, however, are absent from the substrate; the article records no retailer response to the report’s projections.

Electronic shelf label manufacturers hold a substantive interest in selling hardware into a multi-thousand-store U.S. grocery footprint, with a fallback position in European and other markets where adoption is further along. The report draws on the label manufacturers’ own marketing materials to argue that ESL infrastructure is designed to enable individualized pricing — a sourcing choice that raises the evidentiary floor on the report’s conceptual claim, since the manufacturers are the parties best positioned to know their own product’s capability. Manufacturer responses to that framing are not recorded in the substrate.

Consumers hold a substantive interest in lower and more predictable grocery prices and an identity-and-recognition interest in equal treatment at the point of sale; their alternatives include shopping at non-ESL competitors where they exist, retailer switching, and ballot-driven regulatory responses. The May poll reports 67 percent support for banning digital price tags and surveillance pricing, with 26 percent opposed.

A position-versus-interest analysis surfaces a trade the stated positions do not admit: a statute preserving uniform in-store prices while permitting time-of-day dynamic pricing on non-staple goods is conceptually available, but is not the position either side has declared. Whether such a statute would satisfy the labor coalition’s identity interest in uniform pricing at the shelf while accommodating retailers’ margin interest in dynamic-pricing infrastructure is a question the parties have not answered on the record. Whether the parties are negotiating on merits or running a positional game cannot be confirmed from the substrate, which contains no substantive retailer or manufacturer response to the report.

What happens next

The regulatory pathway is shaped by a documented 2026 state-level record. Maryland became the first state in the United States to ban surveillance pricing in April, Connecticut signed a ban into law in June, and New Jersey signed a ban into law in July. The substrate does not specify whether these statutes regulate the technology, the data inputs, the algorithmic outputs, or some combination — a distinction that will determine their precedential reach.

Four elements structure any forward path regardless of regulatory outcome, per the substrate: continued ESL manufacturer investment in U.S. grocery; the existing state-level bans in Maryland, Connecticut, and New Jersey; the AFL-CIO/UFCW organizational coalition; and the Instacart pilot as documented precedent. Three critical uncertainties open the scenario space. The first is the federal-versus-state regulatory pathway — whether additional legislatures follow the Maryland-Connecticut-New Jersey precedent, whether federal preemption engagement occurs, or whether the AFL-CIO’s call for a federal ban advances. The second is the primary ESL use case in any given retailer — contextual dynamic pricing, personalized pricing using identified customer data, or instantaneous price-change infrastructure that includes neither. The third is the labor-cost-to-price translation — whether the labor-cost reductions projected in the report’s job-loss and wage-loss figures translate to consumer price decreases, increases, or margin absorption; the substrate supports the projection’s direction but not its specific magnitudes, which depend on methodology the report does not detail.

Four scenarios capture most of the variance across two axes — regulatory stringency and ESL adoption velocity. Under high stringency and high adoption, regulatory frameworks constrain how ESL infrastructure can be used even with hardware installed. Under high stringency and low adoption, the three state bans extend into federal action and ESLs remain largely confined to non-grocery retail. Under low stringency and high adoption, ESLs become standard U.S. grocery infrastructure and the algorithmic pricing capability the report warns of becomes operationalized. Under low stringency and low adoption, voluntary industry restraint holds the line on uniform pricing without statutory force.

Per the substrate’s conditions, the low-stringency/high-adoption trajectory is the most plausible absent federal action, given the manufacturers’ own marketing materials the report cites and the labor-cost pressures the report does not dispute; the low-stringency/low-adoption path would require industry behavior inconsistent with those marketing materials; the high-stringency/low-adoption path requires congressional movement not currently on the record; and the high-stringency/high-adoption path requires adoption momentum and regulatory constraint working simultaneously, the hardest of the four to coordinate. The labor coalition’s framing — uniform in-store pricing as a consumer-protection norm — travels across all four scenarios because it does not depend on deployment outcomes; scenario-dependent strategies include retailer voluntary restraint (effective in the low-adoption/low-stringency case, ineffective in the low-stringency/high-adoption case) and a federal ban (effective in the high-stringency/low-adoption case, irrelevant in the low-stringency/high-adoption case).

Leading indicators to monitor, per the substrate’s record: surveillance-pricing legislation in additional states; federal preemption bills; the share of U.S. grocery stores with ESL infrastructure installed; ESL manufacturers’ public statements on individualized pricing capability; and follow-up UFCW polling at the levels already reported. Two wild cards sit outside the matrix: a high-profile incident in which an algorithmic price spike on a staple good is publicly attributed to ESL infrastructure, analogous to the Instacart-pilot dynamic the substrate records; and a federal preemptive statute authorizing ESL use nationwide and foreclosing the state-level coalition strategy. Both are second-tier in probability relative to the four scenarios but high-impact if they occur.

Unresolved questions on the record

The published record leaves several questions open. The methodology behind the report’s 44,223-to-191,633 job-loss projection and its $1.6 billion-to-$6.9 billion wage-loss projection is not described in the substrate. The state statutes’ specific definitions — whether they regulate the technology, the data inputs, the algorithmic outputs, or some combination — are unspecified. Retailer counter-arguments and operational data are absent, as are manufacturer responses to the report’s framing of their own marketing materials. And the status of the pollen-count allergy-medication example — documented practice or hypothetical — is not specified. The most consequential gap is the first: the substrate contains no retailer or manufacturer counter-positioning on the merits, so the evidentiary symmetry check on the report’s central claim cannot be completed from the supplied material.

Analytical techniques used in this piece

This analysis applies the methods below. Each links to a short, plain-English explainer you can read and reuse.

Conceptual Engineering
Asks not just what a concept means but what it should mean, and re-engineers it.
Principled Negotiation
Works a negotiation from interests, options, and objective criteria rather than positions.
Scenario Planning
Builds a small set of distinct, plausible futures to plan against.