The drivers and impacts of subsidies to steel firms

2. Subsidies are pervasive but very unequally distributed across time, jurisdictions, and firms

Subsidies are pervasive and persistent and can be provided through various instruments

This report makes extensive use of subsidy data and estimations from the OECD MAGIC database (‎Annex B). The data accurately capture three instruments largely used by governments and government-related entities to channel subsidies towards steel firms: cash grants, below market borrowings (BMB), and even, for about two thirds of firms in the sample, corporate income tax concessions.

Subsidies in the steel sector seem both pervasive and persistent. Only 6% of the 742 firm-year observations contained in the data correspond to zero subsidies, either in the form of grants, tax concessions, or below-market borrowings. Meanwhile, nearly 90% of the firms covered received subsidies for at least half of the period during which they are included in the database. This indicates that very few steel firms among those covered did not obtain any subsidies between 2005 and 2022 and that firms which receive subsidies tend to receive them for several years in a row.

Overall, steel firms appear to receive relatively large subsidies as a share of their revenue compared to firms in other sectors (Figure 1). Furthermore, as is also the case for aluminium smelting and to some extent cement and shipbuilding, BMB account for the largest share in the total support steel firms receive. This stems in part from heavy industries’ reliance on debt for funding their capital-intensive business operations.

Figure 1. The steel sector appears to be a relatively large recipient of subsidies

Note: See ‎Annex B for more information on the firm sample and full sector names. Sector averages use weights based on annual firm revenue.

Source: OECD MAGIC database.

Cash grants and cash transfers

Governments can provide money to a steel firm by distributing grants and awards, either directly or through financial funds or research projects set up for that purpose. A grant can be tied to assets purchased with the grant, or to a specific project (renovation, transformation, technological upgrading, etc.). It can also be given to the company without any conditions attached. Governments can use cash awards to distinguish companies that have excelled in a particular domain. Cost refunds are cash injections provided by the government to compensate for specific costs already paid by the steel company.1

BMB

Governments can also provide loans at below-market rates, either directly or indirectly through a state-controlled bank or other public financial institutions. Preferential loans are difficult to identify as the contractual rates of interest are rarely disclosed to the public. Banks are not willing to share this type of information and can even be forbidden by statutory or regulatory requirements to disclose it. Preferential loans can also be loans that have features that make them particularly attractive for the borrower and would not be obtained at similarly advantageous conditions from private lenders. For example, revolving loans allow the company to avoid repaying their principal at maturity, as those loans are automatically renewed by the lender on maturity.

Because individual loans and loan rates are not disclosed, estimations are carried out on the basis of the average effective interest rates applicable to a steel firm’s debt, calculated using assumptions on its average debt maturity and additional information on debt structure and currency, as well as specifically built interest rate benchmarks (‎Annex C). This approach has the advantage that if preferential treatment is provided even without being disclosed by the lending institution (as there is ample evidence this is surely the case in some jurisdictions), the BMB estimated subsidy amount will still capture the implied subsidy contained in the loan. This estimation methodology results in BMB capturing the part of implied subsidies entail by loans from private banks, to the extent that those loans were provided at lower rates due to the government implicit guarantees propping up a steel firm’s ratings.

Income tax concessions

Other forms of targeted government support are tax exemptions, reductions, and credits, when directed specifically at the steel industry or at specific steel firms. Tax benefits can be given in the form of tax credits, tied to the purchase of some equipment (domestically produced or not), to the completion of a project, or even given to the steel firm as a reward for a specific contribution. In several cases, firms benefit from preferential income tax rates for some of their subsidiaries. Tax concessions may also at times concern the property tax of the factories, export tariffs, and the firm’s contribution to workers’ social security scheme, etc. The data research provided an estimate of corporate income tax concessions for approximately two thirds of the companies in the sample.

Other instrument types

Subsidies, in the sense of government support, are not limited to grants, BMB and tax concessions. There are many more instruments that are regularly used to channel subsidies to steel firms, although many of these instruments are not captured in the recorded and estimated data. Figure 2 below illustrates the main channels, or “instruments”, that can be used to provide subsidies to a steel firm, while Box 1 below illustrates an example of a difficult-to-quantify subsidy, land-use. ‎Annex A provides an indicative OECD matrix of support measures, with general examples.

