This section highlights the characteristics of steel firms that increase their likelihood of receiving subsidies through grants and BMB. Section 0 presents a number of graphs related to the impact of government ownership, while Section 0 uses various estimation techniques to try to disentangle the individual contributions of each factor.1
The drivers and impacts of subsidies to steel firms
Higher government ownership, higher subsidies
SOEs generally benefit from more subsidies and government support than privately owned firms across many industrial sectors (OECD, 2024[16]). As indicated in Figure 8, the data shows this is also the case in the steel sector. This result holds across instrument types. Figure 9 below depicts the proportion of subsidies received by a firm relative to its asset size. It shows that subsides almost continuously increase as the government ownership share increases, and across all instruments. For example, steel firms with government ownership exceeding 25% receive, relative to their asset size, subsidies in the form of grants and corporate income tax concessions that are twice as large as those received by firms with less than 10% government ownership. Even more notably, these firms receive three times more subsidies in the form of BMB (Figure 9).2 This finding indicates that a larger government ownership share is associated with a higher average subsidy provided. This can occur through direct cash transfers (e.g., grants) or through more indirect instruments such as income tax concessions and BMB. Additionally, higher BMB may result from government financial institutions offering cheaper finance or from implicit guarantees that reduce the risk profile of lending to a steel firm, thereby attracting subsidised borrowing from private market participants.
Government ownership increases the amount of subsidisation received per unit of a steel firm’s assets
Note: Each company is equally weighted in the average above. Subsidies are estimated for each different instruments, and compared to a firm’s asset size, although comparing to a firms’ revenue as in Figure 8 would likely provide a similar picture.
Source: OECD MAGIC database.
Higher government ownership also implies higher indebtedness and lower profitability
The extend of government ownership also correlates with steel firms’ financial performance. For example, Figure 10 shows that a higher government ownership share usually entails higher indebtedness (debt to assets ratio), and lower return on equity (RoE) and return on assets (RoA). Those results collaborate findings from (OECD, 2025[17]) for the specific case of the steel sector. Steel firms owned by more than 50% by the government have RoE and RoA that are almost three times less than those for which the government ownership is less than 10%, while their indebtedness is about 30% higher. Furthermore, in China SOEs' average debt-to-capacity ratio is 2.5 times that of private enterprises, suggesting that significant financial imbalances are driven by state support.
Financial performance worsens with higher degrees of government involvement
Source: OECD MAGIC Database, merged with ownership database.
The interest coverage ratio,3 another complementary financial indicator, provides a similar picture (Figure 11). The interest coverage ratio is on average half less when governments own a majority share in the company compared to when they own less than 10% of the steel firm, although the relationship between interest coverage ratio and government ownership is not monotonously decreasing.
Note: The interest coverage ratio is a financial ratio that measures a company's ability to meet its interest payments on outstanding debt. Interest coverage is typically used by investors, lenders, and analysts to assess a company's risk profile and financial stability. It is defined as EBITDA divided by Interest Expense (the total amount a company pays in interest on its outstanding debt over the year).
Source: OECD MAGIC Database.
To conclude, there is a partial correlation between a firm’s indebtedness, its RoA and RoE, other financial variables (e.g. the interest coverage ratio), a firm’s percentage share of government ownership, and the subsidies it benefits from. Hence, there is the need to disentangle their effects (Section 10345021).
Multivariate regression analysis can attribute the amount of additional subsidies to individual characteristics of firms, by controlling for other non-firm-specific factors included in the regression.4 This is the approach chosen in this section to address the question: do governments provide more subsidies to steel SOEs due to their government ownership status, or because they generally tend to subsidise poorly financially performing firms more? The findings indicate that it is primarily the share of government ownership that matters (Section 0); financial performance does not significantly impact the provision of subsidies once government ownership is properly accounted for (Section 0). To assess the impact of firms’ characteristics on their provision of grants and BMB, this section uses Pseudo-Poisson Maximum Likelihood (PPML) estimations, in a simplified framework similar to (Branstetter, Li and Ren, 2023[18]),5 and complemented by multivariate regressions to further confirm the sign of the impacts. The results of the different estimations are presented in Annex E and are summarised here.
