The night before GDP · Romania, Q2 2026

THE CURRENT READ

The figure landed close to the dial’s centre.

On the basis followed by the nowcast, INS published minus 0.4% year on year. Our estimate was minus 0.5%, so the difference was plus 0.1 percentage points and stayed within the range. The high-frequency picture caught the broad direction, not every GDP component.

LIKE-FOR-LIKE

Gross, Q2 2026 against Q2 2025: INS minus 0.4%; nowcast minus 0.5%.

KEEP SEPARATEOn the seasonally adjusted series, GDP was 0.0% against the previous quarter and minus 2.0% against Q2 2025. These are different bases: we do not mix them with the nowcast’s gross target.

THE PUBLISHED RESULT

The official figure and our estimate, on the same basis.

INS published its flash estimate on 14 August. We place it beside the nowcast frozen before the release and keep the seasonally adjusted reading separate.

INS · GROSS, Y/Y

-0.4%

Q2 2026 GDP against Q2 2025. This is the direct comparison with the nowcast target.

NOWCAST · FROZEN

-0.5%

Our estimate published before the flash, with a -1.3 ... +0.3% working range.

DIFFERENCE

+0.1 pp

The official figure was 0.1 percentage points less negative. It landed inside the range.

INS · ADJUSTED, QUARTERLY

0.0%

Q2 against Q1 in real terms. Our orientation was -0.2 ... +0.5%.

WATCH THE BASIS: ON THE SEASONALLY ADJUSTED SERIES, Q2 WAS 2.0% BELOW Q2 2025. THAT IS NOT THE SAME MEASURE AS THE NOWCAST’S GROSS YEAR-ON-YEAR TARGET.

THE MEASURING INSTRUMENT

What stayed frozen before publication?

Each indicator covers a different part of the economy and arrives with a different lag. These controls stress-test the pre-release estimate: removing one does not change the INS figure or erase an official GDP contribution.

estimate frozen · 13 Aug 20265mechanisms

OUR CALCULATION-0.5%an/an · year on year

WORKING RANGE-1.3 ... +0.3%working range

With all five mechanisms active, the frozen estimate is -0.5%.

Editorial sensitivity, not a GDP contribution.

purchasing power

Wages

-7.0% real May 2026 vs May 2025 pulls down sensitivity -0.2 pp

Net pay was 5,684 lei, 3.2% above last year in nominal terms, but the real-wage index was 93.1. The 2.7% monthly fall also follows bonuses paid in April.

INTERPRETATIONIncome looks better on paper than it does at the checkout. This is a demand signal, not a measure of the median income.

INS · wages

adjusted retail

Consumption

-4.7% May 2026 year on year strong drag sensitivity -0.4 pp

The volume of retail turnover, adjusted for working days and seasonality, fell 4.7% in May and 5.3% in the first five months.

INTERPRETATIONDomestic demand remains the clearest cold-weather signal for households, even though retail is not all final consumption.

INS · retail

adjusted output

Industry

-5.3% May 2026 year on year pulls down sensitivity -0.3 pp

Adjusted industrial production was 5.3% below May 2025; the adjusted decline was 3.1% across the first five months. Manufacturing fell more than the total.

INTERPRETATIONFactories do not yet confirm a broad recovery. Output is not the same as value added, but the direction is hard to ignore.

INS · industry

work volume

Construction

+16.4% May 2026 year on year, adjusted lifts sensitivity +0.6 pp

Construction volume rose 16.4% in May on the adjusted series and 11.3% across the first five months. The gross cumulative figure was +12.4%.

INTERPRETATIONBuilding sites are the visible counterweight: investment can support the quarter while everyday demand weakens.

INS · construction

FOB value

Exports

+2.4% January–May 2026 year on year small lift sensitivity +0.3 pp

FOB exports reached 40.337 billion euro, 2.4% above the comparable period. In May, exports rose 4.5% to 8.324 billion euro.

INTERPRETATIONForeign demand offers nominal support, but prices, the exchange rate and the product mix can hide weaker exported quantities.

INS · foreign trade

Select an indicator to remove it from the estimate. Its displayed sensitivity moves the centre; each removal widens the band by 0.1 percentage points.

WHY EXPERIENCE DIFFERS

People can see one economy. GDP can count another.

GDP adds production, consumption, investment and foreign trade, then adjusts for prices and seasonality. A household feels real pay and the shopping basket. A building site enters the accounts through investment. Both statements can be true in the same months.

FACT

The household

Prices faster than pay

Net pay rose 3.2% in nominal terms, but real pay was 7% lower. Adjusted retail fell 4.7% in May.

FACT

The building site

Work volume rising

Adjusted construction was 16.4% above May 2025. Investment can lift the quarter without immediately repairing a household budget.

INTERPRETATION

The factory and the border

Weak output, better nominal exports

Adjusted industry fell 5.3% in May, while FOB exports rose 4.5%. Prices and the product mix can explain some of the gap.