Figure 2. Government support can be delivered to steel firms through numerous instruments of different incidence

Only the transfers in red, namely cash grants, BMB and corporate income tax concessions (to some extent) are captured in the data used

Source: OECD Secretariat desk research.

Box 1. Below-market-price land transfers in China

In some jurisdictions like China, land ownership is highly centralised, with all land either state-owned (urban areas) or collectively owned by rural communities. Land transfers for industrial use typically involve local governments leasing land to enterprises through long-term land-use rights rather than outright sales. This leasing system is an important tool for local governments for both revenue generation and economic development. Although the law prohibits the use of low land prices to attract investment (State Council, 2004[4]), there is evidence that many local authorities do transfer land at below-market rates to encourage industrial expansion and incentivise investment, thereby providing a significant, yet indirect, form of subsidy to industries (Zhang et al., 2024[5]).

The impact of such subsidies is particularly hard to track because it requires access to local-level transaction data and a clear benchmark of what the market price should be. Moreover, the alignment between local governments and firms - especially state-owned enterprises (SOEs) - further blurs the line between public and private interests. When land is transferred to firms that support the broader strategic goals of the government, such as advancing industrial capacity or technological innovation, it becomes nearly impossible to determine whether the transaction was commercially driven or politically motivated.

There has been some gradual improvement of this situation with the implementation of market-oriented land reforms at the end of 2006 (The PRC’s Government, 2006[6]). These reforms aimed to reduce the preferential treatment historically given to state-owned enterprises (SOEs) in acquiring industrial land at below-market prices. Although some research points out that the ability of SOEs to purchase land at lower prices and in prime locations has diminished over time due to more transparent methods like tenders, auctions, and listings (Zhang et al., 2024[5]), the issue persists, with evidence suggesting that preferential treatment for SOEs in land transactions still exists in certain regions (The Paper, 2019[7]).

How can econometric analysis account for all hidden instruments implied in Figure 2? A possible solution to alleviate concerns related to missing subsidy variables is to recall that in econometric analysis, any given variable can represent not only the obvious quantity it reports, but also serve as a signalling device for the presence of other, unobserved components of subsidisation. In that sense, a cash amount may represent not only the verifiable money amount that a steel firm receives in a given year, but also the amount of other types of support it probably received in that year. Hidden variables such as the political connectedness of the firm undoubtedly influence cash grants, BMB, tax concessions, and other types of subsidies not captured in our data.

To informally “test” for this hypothesis, the conditional probabilities for a steel firm, in a given year to receive BMB knowing it has received a grant in that year, and vice-versa, are presented below (Table 1). Steel firms that are recipients of a cash grant in a given year have an 80% chance of also benefitting from BMB during that given year, and steel firms that benefit from BMB in a given year have a 72% chance of also being recipients of a cash grant on that year. Correlations are strong for partner economies, but less so for OECD Member countries, where BMB is provided much less often, even when grants are provided. This can have consequences when discussing econometric regression results (Box 2).

Table 1. Conditional probabilities show that subsidy instruments are often used in conjunction

	All sample	OECD Member countries observations	partner economies
Total number of observations in the sample	742	204	538
Total number of observations with non-missing grant field	660	178	482
Total number of observations with non-missing BMB field	659	178	482
Number of observations associated with a non-zero cash grant amount	444	110	334
Number of observations associated with a non-zero BMB	403	47	356
Number of observations associated with neither a missing cash grant field nor a missing BMB amount	626	170	456
Number of observations associated with both a non-zero cash grant amount and a non-zero BMB	305	28	277
Conditional probability for a firm in the sample to receive a cash grant knowing it received BMB (on a given year)	(305/626)/(403/659) = 80%	(28/170)/(47/178) = 62%	(277/456)/(356/482) = 82%
Conditional probability for a firm in the sample to receive BMB knowing it received cash grants (on a given year)	(305/626)/(444/660) = 72%	(28/170)/(110/178) = 26%	(277/456)/(334/482) = 87%

Source: OECD MAGIC database

If the correlations highlighted in Table 3 between the use of grants and BMB can be extrapolated to all other instruments of government support, then arguably cash grants can be used as a proxy, or signalling device, for the total subsidisation benefiting steel firms, through all available instruments including the “hidden” channels depicted in Figure 2. This has implications when discussing impact results (Box 2), but may also help justify the validity of performing robustness checks for later capacity regressions by using state ownership as an instrumental variable (IV).2

Box 2. Interpretation of coefficients in an econometric study for proxy variables

Under the assumption – not proven but suggested by Table 3– that in partner economies a high degree of correlation between all instruments exists, then cash grants, the most transparent and easy-to-capture instrument, can be used as a proxy variable for other, more “hidden” type of subsidies.