Share of state ownership in the steel company
Multivariate regressions confirm what the visual inspection of the data suggested: a steel firm’s government ownership share is a very significant variable in explaining subsidies received by firms, both through grants and through BMB (Annex E). The result holds when all observations are taken together. When observations from OECD Member countries and partner economies are used as separate panels for the regression estimations, noticeable differences with respect to grants emerge. Government ownership had a positive impact on the total amount of grants provided to a steel firm in partner economies, whereas it had no statistically significant impact in OECD Member countries. An explanation may be that in OECD Member countries the provision of grants follows more transparent and equitable processes. This could be the result of a successful implementation of the principles set out in the OECD Recommendation on Competitive Neutrality (OECD, 2021[19]) with regard to the provisions of grants, as well as the revised guidelines 2024 OECD Guidelines on Corporate Governance of State-Owned Enterprises.
The OECD Recommendation on Competitive Neutrality aims to ensure that SOEs and private businesses compete on a level playing field. It promotes principles that reduce distortions arising from government ownership or involvement in markets. The recommendation encourages governments to:
Ensure SOEs do not benefit from undue advantages such as favourable regulations, subsidies, or preferential treatment.
Maintain transparency regarding the relationship between governments and SOEs, particularly in financial flows and objectives.
Implement competition law equally across both SOEs and private firms.
Ensure governance frameworks for SOEs foster commercial and competitive neutrality principles, while minimising market distortions.
These principles help improve market efficiency, encourage fair competition, and enhance economic performance globally.
The Competitive Neutrality Toolkit, which was developed by the OECD Competition Committee1 to support the implementation of the principles set out in the OECD Recommendation on Competitive Neutrality, provides a set of good practices, based on examples from international experience, to support public officials in identifying and reducing distortions to competition due to state intervention (OECD, 2024[20]).
1. Working Party No. 2 on Competition and Regulation.
The OECD Guidelines on Corporate Governance of State-Owned Enterprises provide internationally agreed benchmarks designed to help governments improve the governance and performance of their SOEs. These guidelines address the unique challenges of state ownership, promoting transparency, accountability, and integrity while ensuring that SOEs operate efficiently and fairly alongside private enterprises.
The 2024 Guidelines revision emphasise the integration of climate-related goals and resilience into state ownership practices, with recommendations for SOEs to set ambitious sustainability objectives, align their operations with these targets, and enhance risk management systems to include sustainability considerations.
Governments are encouraged to professionalise their ownership role by clearly defining the rationales for state ownership and setting strategic mandates for SOEs. The guidelines advocate for granting SOE boards sufficient autonomy to make decisions while establishing frameworks to ensure transparency and accountability. Boards are also expected to possess the skills, independence, and integrity required to oversee corporate strategy effectively and fulfil their responsibilities without political interference. SOEs are called upon to adhere to high standards of disclosure and reporting, comparable to those of listed private companies. This includes financial performance, governance structures, and any public policy objectives they are tasked to fulfil. The guidelines also highlight the importance of stakeholder engagement, requiring SOEs to be accountable not only to shareholders but also to the broader public, given their role in serving societal interests. By fostering competitive neutrality, the guidelines ensure that SOEs do not receive undue advantages such as preferential financing or regulatory exemptions. Instead, they should operate under the same market conditions as private enterprises, promoting fair competition and efficiency in the marketplace.
The revised Guidelines place a strong emphasis on fostering a level playing field between SOEs and private enterprises to avoid market distortions and ensure fair competition. They encourage governments to eliminate practices that confer undue competitive advantages to SOEs, such as preferential financing, tax exemptions, or lenient regulatory treatment. By reinforcing the principles of competitive neutrality, the Guidelines ensure that SOEs operate under market conditions that mirror those faced by private firms, promoting efficiency and accountability. This approach is vital not only for supporting vibrant and competitive markets but also for maintaining public trust in the integrity of state ownership practices.
In summary, the Guidelines provide a robust framework for governments to manage their SOEs in a way that supports economic security, competitiveness, and sustainable development while ensuring fairness, transparency, and public accountability.
Nevertheless, and contrary to grants, government ownership is correlated to BMB in OECD Member countries, implying firms with higher government ownership are more likely to receive BMB, even in OECD Member countries (Annex E). In partner economies, government ownership is correlated to both grants and BMB: firms with higher government ownership are more likely to receive subsidies through both cash grants and BMB.
