How preferential trade agreements address market distortions from state enterprises

5. How do preferential trade agreements deal with the issue of state enterprises as recipients of subsidies?

5.1. Specific provisions disciplining subsidies received by state enterprises

Evidence from the OECD MAGIC database shows that SEs receive on average relatively more subsidies than their private competitors (Figure 2). While the current WTO rules, including the WTO subsidy rules, embrace an ownership neutral approach by applying without distinction to private and state enterprises, a handful of PTAs disciplining SEs contain rules addressing the issue posed by adverse trade effects of subsidies provided to SEs specifically. The revised OECD SOE Guidelines also specify the case where SOEs are recipients of government support (Box 3).

The CPTPP and, subsequently, the Australia-Peru FTA, the USMCA, the United Kingdom-New Zealand, and the United Kingdom-Australia FTA, include rules governing the provision by a Party of government support – referred as to non-commercial assistance (NCA) – to its SOEs.1 These five agreements define NCA as assistance limited to certain SOEs2 bestowed directly or indirectly3 by governments, its SEs or its SOEs. Similar to the SCM definition of “financial contribution” in Article 1.1(a)(1), the concept of NCA covers various forms of government financial contribution, namely: (i) direct transfer of funds or potential direct transfer of funds or liabilities (e.g. grants, debt forgiveness and preferential loans, loan guarantees, and equity investments) and (ii) the provision of goods or services on terms more favourable than those commercially available to the enterprise concerned. The USMCA includes a third category, which also follows Article 1.1(a)(1) of the SCM Agreement, namely (iii) the purchase of goods on terms more favourable than those commercially available to the enterprise.4 This presumably covers situations where the government provides support by purchasing the goods manufactured by a SE with a view to commercial resale, inconsistent with market considerations. For an NCA to be prohibited, the NCA must cause “adverse effects to the interests of another Party.”5

By inserting the concept of NCA into their PTAs, these agreements have the merits of giving their Parties the legal tools to address the issue posed by subsidies provided by governments and SEs to other SEs without the need to demonstrate that the providing SE is a “public body”. They may notably permit a Party to collect evidence on alleged NCA through a request for information directed to another Party6 and, potentially, to formulate a request for consultations under the Agreement’s dispute settlement mechanism. In other words, although the SCM Agreement rules apply to subsidies received by all enterprises, irrespective of their ownership status, the transparency and adjudicatory mechanisms in place under certain PTAs targeting support benefitting specifically SEs, and granted by governments or SEs, may facilitate their efforts to assemble evidence, as well as to increase the incentives of a Party to a PTA to inquire into such support measures.

Box 3. The treatment of state enterprises as recipients of subsidies under the OECD SOE Guidelines

The OECD SOE Guidelines recommend governments to implement measures to ensure that SOEs’ economic activities do not benefit from any direct or indirect financial support or in-kind inputs that confers an advantage over private competitors. The different types of government support envisaged by the Guidelines are wide-ranging. They include tax concessions and more lenient enforcement of tax rules, below-market lending or financing, state guarantees on SOE debts, trade credits, below-market equity infusions and below-market equity returns, as well as the provision of in-kind inputs, such as goods, energy, water, real estate, data access, land or labour or arrangements (including rights-of-way and concessions) at prices or conditions more favourable than those available to privately owned competitors.

Such a list would permit to capture all forms of subsidies identified under the OECD MAGIC database (i.e. grants, income-tax concessions and below-market borrowings), as well as other forms of support that the OECD has quantified for a sample of sectors or companies (i.e. below-market equity infusions and returns, and below-market energy inputs) (OECD, 2021[5]; OECD, 2023[3]). The list also covers subsidies, such as the provision of below-market inputs or land, for which the OECD has not thus far provided estimates due to data limitations and methodological difficulties.

5.2. Specific provisions disciplining subsidies received by non-financially sustainable state enterprises

The concerns surrounding government support for SEs go beyond the fact that these companies receive relatively more subsidies to include what this support might mean for market entry and exit. Continued subsidisation to non-financially sustainable firms can prop up uncompetitive firms and prevent their exit from the market while warding off new market entries, thus contributing to maintaining capacity that would otherwise not be economically viable. In turn, such subsidies can exert downward pressure on global prices, depress the profits of foreign competitors and prompt the exit of non-subsidised producers, which further increases the weight and participation of SEs in manufacturing value chains.