This has consequences on the interpretation of the estimated coefficient for cash grants. For example, the econometric analysis in Section 4 shows that in partner economies, USD 1 million in additional grants is correlated to an increase of a steel firm’s capacity by about 7 500 to 9 500 metric tonnes in the following years (controlling for various other factors including capacity development trends). Interpretation would not be that the cash grant subsidies “by its own” increased capacity to said extent, but that it revealed a “subsidy mix” that resulted in said capacity expansion. Grants are a “signalling device”, but just the “top of the iceberg”, and the use of only grants (with no other instruments) would likely have resulted in lower, or no capacity increase (as in the case of OECD Member countries).

Subsidies have varied sharply over time

Subsidies provided to steel firms, be they in the form of grants, income-tax concessions or BMB, have varied significantly over time. BMB, but also other types of subsidies, tend to increase sharply at times of crises, as shown in Figure 3 below.

Figure 3. Steel subsidisation, as a percentage of total firm revenue

Source: OECD MAGIC database

The two last major steel crises were the 2009 global financial crisis and the 2015-2016 economic slowdown in China. Both crises were characterised by sharp declines in global and domestic demand, leaving steel companies in financial distress. In both instances. BMB were extensively used to channel subsidies to steel firms. Although grants were also used, BMB are quicker to deploy as lending institutions usually already have established lending relationships with concerned steel firms, veiled interest in avoiding their bankruptcy, and do not require any specific framework to provide BMB to their client steel firms. This was especially the case in China (Box 3).

Box 3. The strong counter-cyclical use of BMB in steel business cycles

The Chinese Government used BMB as a key instrument to support Chinese steel firms, allowing them to access financing at favourable rates, thereby keeping production lines running and avoiding widespread bankruptcies.

During the 2009 global financial crisis, as demand for steel plummeted both internationally and domestically, China launched a 4 trillion RMB stimulus package focused on infrastructure investment (The Central People’s Government of the PRC, 2008[8]). This stimulus significantly increased demand for steel, and BMB was extended to firms to help them weather the downturn (The Central People’s Government of the PRC, 2012[9]). The availability of cheaper loans allowed steel firms to continue operating despite the crisis, supporting employment, steel production, and broader economic activity (The State Council of the People’s Republic of China, 2009[10]).

Similarly, in 2015-2016, China’s steel industry faced another crisis due to excess capacity, a slowing housing market, and a stock market crash. The slowdown in housing, a critical driver of steel demand, led to excess inventory, particularly as millions of empty apartments were left unsold, exacerbating the challenges for steel producers (The People’s Bank of China, 2015[11]). Additionally, the stock market collapse further strained financial stability, with the Shanghai Composite Index losing nearly half its value by early 2016 (Shanghai Bureau of Statistics, 2022[12]). The capital outflows during this period, as a result of fears over exchange rate shifts, added pressure on the broader economy.

In response, the government once again turned to BMB as a key element of its broader financial strategy, providing liquidity to struggling steel firms. The availability of BMB may have allowed Chinese steel firms to continue operating even at low-capacity utilisation rates, as domestic demand sharply declined. This led to a surplus in production, which was offloaded through exports, peaking in 2015 (Figure 4). The surge in exports, supported indirectly by BMB contributed to global oversupply, depressed international steel prices, and increased trade tensions. Going forward, there is growing concern that the slowing economy and the continued downturn in Chinese real estate could lead to a resurgence of Chinese BMB or similar targeted borrowing programmes in 2023-2024, as authorities seek to provide support through more favourable financing options (Reuters, 2024[13]).

