Statistical agencies often revise growth figures a posteriori, and generally upwards. But this takes time, as their data often become more comprehensive and accurate after two years. France is seeing the largest upward revisions over this timeframe, and, generally, these revisions are more pronounced in Europe than in the United States.
Our study focuses on 2023 and 2024, two years for which statistical agencies have more information (growth figures have already, in most cases, been revised), and the figures are compared to pre-COVID levels (an average of 2017–2019). The COVID-19 period, which was inherently disruptive, is excluded from the study.
Growth figures are usually revised upward: more in Europe than in the US, particularly in FranceRepeated shocks delay the positive diagnosis
There have been many shocks in recent years — some of them significant — which have made estimating national accounts a complicated task for statistical agencies. The downward revision of French growth figures for the first and second quarters of 2026 (from -0.1% to -0.2% QoQ and from 0.2% to 0% QoQ, respectively) is a recent example. Why? Primarily because one of the most severe droughts in history had a greater impact on agricultural production than initially estimated. This impact was hard to gauge when the first estimate was released.
The most significant revisions occur much later and are often upward revisions
Growth figures can be revised several years later. By then, the data are more comprehensive. Also, it is easier to distinguish long-term trends from cyclical shocks. Therefore, growth can finally surprise on the upside, particularly due to the role played by innovation, which improves the quality of goods and services and/or lowers their prices.
However, the distinction between volume and price is central to GDP statistics. Most of the data which provide the basis for these statistics come from value-based series, such as turnover figures. In order to publish volume-growth figures, statistical agencies must choose a base year — i.e. the estimated price level during that year — and then calculate the change between that base year and subsequent years. This chained-index methodology is used to calculate volume series at constant prices.
But innovation, which improves the quality of goods and services, means that the price for a given quality decreases. By how much, exactly? Statistical agencies face difficulties when trying to take it into account, until they have more information. This is why their estimates become more accurate over time — often two years later — when the available data are more comprehensive. This usually leads them to revise GDP growth upwards, as shown in our chart, though the size of these revisions varies by country and period.
France enjoys the strongest upward revisions
The size of the revisions is, on average, more moderate for the years 2023–24 than it was before the COVID-19 pandemic. However, it is still significant for some countries. This is the case for France, which ranks first irrespective of the period. We see this result as another positive sign of the ongoing transformation of the French economy, which is also reflected in the increase in business creations (see our analysis). For example, in 2023, growth reached 1.4% after revision, an upward revision of 0.5 percentage points.
Apart from 2023 — which was an exceptional year when Germany was hit harder by the energy crisis than any other country — these positive revisions are more pronounced in Europe than in the United States. This is not closing the growth gap between the two regions, but it is narrowing it somewhat. The development of innovations and their contribution to growth may thus be harder to capture, especially in the short term, but they have a positive impact.
Growth figures reflect, in particular, long-term trends: structural GDP growth, or so-called “potential” growth. Our analysis shows that, as soon as statistical agencies are able to estimate it more accurately, growth figures come out higher.