Evidence from the OECD MAGIC database shows that disciplining support for non-financially sustainable companies,7 notably to firms in which the government owns either a majority share or a minor yet significant share may prove particularly relevant. Previous OECD work has shown that despite receiving relatively more subsidies as compared to private firms, companies having at least 25% government ownership have lower return on assets and equity than firms that are less than 25% government owned (OECD, 2024[1]). The gap between these two groups of companies further widens once the companies’ return on assets and equity are calculated by subtracting, under a purely hypothetical scenario, grants and BMB from their profits before tax (ibid).8 This shows that the greater the extent to which companies are government owned, the lower their performance and, moreover, that subsidies may contribute to obscuring their true performance. That said, this does not show the full extent of the poor performance (in terms of financial difficulties) of some of the companies with at least 25% government ownership.

In this context, the OECD has examined, for the purpose of this report, the presence and magnitude of companies experiencing prolonged financial distress – often described as zombie companies – in the firms covered by the MAGIC database. Aligned with the methodology developed by (Adalet McGowan, Andrews and Millot, 2017[17]), the OECD classifies a firm as a zombie if its interest expenses exceed its earnings before interest and taxes (EBIT) for three consecutive years (the baseline scenario) (Annex E). Figure 4 shows, in this respect, the share of zombie firms among all firms covered by the OECD MAGIC database (left), as well as among those firms with at least 25% government ownership (right), for three different triennial periods (i.e. 2016‑18, 2019‑21, and 2022‑24). Results are also presented in terms of firm revenue to capture better the economic weight of these companies.9 These results show that the number of zombie companies between the two categories does not significantly differ across the three different time periods, thereby suggesting the absence of an evident relationship between government ownership and zombie status.

However, the gap between private firms and firms with more than 25% government ownership widens once, under a purely hypothetical scenario, government grants are removed from EBIT and BMB added back to interest expenses, thus assuming that companies face market-based interest rates (Annex E). In this hypothetical “no-subsidy” counterfactual, the share of zombie firms increases substantially for companies with more than 25% government ownership both in term of firm count and revenue. Such a difference suggests that zombie companies may be more prevalent among those companies that are at least partly government owned and that subsidies may contribute to masking their zombie condition.10 Interestingly, under the “no-subsidy” counterfactual, a few companies remain zombies across the three consecutive time periods (i.e. nine years), hence indicating the presence within the MAGIC database of companies that are artificially kept alive. The fact that most of these companies are majority owned by a government signals that government ownership or government intervention may play a role in their maintenance on the market.

Figure 4. Subsidies can mask the true extent of zombie companies that are at least 25% government owned

Number of zombie firms as a percentage of all MAGIC companies (left) and as a percentage of all MAGIC companies with at least 25% government ownership (right) between 2016 and 2024

Note: Firms lacking the necessary data to calculate EBIT or below-market borrowings are excluded. Readers are advised that the counterfactual “no subsidy scenario” shown here is a hypothetical construct and does not represent what would actually happen were subsidies to be removed. The OECD classifies a firm as a zombie if its interest expenses exceed its earnings before interest and taxes (EBIT) for three consecutive years (the baseline scenario). The definition of zombies relies solely on interest coverage ratios; therefore, some young firms may be classified as zombies.

Source: based on the OECD MAGIC database (accessed 9 March 2026).

Subsidies also appear to be persistent in the case of zombie firms. In this respect, the data also show that companies with at least 25% government ownership that are classified as zombies under the hypothetical scenario during at least one of the time periods covered continue to receive subsidies, most prominently BMB (Figure 5). This suggests that subsidies may contribute to keeping firms in the market that are at least partially state-owned and experience prolonged financial difficulties for at least three years, thus acting as a lifeline.

Figure 5. Below-market borrowings appear particularly pervasive for zombie companies having at least 25% government ownership

Percent of subsidy as of revenue for zombie firms with over 25% government ownership (non-subsidy scenario)

Note: Firms lacking the necessary data to calculate EBIT or below-market borrowings are excluded. The OECD classifies a firm as a zombie if its interest expenses exceed its earnings before interest and taxes (EBIT) for three consecutive years.

Source: based on the OECD MAGIC database (accessed 9 March 2026).

Importantly, zombie firms do not necessarily lose market share despite experiencing prolonged financial distress.11 These findings imply that zombie firms supported by subsidies (as well as firms that would appear as zombies absent subsidies) may be actively competing in the market by expanding or at least maintaining their capacity, with government support acting as a soft budget constraint. The possibility that continuous subsidies to loss-making firms may be distorting the balance between market demand and supply capacity and discouraging new entry is therefore non negligible.

While the issue of zombie companies may also concern private firms, the evidence above suggests it is more prevalent among firms with partial or substantial government ownership. The question is then whether this argues for provisions specifically targeting SEs. Current practice suggests two different models, with some PTAs opting for an SE-specific approach and another set embracing an ownership-neutral approach.

