Read beyond the noise.

A base effect can make an annual change look unusually large

A newsroom desk with two printed neutral bar-chart silhouettes and a calendar with unreadable small text, no fabricated numbers.
AI-generated photographic illustration; not documentary photography.

A headline reports that annual growth has slowed sharply. The latest level, however, has barely changed. Both statements can be correct when an unusual movement a year earlier drops out of the comparison. That is the basic idea behind a base effect.

The European Central Bank’s explanation of inflation base effects describes how atypical movements in the comparison period can influence changes in an annual rate. The denominator is part of the story, not merely a number beneath the fraction.

Work through an invented index

Imagine an index at 100 in January of one year and 110 in February. A year later, it is 121 in both January and February. The newer monthly level has not moved, but the annual comparisons differ:

  • January: (121 ÷ 100 − 1) × 100 = 21%.
  • February: (121 ÷ 110 − 1) × 100 = 10%.

The annual rate falls from 21% to 10% because the comparison base rises from 100 to 110. The example is deliberately simplified and is not a claim about any published economy or price series.

Notice what did not happen: the latest index did not fall from 121. A slower rate of increase should not be casually rewritten as a fall in the level.

What to check in a real report

Find the current level, the previous month’s level and the year-earlier comparison. Then check whether the article is discussing a monthly change, an annual change or a change in that annual rate. Those are three different calculations.

Seasonal adjustment, revisions and the composition of an index may add further complications. Use the statistical release’s definitions rather than reconstructing a precise result from a rounded headline alone.

A helpful annotation is: “How much of this change comes from what happened now, and how much from what happened then?” A base effect does not make a statistic fake. It explains why a true percentage can still give an incomplete impression when its comparison period remains invisible.