The Skimpflation of Everything: Quality Loss at Unchanged Prices, the Limits of Price Statistics, and Generative AI
Skimpflation is a fall in the quality of a good or service without a matching, disclosed cut in its price and without the change being disclosed before purchase. It differs from inflation, which changes the price openly, and from shrinkflation, which changes a quantity printed on the package. This conceptual paper makes three arguments. First, skimpflation largely escapes price measurement, not because particular items lack a quality adjustment method, but because quality adjustment in a consumer price index is made when a recorded characteristic of the priced item changes, and skimpflation mostly changes attributes that are not recorded. Services, where much of what is bought is a person's time and attention, are likely to have more such attributes. Second, skimpflation is better understood as an information problem than as a price problem: it weakens the information a price carries and may shift a verification cost onto buyers. Third, generative AI lowers the cost of one form of skimpflation, the undisclosed replacement of human attention by machine output, across many services at once. The paper sets out the strongest objections, including evidence that AI assistance raises the productivity of less experienced workers and the argument that reputation disciplines hidden quality cuts, and states the evidence that would refute its claims. It closes with five testable propositions and a minimal disclosure standard. No new data were collected. This paper develops the argument of an opinion essay by the same author, published on AI Tools Police on 20 October 2026.
Authors
- Mucahit Kaya (ORCID: https://orcid.org/0009-0007-6379-0416)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-21
- DOI
- https://doi.org/10.5281/zenodo.24418529
- Primary Topic
- Consumer Market Behavior and Pricing
- Type
- preprint