CBAM Country Intelligence Moldova 2026: The Dniester Evidence Border, Steel’s Default Route Inversion and Cement’s Internal Carbon Border

Executive Summary

Moldova is a small exporter in global industrial terms, but it is an unusually revealing case for the European Union Carbon Border Adjustment Mechanism (CBAM). A single customs origin, the Republic of Moldova, overlays several operational spaces that do not fully coincide. Industrial production, official statistics, customs administration, electricity supply, environmental supervision and access to installation-level emissions evidence follow different territorial and institutional boundaries. The definitive CBAM regime makes these differences economically relevant because actual embedded emissions depend increasingly on evidence generated at the installation rather than on a country label alone. [1–5]

This report names the resulting Moldovan structure the Dniester Evidence Border: the discontinuity between the circulation of goods and the circulation of carbon evidence across Moldova’s divided governance space. Economic operators from the Transnistrian region file declarations with Moldova’s Customs Service, which confirmed in January 2024 that left-bank firms were operating under the new Customs Code. Moldova’s National Bureau of Statistics, by contrast, excludes enterprises on the left bank of the Dniester and in Bender from the statistical territory used in its standard foreign trade releases. A shipment can enter Moldovan customs administration while the producing enterprise remains outside the statistical universe normally used to describe Moldova’s exports. [6,7]

The implications become more significant when CBAM reaches the installation. The Moldavian Metallurgical Plant, known as MMZ or Moldova Steel Works, is located in Rîbnița on the left bank. Its physical technology is well established: a scrap-based electric arc furnace route with an active 120 tonne EAF, a 95 MVA transformer, secondary metallurgy and continuous casting. The plant’s physical identity is fundamentally different from integrated blast furnace and basic oxygen furnace steelmaking. [8,14]

Yet the corrected definitive default table does not provide a Moldova-specific value for MMZ’s principal CN 7213 and CN 7214 20 00 long steel products. Moldova’s country table contains only cement and fertiliser sections. Under Regulation (EU) 2026/1740, a listed country with no value for the relevant product falls back to the “Other countries and territories” table, where the base default for these products is 3.996 tCO₂e per tonne and the associated production route is Route C, carbon steel based on BF/BOF. The 2026 steel mark-up is 10 per cent, which applied mechanically to the published base value gives approximately 4.40 tCO₂e/t before any registry rounding. [4]

The report defines this outcome as Default Route Inversion. MMZ is physically Scrap/EAF, while the fallback default treats its long products through a BF/BOF regulatory route. The inversion originates in the default architecture and the absence of a Moldova-specific steel value; it does not imply that European authorities believe MMZ operates a blast furnace. Its importance lies in the financial difference between a verified actual route and the regulatory fallback.

Free allocation moderates the gross default without removing the discontinuity. Regulation (EU) 2025/2620 links default emissions to a route-specific default CBAM benchmark. For CN 7213 and CN 7214 20 00, Route C carries a default benchmark of 1.364 tCO₂e/t. With the 2026 CBAM factor of 97.5 per cent and a 100 per cent cross-sectoral correction factor, the corresponding free allocation adjustment is about 1.33 tCO₂e/t, so a default emissions value of roughly 4.40 tCO₂e/t implies a net quantity of approximately 3.07 certificates per tonne before any eligible deduction for a carbon price effectively paid in the country of origin. [3,18,19]

The actual route is more installation-specific. Regulation (EU) 2025/2620 assigns a process-related benchmark of 0.038 tCO₂e/t to the relevant rolling products when actual data are used, but long steel is a complex good for free allocation purposes, and its actual specific embedded free allocation must also incorporate the verified free allocation embedded in relevant precursors. A credible actual route requires a complete chain through steelmaking and rolling rather than a simple subtraction of 0.038 from the plant’s direct emissions. [2,3]

This distinction underpins the report’s broader concept of Evidence Gated Carbon Identity. Physical production creates emissions, monitoring converts production into data, verification tests those data, and legal admissibility determines whether the data can enter the CBAM declaration. When the actual evidence chain is usable, the regulatory carbon identity can reflect the installation; when it is not, the default system supplies another identity. In Moldova the two identities can be far apart.

MMZ’s location makes verification a material open question. Definitive CBAM verification normally relies on physical site visits in the first verified year, with virtual approaches permitted under defined exceptional circumstances. As of September 2026 the CBAM verification market was only becoming operational, and the first definitive-period verification reports were expected from 2027. The available evidence supports an elevated evidence access and admissibility risk for MMZ without establishing that verification will fail. [5,9]

Cement provides a different test of the same national structure. Holcim Rezina operates on the right bank within Moldova’s central environmental system and the corporate governance of an international cement group. Rîbnița Cement operates on the left bank within a distinct regulatory and energy environment while maintaining documented commercial links with Moldova and Romania. The plant-level distribution of Moldovan cement exports is difficult to reconstruct from public statistics because standard national trade data exclude left-bank enterprises, while EU mirror data identify country of origin rather than individual installation. [20–28]

The cement comparison supports a second sector concept, the Internal Carbon Border: the uneven environmental, energy, monitoring and market conditions faced by installations inside Moldova’s internationally recognised economic space. It is a regulatory discontinuity, not a fiscal frontier, and its economic relevance is likely to increase as Moldova moves toward EU climate legislation, installation-level MRV and eventual carbon pricing.

Steel and cement together show why physical carbon intensity alone cannot predict Moldova’s CBAM outcome. That outcome depends on the interaction of production technology, trade geography, evidence admissibility and regulatory treatment. Moldova is best viewed as an extreme conjunction case: none of these mechanisms is unique in isolation, but their concentration within a small economy, with major industrial assets divided across the Dniester and trade increasingly oriented toward the European Union, makes them unusually visible.

For Moldova’s EU accession, the central issue is moving from legal alignment to operational carbon governance. The country has adopted climate legislation, begun constructing an EU-aligned monitoring, reporting, verification and accreditation system, integrated its electricity market more closely with Europe and placed carbon pricing on its reform path. [10–17] The long-term question is whether those institutions can produce credible and enforceable carbon information across the industrial economy associated with Moldova as a whole.

The commercial implication follows from the same structure. Public CBAM formulas and default values make generic carbon calculation increasingly easy to automate. The harder decision problem is identifying the producing installation, assessing the quality and legal usability of emissions evidence, determining whether actual data can survive verification, and calculating the regulatory fallback if they cannot. For Moldova steel in particular, CBAM Evidence Admissibility Due Diligence is a plausible commercial use case with a clear value proposition, although demonstrated willingness to pay for it as a standalone service remains limited.