The energy witness

Final electricity consumption fell 3.4% in the first five months. We do not count it separately, to avoid counting the same demand cooling twice.

-3,4%final energy consumption, Jan–May

HOW WE CALCULATED IT

How we got to the estimate.

We built the estimate in five steps. First, we fixed the anchor: Q1 2026 was 1.2% below Q1 2025 and flat against Q4 2025 on the seasonally adjusted series. Then we tracked five indicators over January to May: real wages, adjusted retail, adjusted industry, construction work and FOB exports. We keep their percentages separate because they measure different things and none is a GDP contribution. Three signals point to weakness and two offer support. The NBR is an external reference for a small quarter-on-quarter recovery. From this picture, we set -0.5% as the working centre. For seasonally adjusted quarter-on-quarter GDP, the orientation is near zero to mildly positive, roughly -0.2% ... +0.5%. That is a separate orientation. The nowcast and range remain frozen on 13 August; we compare them with the published figure rather than rewriting them.

ESTIMATE LOGICQ1 anchor · direction of five signals · editorial choice

01 · ANCHOR
Q1: -1.2% year on year and 0.0% quarter on quarter.
02 · SIGNALS
Wages -7.0% real; retail -4.7%; industry -5.3%; construction +16.4%; exports +2.4% nominal.
03 · DIRECTION
Three signals point to weakness, while construction and exports offer support. The NBR provides an external reference for a small quarter-on-quarter recovery.
04 · CENTRE
-0.5% is our editorial working point, between the Q1 anchor and the direction of the signals.
05 · RANGE
-1.3 ... +0.3% is the working band. It leaves room for components that monthly data cannot yet show.

WHAT WE CANNOT SEE YET

full services, inventories, agriculture, real imports, the GDP deflator, statistical discrepancy and INS revisions

UNKNOWN

A good signal in one sector can be hidden by weakness in another. That is why the dial does not promise a precise number.

The formula shows how we reached the estimate. We do not add indicator percentages or invent GDP weights. The dial sensitivities are stress tests, not official contributions and not an econometric model. The published result is compared separately on the same gross year-on-year basis.

SOURCES AND LIMITS

Every hand comes with a footnote.

We link each series, period and transformation to a place where it can be checked. For the published result, we cite the INS release through the AGERPRES report; TEMPO is the methodological ledger.

PUBLISHED RESULTINS · via AGERPRES

Q2 2026 GDP, flash estimate

GDP was flat against Q1 on the seasonally adjusted series; against Q2 2025, it fell 0.4% on the gross series and 2.0% on the adjusted series. First-half GDP: minus 0.8% gross and minus 1.6% adjusted.

published 14 Aug 2026, 09:51

Open source

OFFICIAL SERIESINS · TEMPO

Quarterly GDP, CON106J

The aggregate definition, estimation methods and sources used in Romania’s quarterly national accounts.

metadata checked on 13 Aug 2026

Open source

OFFICIAL SERIESINS · TEMPO

Wages, retail, industry, construction

FOM106D + FOM110A, COM1073, IND104P and CNS105N: monthly values and volume indices with the transformations listed in the ledger.

January–May 2026

Open source

FOREIGN TRADEINS · TEMPO

FOB exports, EXP101A

Goods exports at FOB value. The dial uses a nominal value, not a real-volume export measure.

January–May 2026

Open source

EXTERNAL CHECKNBR

8 July monetary-policy decision

The NBR saw a slight Q2 recovery from Q1, with uneven movement: less weakness in retail, stronger construction and weaker industry.

Q2 2026, information available on 8 Jul 2026

Open source AGERPRES report

FISCAL CONTEXTMINISTRY OF FINANCE

Budget execution at 30 June

The first-half deficit was reported at 2.0% of GDP, versus 3.64% in the same period of 2025. It is context, not a separate input to the calculation.

January–June 2026

Open source

DEMAND WITNESSINS · ENERGY

Final electricity consumption

Final consumption fell 3.4% in the first five months; households fell 11.3%, while final consumption in the economy fell 0.8%.

January–May 2026

Open source

EUROPEAN CHECKEUROSTAT

Eurostatistics, June 2026

Romania’s flat quarter-on-quarter GDP reading in Q1 2026 and European reference points for production and consumption.

published in June 2026

Open source

CALENDARPUBLICATION REFERENCE

Q2 GDP flash estimate

The publication calendar followed for 14 August 2026. The event is not an input to the estimate frozen on 13 August.

editorial window

Open source

READING LIMITS

Average net pay is not median pay and includes bonus, sector-mix and statistical-coverage effects.
Retail is a consumption proxy, not all final consumption; population services have a separate path.
Industrial production and construction work are activity indicators. GDP measures value added and uses other sources too.
The export hand is in euro and FOB. It is not real exports and does not replace net exports in the national accounts.
Data can be revised. The nowcast uses only information available on 13 August; the adjusted-series minus 2.0% is not compared directly with the gross year-on-year target.