Quarterly GDP: an experimental model
Experimental. None of these models forecasts GDP significantly better than a naive forecast (last quarter’s growth). They are published to show the work and its limits, not as a reliable estimate.
For Q3 2026, the models range from +0.03% to +0.38%. INSEE’s first figure is expected around 30 October 2026.
| Model | Estimate | Past error (RMSE) | Right direction | p vs naive |
|---|---|---|---|---|
| Naive forecast (last quarter) | +0.25% | 0.313 | 82% | — |
| Historical average | +0.38% | 0.336 | 82% | — |
| Industrial production only | +0.03% | 0.313 | 76% | 0.50 |
| Production + services climate | +0.05% | 0.319 | 79% | 0.57 |
| Production + services + consumption | +0.12% | 0.308 | 79% | 0.44 |
| Production, business climate, past GDP (ridge) | +0.12% | 0.332 | 82% | 0.71 |
Past error: test from Q1 2016 to Q2 2026, 34 quarters excluding 2020–2021, using only what was known on the day of INSEE’s first figure. p: the probability that the gap with the naive forecast is due to chance.
GDP growth: published figures and past estimates
Quarterly change in real GDP (%), since 2016, excluding 2020–2021. Each estimate uses only the data available on its date.
- Published by INSEE
- Best model (production + services + consumption)
- Naive forecast
Show the data
| Quarter | Published by INSEE | Best model (production + services + consumption) | Naive forecast |
|---|---|---|---|
| Q2 2026 | +0.04% | +0.50% | +0.17% |
| Q1 2026 | -0.18% | +0.05% | +0.34% |
| Q4 2025 | +0.26% | +0.51% | +0.40% |
| Q3 2025 | +0.42% | +0.54% | +0.33% |
| Q2 2025 | +0.24% | -0.04% | +0.30% |
| Q1 2025 | +0.17% | +0.47% | +0.24% |
| Q4 2024 | -0.01% | +0.26% | +0.41% |
| Q3 2024 | +0.43% | +0.41% | +0.33% |
| Q2 2024 | +0.22% | +0.29% | +0.34% |
| Q1 2024 | +0.24% | +0.48% | +0.44% |
| Q4 2023 | +0.50% | +0.32% | +0.38% |
| Q3 2023 | +0.37% | +0.32% | +0.60% |
| Q2 2023 | +0.93% | +0.61% | +0.29% |
| Q1 2023 | +0.14% | +0.60% | +0.40% |
| Q4 2022 | +0.40% | +0.13% | +0.44% |
| Q3 2022 | +0.51% | +0.54% | +0.39% |
| Q2 2022 | +0.40% | +0.27% | +0.23% |
| Q1 2022 | -0.01% | +0.57% | +0.44% |
| Q4 2021 | +0.51% | +0.64% | +1.39% |
| Q3 2021 | +2.92% | +0.66% | +0.71% |
| Q2 2021 | +1.20% | 0.00% | +0.42% |
| Q1 2021 | +0.45% | +0.27% | +0.24% |
| Q4 2020 | -0.01% | +0.18% | +5.77% |
| Q3 2020 | +14.10% | +6.46% | -4.87% |
| Q2 2020 | -13.03% | -7.83% | -1.77% |
| Q1 2020 | -5.13% | +0.63% | +0.04% |
| Q4 2019 | -0.50% | +0.37% | +0.27% |
| Q3 2019 | +0.04% | +0.07% | +0.44% |
| Q2 2019 | +0.49% | +0.77% | +0.62% |
| Q1 2019 | +0.95% | +0.72% | +0.53% |
| Q4 2018 | +0.72% | +0.35% | +0.46% |
| Q3 2018 | +0.55% | +0.57% | +0.33% |
| Q2 2018 | +0.23% | +0.23% | +0.23% |
| Q1 2018 | -0.06% | +0.16% | +0.48% |
| Q4 2017 | +0.60% | +0.88% | +0.54% |
| Q3 2017 | +0.75% | +0.52% | +0.53% |
| Q2 2017 | +0.73% | +0.80% | +0.50% |
| Q1 2017 | +0.66% | +0.36% | +0.49% |
| Q4 2016 | +0.64% | +0.72% | +0.31% |
| Q3 2016 | +0.18% | +0.34% | +0.15% |
| Q2 2016 | -0.26% | +0.28% | +0.37% |
| Q1 2016 | +0.31% | +0.75% | +0.35% |
How it is built
A bridge equation links quarterly GDP growth to industrial production growth. Months INSEE hasn’t published yet are filled with this site’s monthly estimate. Two variants add the services business climate and household consumption of goods. All were tested as on the day of INSEE’s first estimate, about 30 days after the quarter ends.
Why it doesn’t work yet
Since 2016, French quarterly GDP growth has been small and steady: last quarter’s growth already captures most of it. Industrial production really helps when all three months of the quarter are known, but the third is published after GDP, and estimating it erases the gain. Surveys add little at this horizon.
What could change that
A better monthly estimate of production, or real-time consumption data such as the card payments the Banque de France uses, which aren’t openly available. The test reruns every day: if a model takes the lead, this page will say so.