Nowcasting

GDP growth and HICP inflation · sparse-group Lasso

Issued estimate

Finland · 2026Q4

2026Q4Current period

1.7%

Year-on-year · SG-Lasso

Indicative 80% range: 0.0 to 3.5%

0.0 pp since 19 Oct 2026

Snapshot available from 20 Oct 2026 · 16:52 UTC+3. Indicative ranges use revised-data errors, without historical PIT calibration.

Daily nowcast history

Last issued estimate per UTC day · same target period

5 days with issued estimates. Missing days remain gaps.

Latest data releases

Observed updates and their effect on the selected nowcast

Release / input	Reference period	Latest value	Recorded	Nowcast effect, pp
GDP growthRevision · 13 changed periods	Apr 2026	1.4 % y/y	20 Oct 2026First seen by AEIS	0.000
Construction output volumeNew observation · 18 changed periods	Aug 2026	12.9 % y/y	19 Oct 2026First seen by AEIS	+0.108
Headline inflationRevision · 1 changed period	Sept 2026	2.7 % y/y	16 Oct 2026First seen by AEIS	—
EnergyRevision · 1 changed period	Sept 2026	13.1 % y/y	16 Oct 2026First seen by AEIS	—
FoodRevision · 1 changed period	Sept 2026	2.4 % y/y	16 Oct 2026First seen by AEIS	—
Services pricesRevision · 1 changed period	Sept 2026	1.7 % y/y	16 Oct 2026First seen by AEIS	—

Effects cover all changed observations for that input between issued snapshots. A dash means no comparable effect is available for this target. Source publication and first-seen dates are distinct.

What changed the forecast?

Change since the previous issued snapshot · percentage points

Previous 1.7%Change 0.0 ppCurrent 1.7%

No change at the displayed precision.

Input effects include new observations and revisions, evaluated with the previous fitted model. Calendar and refit effects are separate. These are model contributions, not causal shocks or survey surprises.

Inputs behind this estimate

Construction output volumeConstruction confidenceConsumer confidenceEuro-area industrial productionIndustrial productionIndustrial confidenceRetail sales volumeServices confidenceUnemployment

35% of current lag features are missing and use training-sample medians. Unsupported series are omitted.

View omitted inputs (0)

Model checks

Chronological diagnostics · revised data

1.47RMSE, pp

0.95MAE, pp

24Validation points

The tuning sample is also used for these diagnostics. This is not an independent historical PIT backtest or a live accuracy claim.

How the model works