Romania registered 8,103 new businesses in August 2026, the second-lowest August total in the ten-year window for which the Trade Register holds comparable data, surpassed only by the anomalous trough of 5,288 recorded in August 2024. The figure sits 29% below August 2025’s 11,435 and 27% below July 2026’s 11,049. It is also the ninth consecutive month in which business exits have outnumbered new formations — a streak running since December 2025 that has erased a cumulative net of 16,756 businesses from the registry.

That convergence of facts is striking. The question this article sets out to answer — the one the registry data is now explicitly forcing — is whether August 2026 represents a genuine, durable reduction in Romanian entrepreneurial activity, or simply the calendar doing what it normally does in summer. The answer, on close examination, turns out to depend on a distinction the data is structurally incapable of resolving cleanly.

---

What Ten Augusts Actually Look Like

The seasonality record from 2017 onward is unambiguous on one point: August is one of Romania’s weakest registration months of the calendar year, ranking 10th out of 12 with a long-run mean of 10,681 formations. Only January (10,531) and December (9,680) score lower. August is not modestly soft — it sits roughly 1,000 formations below the next-strongest month (April, at 11,681) and about 2,600 below the peak month of March (13,289).

That makes the seasonal framing initially plausible. Except that individual August totals tell a more complicated story:

Year	August Registrations
2017	9,451
2018	10,242
2019	9,059
2020	11,238
2021	12,677
2022	17,586
2023	11,731
2024	5,288
2025	11,435
2026	8,103

Three things stand out immediately. First, the variance across Augusts is enormous — a spread of more than 12,000 between the weakest and strongest years. Second, 2024 was itself an extreme outlier at 5,288, making 2025’s bounce to 11,435 look like a recovery and 2026’s 8,103 look like something in between. Third, 2022’s 17,586 — the post-pandemic registration surge — sets a ceiling that subsequent years have not approached.

This variance is the core analytical problem. When the month’s natural range runs from 5,288 to 17,586, a total of 8,103 is not automatically alarming. Excluding 2024’s anomalous trough, August 2026 is the lowest August in the ten-year window — below even 2017 (9,451) and 2019 (9,059), the two quietest pre-pandemic Augusts. But 2026 is only modestly below those earlier readings, not dramatically outside a plausible range. The seasonality mean of 10,681 implies August 2026 is roughly 2,600 registrations below the historical average for the month — a meaningful gap, though one that the 2022 post-pandemic surge inflates considerably.

---

The Entity-Type Split Focuses the Problem

The broadest year-on-year decline belongs almost entirely to one legal form. SRLs — Romania’s standard incorporated limited liability company — fell from 7,613 in August 2025 to 4,419 in August 2026, a drop of 3,194 entities or −42%. PFAs — the sole-trader form — fell by only 65 registrations, a statistically negligible −1.9%.

This bifurcation matters here because it complicates the seasonal frame. If the August slowdown were primarily a calendar effect — people on holiday, notaries underworked, administrative processing slow — it should affect all entity types roughly proportionally. It does not. PFAs, which require less administrative effort to register, held nearly flat. SRLs, which require more formal incorporation steps, collapsed. That asymmetry is consistent with a demand-side explanation specific to incorporated entities rather than a supply-side explanation attributable to August’s calendar position.

Yet the asymmetry does not clinch a structural diagnosis either. PFA formation has been rising relative to SRL formation for months, driven partly by the ongoing regulatory and cost environment that makes PFA status more attractive for sole operators. August may simply be revealing that shift in sharper relief because the absolute SRL numbers were already trending down before August arrived.

---

The Exits Window: Useful, But Bounded

On the other side of the ledger, August 2026 recorded 8,744 total business exits, comprising 1,015 suspensions, 3,198 dissolutions, and 4,531 deregistrations. Against 8,103 registrations, this produces a churn rate of 107.9% — meaning exits exceeded entries by 641 businesses.

That net loss of 641 is, notably, the second-smallest monthly deficit in the nine-month negative streak, after February’s −119. In December 2025, net growth hit −4,292; by August 2026 it has narrowed to −641. The health ratio — registrations divided by exits — ticked up by 0.01, and the net growth trend improved by 389 entities month-on-month. The ecosystem is still contracting, but the rate of contraction is easing.