On the one hand, the USMCA contributes to addressing the issue of non-financially sustainable SEs by specifying that three types of NCA bestowed to SOEs by the government of a Party, its SEs or SOEs are presumed to cause adverse effects and thus prohibited outright.12 According to Article 22.6(1) of the USMCA, (i) loans or loan guarantees to uncreditworthy SOEs, (ii) non-commercial assistance to an SOE that, namely is insolvent or on the brink of insolvency without a credible restructuring plan, and (iii) conversion of the outstanding debt to equity when inconsistent with the usual investment practice of a private investor are considered as prohibited NCA without the need to show adverse effects when provided to SOEs primarily engaged in the production or sale of goods other than electricity. This provision captures subsidies provided by SEs and SOEs,13 without the need to prove that the entity concerned is a “public body”.14

On the other hand, a few PTAs recently concluded by the European Union and the United Kingdom have instead introduced an additional prohibited subsidy category on the model of the SCM Agreement. This consists of two types of subsidies, i.e. subsidies in the form of an unlimited guarantee and subsidies bestowed to non-financially sustainable enterprises. However, contrary to the USMCA, this additional category follows the SCM ownership neutral approach: it does not target subsidies to and by SEs and concerns only subsidies by governments or public bodies to all enterprises, irrespective of their ownership status. The EU-Japan Economic Partnership Agreement, for instance, prohibits: (i) arrangements “whereby a government or a public body is responsible for guaranteeing debt or liabilities of an enterprise, without any limitation regarding the amount or duration of such guarantees”15 and “subsidies for restructuring an ailing or insolvent enterprise without the enterprise having prepared a credible restructuring plan.”16

5.3. Potential gaps remain within provisions disciplining subsidies specifically received by state enterprises

By offering Parties to these agreements legal tools to constrain subsidies to SEs causing adverse effect, the few PTAs with provisions on NCAs contribute to addressing the specific issue posed by subsidies given to SEs by governments and, most notably, by other SEs. Yet, notwithstanding that such PTAs deal surgically with the issue posed by subsidies given to SEs, potential gaps within such provisions remain.

First, evidence from the OECD MAGIC database shows that while companies having at least 25% government ownership receive relatively more subsidies as compared to private companies (Figure 2), private firms are also important recipients of government support. This is notably the case in downstream sectors producing high-tech or clean-energy final products, such as semiconductors, telecommunication equipment, solar panels, wind turbines, and electric vehicles where government ownership is less prevalent. For subsidies provided by a government or its SEs to private companies, governments can either make use of the WTO subsidy rules included in the SCM Agreement or, in case they have a PTA in force with the subsidising government, the SCM-based subsidy provisions contained in the PTA.17 As a result, in cases where one SE of a Party to a PTA provides subsidies to a private company, the other affected Party will need to prove that the entity providing the subsidy is a “public body”. Second, the exact reach of the provisions on NCA depends upon the forms of assistance covered, as well as the potential presence of exemptions. It is notable in this respect that the forms of assistance envisaged under the PTA provisions do not include tax concessions, be they in the form of tax credits or tax exemptions.18 In addition, most of the provisions on NCA do not yet apply to sub-central SEs.19 This exclusion may be problematic in countries where companies may be owned by local authorities, at the state, provincial, or municipal level. By way of illustration, the OECD MAGIC database, includes, at the time of writing, 115 companies that are at least 25% government owned; out of these companies, 35 are owned by local governmental authorities.20

Notes

← 1.  The term SOE is used in here because the provisions distinguish between SOEs as recipient of NCA and both SOEs and SEs as providers of NCA. More specifically, the agreements prohibit NCA to SOEs provided by governments, its SEs or its SOEs, which cause adverse effects. While the definition of SOEs under the five agreements concerned corresponds to either Category B or C, (Table 1) SEs are defined as enterprises “owned or controlled through ownership interests by a Party”.

← 2.  The agreements draw from the concept of de jure and de facto “specificity” under the SCM Agreement by specifying that non-commercial assistance is considered to be limited to certain SOEs when access to the assistance is expressly limited to certain SOEs, when the assistance is provided to a limited number of SOEs or used predominantly by certain SOEs, or when a disproportionately large amount of the assistance is provided to certain SOEs. Instances where a Party favours certain SOEs by using its discretion in the provision of assistance are also deemed to be considered as “limited to certain SOEs.”

← 3.  Indirect provision of NCA refers to situations where a government directs or entrusts a private entity to provide support. This is similar to Article 1.1 of the SCM Agreement.

← 4.  Article 22.1 USMCA.

