The finding, stated plainly

Romania recorded negative net business formation in all three months of the June–August 2026 quarter. But the three months did not behave the same way. The deficit narrowed in each successive month — from −1,748 in June to −1,030 in July to −641 in August.

That is the opposite of an accelerating deterioration. The honest answer to the question of whether this quarter’s net destruction is steady, worsening, or easing is: easing.

The bigger streak, and its limits

The August reading extends a run of negative net months to nine, beginning in December 2025 and cumulating to −16,756.

That number needs a health warning attached. The measurement itself only begins in July 2024, the earliest month for which exit data is considered trustworthy. Nine months of net contraction within a 25-month observation window is not an all-time record and should not be described as one. It is a nine-month window into a specific measurement regime.

Within that window, the nine months split into two distinct phases. December 2025 through May 2026 was erratic and severe: −4,292, −2,686, −119, −1,171, −3,018, −2,051. The most extreme single month was the first one. The current phase — June, July, August — is neither erratic nor severe. It is a monotonic sequence of shrinking losses.

Quarter vs quarter: the trend is improvement

Arranged as calendar quarters within the trustworthy exits window:

Quarter	Monthly net figures	Quarter total
Q1 2026	−2,686, −119, −1,171	−3,976
Q2 2026	−3,018, −2,051, −1,748*	−6,817
Q3 2026	−1,748, −1,030, −641	−3,419

*June sits in Q2 by calendar but is the first month of the quarter under analysis; the June–August total is 3,419.

On a like-for-like three-month basis, the June–August deficit (−3,419, summing the monthly figures above) is roughly half the April–June deficit (−6,817). Even measured strictly against the immediately prior quarter of the same length, net destruction shrank by about half.

Compared with the same quarter a year earlier, however, the picture inverts. In June, July and August 2025, net growth was −836, +1,059 and −274 respectively. Summed, that quarter was roughly net positive by about 50 firms, versus a deficit of 3,419 this year. So: substantially better than the preceding quarter, substantially worse than the same quarter last year.

Both things are true. The improving short-run trajectory is real; so is the fact that the level has not returned to where it stood twelve months ago.

Why the deficit is narrowing: exits are falling faster than entries

The improvement is not being driven by a registrations surge. August registrations were 8,103, down 29.1% year-on-year. June registrations of 11,277 were down only 4.2% year-on-year, and July’s 11,049 were down 27.2%.

What is happening is on the exit side. August exits totalled 8,744 against 13,025 in June and 12,079 in July. All three exit categories fell year-on-year in August: suspensions −31.4%, dissolutions −31.0%, deregistrations −19.0%.

The ecosystem’s registrations-to-exits ratio tells the same story more compactly: 0.87 in June, 0.91 in July and 0.93 in August. A ratio below 1.0 still means contraction, but the distance from 1.0 has been closing for three consecutive months.

One caveat matters here: the August ratio sits below the 0.98 recorded in August 2025. The gradient is favourable; the level is not.

The seasonal question, answered by the data rather than assumed

There is a persistent assumption that Romania’s summer months are a weak period for registrations. The seasonality data supports this partially. August ranks 10th of 12 calendar months, averaging 10,681 registrations. July ranks 7th, averaging 11,926, while October — the third-strongest month — averages 12,992. June is actually the fifth-strongest month, at 12,119.

So August’s low absolute number is partly structural — it is, historically, one of the two weakest months of the year. That is why the sharp 28.2% drop from June to August registrations should not be read as a sudden collapse. The month-to-month decline is normal; the year-over-year decline of 29.1% is the more informative signal, and August 2025 was also a weak August, yet in 2025 the net result that month was only −274. Seasonality explains the level of August registrations. It does not explain why August 2026 ended deeply below the line while August 2025 ended near it.

That gap — not seasonality, not a new shock — is where the year-over-year story lives.

Entity types: the incorporated-to-sole-trader drift continues, and widened

The SRL-to-PFA shift that has been documented in these pages since January remains the quarter’s most consistent structural feature.

In June, SRL formations fell 18.8% year-on-year to 6,233 while PFA registrations rose 23.4% to 4,535. In July, the SRL decline steepened to −45.0% (5,963 units) while PFA rose 20.0% to 4,662. In August, SRL registrations were 4,419, down 42.0% year-on-year, while the PFA count of 3,350 slipped 1.9%.

August is the first month of the quarter in which the PFA side did not expand. That is a meaningful nuance to the standing narrative: the shift toward sole-trader forms is not simply an unstoppable reallocation — in August, both major forms contracted, and PFA’s year-on-year change turned marginally negative for the first time in the quarter. The incorporated segment, meanwhile, contracted by roughly 40% in two of the quarter’s three months. This is a reallocation and a broad contraction: SRL declining much faster than PFA, but with PFA no longer reliably growing.

Industries: the composition is shrinking everywhere

No sector escaped the year-over-year pullback in August. Six of the largest registration categories all declined:

• Retail and wholesale trade: −38.1%
• Other service activities: −39.0%
• Construction: −33.1%
• Professional, scientific and technical activities: −22.5%
• Transport and storage: −6.0% (the mildest decline)
• Information and communication: −2.7%

Transport and storage remains the single largest source of new registrations — 1,515 in August, 2,061 in June and 2,307 in July. It is also the most resilient sector year-on-year. The depth of the declines in trade, construction and miscellaneous services suggests the contraction is concentrated in domestically-oriented, capital-light sectors where sole traders and small SRLs cluster.

The conclusion

The quarter’s three months all ended below the line, but they ended successively closer to it, and the nine-month run of negative net formation has been losing force rather than gathering it.

The quarter is easing. Three consecutive months of shrinking deficits, a health ratio climbing from 0.87 to 0.93, and exits falling faster than registrations across all three exit categories. There is nothing in the June–August sequence that resembles acceleration.

The nine-month run is a plateau, not a deepening. The severe months were December 2025 through May 2026, with a peak deficit of 4,292. The three months since have been progressively milder, and August’s deficit of 641 is the second-smallest of the entire nine-month run. Within the trustworthy window, the trend points toward zero — not away from it.

The caveat is the one the data keeps insisting on: the level remains negative, and remains worse than a year ago, when the same quarter was approximately neutral. Improvement toward the line is not the same as reaching it. Whether the coming quarters complete that approach or stall short of it is, as of the end of August 2026, an open question the data does not yet answer.