Using the Press to Construct a New Indicator of Inflation Perceptions in France

Olivier de Bandt, Jean-Charles Bricongne, Julien Denes, Alexandre Dhenin, Annabelle de Gaye, Marwan Menaa

Key findings

• More than one million articles from 30 French national and regional newspapers, magazines and news agencies are turned into a monthly indicator.
• Supervised machine learning labels each article by the direction of prices it reports: rising, falling or stable.
• The indicator moves closely with households' inflation expectations in the European Commission survey and with headline HICP inflation.
• It helps forecast inflation one quarter ahead in a Phillips curve framework.

Abstract

We build an indicator of inflation perceptions in France from more than one million press articles published since 2004. Articles about consumer prices are selected, then classified by the direction of prices they report using supervised machine learning. The resulting balance of opinions tracks households' inflation expectations and headline inflation, and improves short-term inflation forecasts.

The French press indicator from this paper is published by the Banque de France as an official series on Webstat.

@article{bandt2026ijcb,
  title   = {Using the Press to Construct a New Indicator of Inflation Perceptions in France},
  author  = {Olivier de Bandt and Jean-Charles Bricongne and Julien Denes and Alexandre Dhenin and Annabelle de Gaye and Marwan Menaa},
  journal = {International Journal of Central Banking},
  year    = {2026},
  note    = {Forthcoming}
}