The case for optimism about August 2026 rests on a reasonable instinct: Romania’s registration totals now blend two very different kinds of business formation, and the incorporated company (SRL) share has been thinning. If the headline number is diluted by a growing layer of lighter-weight registrations, the argument goes, then a bad total might conceal a healthier core. The right test is to strip out everything that isn’t an SRL and re-rank the Augusts.

Run that test, and the rebound does not appear. It gets slightly worse.

The raw picture, and the rank

Romania recorded 8,103 registrations in August 2026, against 11,435 in August 2025 — a fall of 29.14%.

Set that against every other August in the series, and 2026-08 ranks 9th of 10, ahead only of 2024’s 5,288. The August series runs: 2022 (17,586), 2021 (12,677), 2023 (11,731), 2025 (11,435), 2020 (11,238), 2018 (10,242), 2017 (9,451), 2019 (9,059), 2026 (8,103), 2024 (5,288).

Two features of that ordering matter. The first is that 2024 — not 2026 — is the weakest August in the ten-year series, at 5,288 registrations. August 2026 is actually 53% larger than August 2024. This undercuts the framing of 2026 as a collapse: relative to that trough, registrations are well up. The second is that the comparison base matters enormously to the −29% headline. August 2025 was the fourth-strongest August in the window, in line with August 2023’s 11,731, so the year-on-year figure is measured against a normal recent August, not an inflated one.

Now rebase to SRL only

The SRL counts for each August are: 2022: 12,295 | 2021: 8,637 | 2023: 8,044 | 2025: 7,613 | 2020: 7,025 | 2018: 6,930 | 2017: 6,237 | 2019: 6,160 | 2026: 4,419 | 2024: 3,846.

The ranking is identical: 9th of 10 — the SRL count for August 2026 was 4,419. Rebasing changes the levels and the percentage gaps, but it does not move August 2026’s position in the same-month hierarchy at all. The reason is mechanical: the SRL share in 2026-08 is 54.5% (4,419 of 8,103), which is genuinely lower than in most prior Augusts — around 69% in 2023, 68% in 2019, 68% in 2018 — but the SRL count itself is also the second-lowest of any August in the ten-year window, above only 2024’s 3,846. Shrinking the numerator while acknowledging a smaller numerator does not produce a better ranking. It just produces a more SRL-specific version of the same story.

Where rebasing does change the picture is in the severity of the year-on-year move. On the total, 2026-08 is down 29.14%. On SRLs alone, it is down 41.95% — from 7,613 to 4,419. On PFA (sole-trader) registrations, it is down only 1.9%, from 3,415 to 3,350. So the better denominator makes the year-on-year headline more alarming, not less. The decline is concentrated almost entirely in the incorporated-company layer, while the sole-trader layer has held essentially flat.

That is the honest answer: the SRL-era denominator shift does not make the August 2026 headline look better. It sharpens the diagnosis by pointing at where the weakness actually sits.

What the mix is doing

The entity breakdown for 2026-08 shows SRLs at 4,419, PFAs at 3,350, and II (individual enterprises) at 283, with the remaining legal forms in single or low double digits. The SRL share of 54.5% is a meaningful change in composition: it means roughly 45% of new business registrations in the month were not incorporated companies.

That composition shift is real and worth watching on its own terms, but it should not be mistaken for a hidden strength. A PFA is a lighter, lower-commitment vehicle than an SRL, and a flat PFA count alongside a sharply lower SRL count describes a registration mix tilting toward less capitalised, more provisional forms. The total falling 29% while both of its two largest components fall — one sharply, one barely — is not dilution masking a robust core. It is one layer weakening and another simply not growing.

Exits fell too — and that matters

Business lifecycle data complicates any simple “things are getting worse” reading. Total exits in August 2026 were 8,744, comprising 4,531 deregistrations, 3,198 dissolutions, and 1,015 suspensions.

Every exit category fell sharply year-on-year: suspensions −31.42%, dissolutions −31.0%, deregistrations −19.0%. The churn rate — exits as a share of registrations — stood at 107.91%, with a health ratio of 0.93 and a net change of −641.

So August 2026 was a month of contracting activity on both sides of the ledger: fewer businesses formed, and fewer wound up. That symmetry is important. It argues against describing the month as a wave of closures. It is closer to a general slowdown in registration and exit activity alike, which is consistent with the seasonal story.

That seasonal story needs stating carefully. August is, by the mean of the available record, the weakest calendar month of the year apart from January and December — it ranks 10th of 12, with a mean of 10,681 registrations across ten observed years. A low August is not, in itself, news. But the seasonal floor does not fully account for a reading 24% below the August mean, nor for a rank 9th of 10 within Augusts.

The net-negative run is now nine months long, beginning in December 2025, cumulating −16,756 net. That streak should be read with care: reliable exit data begins only in July 2024, so a nine-month run cannot be compared meaningfully to the full historical record. What the monthly sequence does show is that August’s deficit of −641 is the second-smallest in the run, after February’s −119 — but it is not a straight line of improvement. The deficit narrowed from December’s −4,292 to February’s −119, then widened again through March and April to −3,018, before falling back over the last four months. August is the mildest month since February, not the seventh step of an unbroken decline in the deficit.

Where the decline lives

The industry data reinforces the SRL-concentrated reading. Nearly every sector fell year-on-year: hotels and restaurants −52.26%, health and social care −50.21%, real estate −48.95%, manufacturing −46.71%, other services −38.98%, retail and vehicle repair −38.11%, construction −33.05%, education −31.35%, professional and scientific activities −22.47%. Transport, the largest single sector, fell least, at −6.02%, and information and communications slipped only 2.69%.

Only two sectors grew: financial intermediation and insurance, +16.2%, and agriculture, +4.2%. Those are small in absolute terms — 165 and 149 registrations respectively — but the pattern of services and consumer-facing sectors falling hardest while transport and IT hold up is consistent with the SRL-heavy sectors taking the brunt.

Geographically, Bucharest alone accounted for 2,276 registrations, followed by Ilfov at 567 and Cluj at 383. No county posted a meaningful gain; the “top growing” list is led by Vaslui at +5.0%, with every other listed county contracting.

The conclusion

Rebasing August 2026 to its SRL component does not rehabilitate the month’s standing. August 2026 ranks 9th of 10 Augusts on totals and 9th of 10 on SRLs alone — the position is invariant to the denominator. If anything, the core measure is the harsher one: the SRL count is down 41.95% year-on-year against a 29.14% fall in the total, because the sole-trader layer has barely moved.

The genuinely mitigating facts are elsewhere. August 2024, not 2026, is the weakest August in the ten-year series, so 2026 is not the low point of the record. Exit activity fell in every category, so this is not a closure wave. And August’s net deficit is the smallest since February in the nine-month negative run.

What the data will not support is the claim that a shift toward the SRL-and-PFA mix reveals a stronger core beneath a weak headline. The mix shift is real; the strength is not in the numbers. The weakness is concentrated in incorporated companies, and rebasing simply makes that concentration visible.