Geographical location
Introducing country dummies resulted in only Luxembourg and China showing statistical significance, with a positive impact for China (i.e., additional subsidies in China relative to baseline) and a negative one (i.e., fewer subsidies) for Luxembourg.6 This suggests, as highlighted in previous research (Mercier and Giua, 2023[3]), that national contexts play a large role in the provision of subsidies.
Firm’s size
There is a noticeable disconnect between a firm's asset size and its revenue when it comes to subsidies. While larger asset size leads to more subsidies, higher revenue does not have the same effect and may even result in fewer subsidies in the case of grants. This suggests that subsidies are more closely tied to the value of a firm's assets rather than its revenue-generating ability. A reason may be that policymakers may prioritise asset size over revenue when allocating subsidies, given the lower volatility of assets compared to revenue. This could have implications for how resources are distributed among firms of different sizes and efficiencies.7 Splitting the sample into OECD Member economies and partner economies observations and re-estimating regressions yield very contrasting results. Total asset is a significant explanatory variable for both grants and BMB for partner economies, whereas it is non-significant for OECD Member countries, and even negatively impacts the provision of grants in one of the estimations (Annex E). This contrast could be the result of a conscious selection in partner economies of the largest (private or public) steel firms, deemed more able to achieve policy goals or compete overseas due to their size. Alternatively, it could be the consequence of non-transparent subsidy application systems that would favour larger firms with better knowledge of government approval processes or better connections to government officials working in responsible agencies, and/or the result of support schemes being more open to smaller firms in OECD Member countries, which could be the result of an attribution of the grants essentially based on proposed projects, rather than the reputation of the size of the firm leading or implementing the project.
Indebtedness
Larger indebtedness negatively affects subsidy provision the following year in OECD Member countries, whereas it positively affects the provision of subsidies in partner economies or is not statistically significant (in the case of linear regressions). When regressions are estimated on all observations collectively, a steel firm's indebtedness - as measured by its debt-to-asset ratio - is not a statistically significant explanatory variable for grants. However, when the data is partitioned and the two panels are estimated separately, higher indebtedness negatively affects the provision of subsidies in the following year in OECD Member countries. In contrast, in partner economies, higher indebtedness either positively affects the provision of subsidies or is not statistically significant (as observed in linear regressions). Since the regressions control for the percentage of government ownership, this greater propensity to subsidise more indebted firms in partner economies, compared to OECD Member countries, cannot be solely attributed to state ownership fulfilling non-market policy objectives. A similar contrast is observed for BMB: indebtedness is not statistically significant in explaining BMB in OECD Member countries, whereas in partner economies, higher indebtedness positively impacts the provision of BMB.
Missing factors: political connections, etc.
There are many explanatory variables that, for lack of data, could not be included as explanatory variables in our regression. For example, political connections could be a meaningful variable to explain the provision of subsidies. It is probably, to some extent, captured by the share of government ownership, but not in its entirety: private companies may have informal links to the government or regulators at the corporate management level, rather than through financial ownership (Box 7) (Chong-en, Chang-Tai and Zheng, 2019[21]).
Political connections, while difficult to quantify, are likely a significant explanatory variable in understanding the provision of subsidies in China’s steel sector and beyond. The rise of state-connected private owners illustrates how these connections extend beyond formal government ownership. Recent analysis shows that in 2019, 65% of China’s largest 1 000 private owners had direct equity ties1 with state-owned entities, while millions more had indirect links through complex investment networks (Stanford University, 2023[22]). These connections blur the line between state and private ownership, suggesting that subsidies may not only be directed towards state-owned enterprises but also to privately owned firms with strong political or regulatory ties.
While the share of government ownership may partially capture this dynamic, informal relationships at the corporate management level- such as guanxi (关系) a Chinese term referring to personal networks and social connections that facilitate mutual favours and trust or other forms of political influence - likely play a key role in subsidy allocation. Guanxi with government officials may provide firms with resource-bridging capabilities, granting access to critical resources such as low-interest bank loans, land, and regulatory approvals (Chen and Wu, 2011[23]). This informal social capital is particularly valuable in many transitional economies like China, where institutional structures are underdeveloped. Moreover, guanxi may enable firms to gain insider information on government policies or industry reforms, further enhancing their ability to access subsidies and other financial support.