Here is where the data encounters its hardest limit. The exit figures are trustworthy only from July 2024 onward, meaning we have at most 26 months of exit data — barely more than two calendar cycles. To determine whether August 2026’s exit volume is seasonally elevated, depressed, or normal for August, we would need multiple Augusts of comparable exit data. Besides this one, we have exactly two: August 2024 and August 2025, when total exits were 11,709 against 11,435 registrations — a churn rate of 102.4%.

Comparing those two data points: exits fell sharply year-on-year in August 2026. Suspensions dropped −31% versus August 2025, dissolutions fell −31%, and deregistrations fell −19%. In other words, August 2026 is a month of fewer new businesses and fewer exits — the registry is simply quieter overall. Whether that joint quietness is August doing what August does, or whether it reflects an underlying cooling of economic activity in both the entry and exit directions, cannot be established from two comparable data points. The exits series is too short.

---

The 12-Month Moving Average Confirms the Downtrend, Not Its Cause

The 12-month moving average as of August 2026 stands at 12,627, down from 12,904 the previous month. The current monthly total of 8,103 sits 4,524 units — 36% — below that moving average. This is a meaningful gap: it indicates that August 2026 is not just slightly below trend but substantially so.

The moving average is the one instrument that genuinely transcends seasonal noise, because it smooths across all twelve months simultaneously. Its direction — downward — tells us that the structural 12-month rate of formation is declining. But it cannot tell us whether August is accelerating that decline or temporarily exaggerating it. A single month that is both seasonally weak and cyclically weaker than recent Augusts will produce exactly the reading we see, regardless of whether the underlying cause is seasonal, structural, or both.

---

Industries Offer Partial Disambiguation

The industry breakdown adds one further data point to the seasonal-versus-structural debate. Transport & Storage led all sectors in August 2026 with 1,515 registrations, down only −6% from August 2025. IT & Communications fell just −2.69%. Both sectors are dominated by PFA-form registrations and appear broadly stable year-on-year.

By contrast, Retail & Wholesale fell −38%, Hotels & Restaurants fell −52%, and Manufacturing fell −47%. These are sectors where SRL formation dominates. The pattern is internally consistent: sectors that register primarily as SRLs are down sharply; sectors that register primarily as PFAs are roughly stable. This again localises the August decline to the SRL form rather than to economic activity as a whole — but it does so in a way that is present every month in 2026, not specifically in August.

---

What the Data Can and Cannot Settle

Whether August 2026’s registration total is a seasonal event or a structural step-down is a question the registry data is definitionally unable to resolve, for three compounding reasons.

First, the seasonal baseline is too noisy. August’s ten-year range spans 5,288 to 17,586. A reading of 8,103 is low relative to the average, and the lowest August in the window outside the 2024 anomaly, but it is not dramatically below pre-pandemic floors. With that degree of variance, no single observation can cleanly separate “this is a seasonally weak month” from “this is a structurally depressed month.”

Second, the exits series is too short. Establishing whether August 2026’s net loss of 641 businesses is seasonally normal would require at least five Augusts of comparable exit data. We have three, counting this one, and the 2024 data sits at the very boundary of the trustworthy window. Any conclusion about whether exits are seasonally elevated or cyclically driven is underdetermined by the evidence.

Third, the SRL decline predates August and persists across all months of 2026. The cumulative nine-month net loss of 16,756 businesses is not a function of August’s calendar position; it is a continuous process that happens to be visible in August as well. August may be amplifying a trend that is already running, but it did not create it, and removing the seasonal effect — if it could be quantified — would not eliminate the underlying negative net growth.

What the data can confirm is narrower but real: excluding 2024’s anomalous trough, August 2026 is the lowest August in the comparable registration window; the SRL form accounts for virtually all of the year-on-year decline; and the exit side of the equation is moving in parallel with entry, suggesting a broadly quieter market rather than a concentrated wave of closures. Whether that quietness is a temporary valley or the new floor is a question the registry, by design, cannot answer from one month’s observation.

The seasonal label explains some of August’s softness and the structural label explains some more — and the data holds enough ambiguity that the precise proportion of each remains, for now, genuinely unknown.