Consumer Price Index — Algeria, continuous monthly series (2002-2026)

One row per month and per scope — Ville d'Alger (city of Algiers) or National. periode is formatted YYYY-MM, and indice uses the base declared in the base column.

Coverage. On base 2001 = 100: 293 uninterrupted months, from January 2002 to May 2026, for both scopes. On base 1989 = 100: 48 months between 2002 and 2009, incomplete.

Honest columns. origine states where the retained value comes from (that month's bulletin, a later bulletin, a multi-year compendium) and flags revisions. sources_concordantes equals 1 for 28 values, all before 2008: they are not wrong, but nothing confirms them, and that should be known before building a calculation on them.

How this series is built, and why it holds up.

The ONS does not publish a series: it publishes one bulletin per month, as a PDF. Restoring continuity means retrieving several hundred documents, extracting their tables and stitching them together — that work is what is sold here, not the raw data, which remains freely available at source.

Three independent sources per value. A given month appears in that month's own bulletin (two decimals), in the comparative tables of the following twelve bulletins (one decimal), and in the multi-year compendia (an entirely different layout). The three paths are checked against each other: across 654 cross-checked months, 98.5% agree. The sources_concordantes column states, for every row, how many distinct documents carry the same value — up to sixteen for recent months.

ONS revisions are tracked. Ten months were corrected by the institute after first publication: the value given here is the most recent one, and the origine column is marked as a revision so the change stays visible.

Nothing is interpolated. No value is computed, smoothed or estimated. A missing month stays missing.

Two index bases, never to be concatenated. The ONS rebased its index from 1989=100 to 2001=100 between July and October 2009. Both scales coexist in the file, told apart by the base column: for the city of Algiers, the same month reads about 700 on the 1989 base and 127 on the 2001 base. Stacking them would show an 80% collapse that never happened. Splicing them requires a coefficient computed over an overlapping period — deliberately not done here, for lack of sufficient overlap in the published documents.