1. Direct equity ties refer to ownership arrangements where both private owners and state-owned entities hold significant stakes (e.g., at least 10%) in the same joint venture or company, establishing a formal financial relationship.
Firm’s revenue, EBITDA and gross profit margins
Profitability indicators exhibit varying significance in explaining subsidy allocation to steel firms. In OECD Member countries, revenue is a significant explanatory factor for grants, whereas EBITDA (Earnings Before Interest, Taxes, Depreciation, and Amortization) is not. In partner economies, the opposite is observed: EBITDA is significant, but revenue is not. In partner economies, revenue has a positive contribution to BMB, while EBITDA has a negative contribution. There is considerable coefficient instability, as different estimation methods produce conflicting results regarding the impact of EBITDA on BMB in partner economies. Due to significant correlations among revenue, EBITDA, and gross profit margins, it is challenging to draw definitive conclusions about whether larger profits attract larger subsidies in either OECD Member countries or partner economies. The only clear finding is that gross profit margins were never statistically significant, even when other profitability indicators like EBITDA and revenue were excluded from the analysis to avoid multicollinearity issues. A plausible reason may be that unlike total assets and indebtedness, which are "stock" variables and are thus more stable variables on which government can base their decisions, profitability indicators are closely tied to a firm's current position in the business cycle, which can change more rapidly. It is thus possible that governments do not consider solely the previous year’s profitability indicators for deciding to subsidise firms, which would negate their statistical significance in the regressions. Furthermore, subsidies provided through channels other than grants and BMB (such as reduced input costs) inevitably boost a firm's profitability. This can result in an unclear impact of profitability on the provision of grants.
Lagging global steel demand in USD
Regressions included a “GDP steel” variable obtained by multiplying world steel demand by a world average steel price and meant as a control variable for the Steel business cycle. The indicator thus represents a very aggregate steel global demand in money terms (USD). Nevertheless, this indicator is often not statistically significant.
Capacity utilisation
Capacity utilisation, as a percentage of capacity used to produce at the firm level, rarely had statistical significance in our regressions. This suggests that, although some jurisdictions set country targets of production, steel firms do not individually receive subsidies for achieving specific and high production targets. It is more likely that production targets can be met through capacity expansions, rather than through higher capacity utilisation of current capacity. However, Chinese SOEs displayed CU rates above 84% between 2011 and 2014, significantly higher than the domestic average of 63-67% and the 76% of POEs counterpart. This suggests that government support likely played a role in sustaining elevated utilisation rates among SOEs, even under less favourable market conditions.
The table below provides a succinct overview of the impact of firms’ characteristics on the provision of subsidies in the form of grants and BMB:
Note: Precise estimate signs for each estimates are provided in Annex E. For the sake of clarity some distinctions and nuances are not reported in the above table, which summarises results across all model specification and for both subsidy and subsidy intensities. “unclear” indicates a significant number of regression specifications yielded non-statistically significant result, and the sign represents the sign of contributions for those regressions that yielded some statistically significant coefficient for the variable.
← 1. Although we would like to explain subsidies, the presence of correlation - even using lagged explanatory variables as in all of our regression estimates (Annex E) - does not inherently imply causation, hence some caution in the interpretation of results is warranted.
← 2. Notice that, although firms with a larger state ownership share do receive considerably more BMB than private firms, the graph does not suggest that the unit impact of BMB, which is not depicted nor assessed in this picture, differs depending on whether it is received by an SOE or privately owned firm.
← 3. An interest coverage ratio of 10 means that the earnings, before interest payments and taxes (EBITDA), of a steel firm, could pay up to 10 times the interest expenses of the company for that year.
← 4. A caveat must be considered: if explanatory variables are too much correlated, then attributing precisely an impact to any of them could be difficult. This will result in coefficient instability when performing the robustness checks of re-estimating the equations after withdrawing one or a number of explanatory variables. For that reason, it is important to re-estimate models with a subset of the correlated variables (e.g. taking only one variable related to profit or sales instead of many).
← 5. The models used in this report do not assess Total Factor Productivity for subsequent inclusion in PPML regressions, and it is in that sense that they are simplified compared to (Branstetter, Li and Ren, 2023[18]).
← 6. Companies whose headquarters are located in Luxembourg are more internationally oriented, which may explain a somehow lower degree of their total subsidisation than other more domestically oriented peers.