← 5.  Article 17.6 CPTPP, Article 16.6 Australia-Peru FTA, Article 22.6 USMCA, Article 19‑7 United Kingdom-New Zealand FTA, and Article 18‑7 United Kingdom-Australia FTA.

← 6.  As explained below and under Table 3, Parties to these five agreements are entitled to request information on any policies or programmes adopted or maintained by the Party providing for non-commercial assistance.

← 7.  The phrase “non-financially sustainable companies” may refer to, amongst others, uncreditworthy companies, companies that are ailing, insolvent, or on the brink of insolvency without a credible restructuring plan, or companies necessitating a debt-to-equity swap.

← 8.  It must be noted that removing subsidies amounts is a simple accounting exercise showing a purely hypothetical counterfactual that does not reflect what the performance of the companies would be absent subsidies. The counterfactual scenario indeed assumes the absence of any behavioural adjustment without the subsidies. In practice, however, firms may alter prices, production volumes, or financing structures if subsidies are withdrawn.

← 9.  Given that the OECD MAGIC database mainly covers large firms, a size-weighted measure is necessary to assess the impact more accurately.

← 10.  Note that there is a significant overlap between zombie firms having at least 25% government ownership and zombie firms based in China. For the three consecutive time periods (2016‑18, 2019‑21, and 2022‑24), Chinese firms account on average for 60% (by number) and 68% (by revenue) of zombie firms among firms with over 25% government ownership in the baseline scenario, and 81% (by number) and 97% (by revenue) in the hypothetical scenario.

← 11.  In the baseline scenario, almost half of zombie firms increase their market shares during the years in which they face financial difficulties. A similar tendency appears in the hypothetical “no subsidy” counterfactual, where over half of zombie firms continue to expand their market presence. Among firms with more than 25% government ownership, this pattern is also observed.

← 12.  The concept of prohibited subsidies comes from the SCM Agreement, which envisages two types of prohibited subsidies deemed per se specific and causing adverse effects, namely export subsidies and import substitution subsidies. See Article 3.1(a) and (b) SCM.

← 13.  See footnote 11 on the definition of SEs and SOEs under the USMCA.

← 14.  See Box 1 on the difficulties attached to prove that an entity is a “public body”.

← 15.  This provision limits the prohibition to instances where a government or a public body guarantees the debts or liabilities of an enterprise without any limitation as to the amount or duration of the guarantees. By contrast, a governmental guarantee of an enterprise’s debts or liabilities that is limited in time or duration would not fall under the prohibition.

← 16.  Article 12.7 EU-Japan Economic Partnership Agreement. See also Article 11.7 EU-Singapore Free Trade Agreement and Article 11.11 EU-Korea Free Trade Agreement.

← 17.  Most of the PTAs including a subsidy discipline model their rules on the SCM Agreement or, in certain cases, simply include a cross-reference to the Agreement. Note that a few PTAs contain additional subsidy rules going beyond the SCM Agreement. This is notably the case of those PTAs, which incorporate a new prohibited subsidy category. Other PTAs include rules modelled on the EU state aid regime.

← 18.  Note that Figure 2 shows that firms with at least 50% government ownership receive fewer tax concessions as compared to firms that are less than 10% government owned. Yet, this lower rate cannot be only explained by the fact that governments offer fewer and smaller tax gifts to SEs. On the one hand, many of these majority government-owned companies covered in MAGIC are active in heavy industries, where, contrary to the high-tech sector, companies are indeed less likely to receive R&D tax credits, given that they depend less on R&D to maintain their competitive position. On the other hand, by being lower-performing than their private peers, SEs can have smaller taxable income and thus by necessity lower income-tax concessions when expressed as a percentage of the taxable income. This does not, however, mean that governments offer fewer tax credits to this category of firms. but rather that their revenue is smaller. Furthermore, some of the majority-owned companies included in MAGIC are located in the GCC countries where governments may exempt their companies from taxation.

← 19.  See, for instance, Article 22.9 and 22-D USMCA. Furthermore, the provisions on NCA do not apply to non-conforming activities of SEs that a Party lists in its Schedule. Note, however, that Article 17‑14 and Annex 17-C(a) CPTPP, as well as Article 22.14 and Annex 22-C USMCA require the Parties to begin further negotiations within six months of the date of entry into force of the Agreement, with a view to extending the application of the obligations imposed on SEs listed in, respectively, Annex 17-D and Annex 22-D to the activities of sub-central SEs. See also Article 16‑14 and Annex 16-C(1) Australia-Peru FTA.

← 20.  Note that some companies are owned by both central and subcentral government authorities.