Figure 4. China’s steel export rose sharply during the 2015 steel crisis

Box 4. The Case of Ansteel Group

Angang (Ansteel Group) is one of China’s most important SOEs and a key player in the steel industry, often called the "Eldest Son of China's Steel Industry." Founded as the first major steel complex, it has been central to the country's economic development. Angang has a production capacity of 53 million tonnes of iron and 64 million tonnes of steel in 2024, with operations spanning iron, vanadium, and titanium production (Ansteel, 2024[14]).

During the steel crisis that began around 2014, Angang faced significant challenges that reflected the broader difficulties of China's steel industry. The sector, once a symbol of prosperity, fell into what many termed a "steel ice age" as overcapacity collided with weakening global demand and declining prices. Angang, as one of China’s leading steel enterprises, experienced a dramatic fall from being a profit-generating powerhouse to recording major losses. In 2015, Angang’s listed subsidiary reported a record loss of 45.9 billion yuan, marking the largest deficit since its listing, as the company grappled with the harshest competition and market conditions in decades (China Youth Daily, 2016[15]).

In the midst of Angang's declining profitability, as evidenced by the sharp decrease in EBITDA between 2014 and 2016, the government stepped in to provide increasing below-market loans, similar to what it did for other SOE steel firms. Figure 5 provides the estimated BMB for Chinese SOEs, their capacity utilisation rates and EBITDA trends. BMB acted as a form of financial support despite Angang's clear operational struggles, with capacity utilisation rates (CUR) consistently below optimal levels. While government support generally helps maintain high CUR rates, Angang’s CUR remained low due to a significant 3.2 Mt increase in steelmaking capacity between 2014 and 2015. This expansion, despite the financial backing, contributed to operational inefficiencies that hindered the firm from reaching optimal utilisation levels.

The continued inflow of BMB suggests that the government's priority was to maintain Chinese SOEs strategic presence in the steel industry, even when market conditions and profitability were under significant strain. The loans helped Chinese SOEs sustain their operations while increasing capacity, despite underutilisation, showcasing how state support in the form of BMB loans can prolong operations even in difficult times.

Figure 5. The Chinese government provided ample counter-cyclical support in the form of BMB to its state-owned steel firms

Note: EBITDA: Earnings Before Interest, Taxes, Depreciation, and Amortisation. It measures a company's operational profitability by focusing on earnings from core business activities, excluding financial and accounting factors. CUR: Capacity Utilisation Rate. This represents the percentage of a company's total production capacity that is being used.

Source: OECD MAGIC database

Subsidies are unequally used across jurisdictions

Subsidies are unequally used across jurisdictions. The pervasiveness of subsidies can be compared across jurisdictions of different sizes by scaling subsidies provided to each firm by some variable capturing the size of the firm, be it in terms of capacity, asset or revenue, thus providing metrics of “subsidy intensity”. Data shows that some jurisdictions tend to subsidise their steel sector more intensively than others, a phenomenon largely depending on the national contexts, including the establishment of “goals” related to steel production. Those targets are very often formulated without taking into account specific market conditions and global steel market excess capacity, and are often set only once for a number of years, typically 4 or 5 years (Mercier and Giua, 2023[3]). Figure 6 below illustrates that subsidy intensities are significantly higher in China compared with other partner economies in the sample. China’s subsidy intensities are five times greater than those of other partner economies, and 10 times higher than those of OECD Member countries. Meanwhile, steel firms in partner economies, excluding China, benefit from subsidy intensities that are double those of steel firms in OECD Member countries.

Figure 6. Subsidies intensities are multiple times higher in China than in other partner economies

Source: OECD MAGIC database

Subsidies are unequally distributed across firms

Subsidies are also provided to steel firms to different extents. Figure 7 below shows the distribution of the subsidy intensities for both grants and BMB through its median and quartiles. Distributions happen to be very fat tailed, with their lower quartile receiving multiple times lower levels of subsidies compared to asset size than the higher quartiles. This is evidence of the wide disparity in the provision of subsidies to steel firms across jurisdictions.

Figure 7. Distribution of subsidies intensities shows a wide dispersions

 

Note: The dotted lines provide information on the distribution (first and third quartiles) of the represented values across the firms in the sample: 25% of the companies have a value below (respectively, above) the first (respectively, third) quartile line. The continuous line provides information on median values for the firms in the sample: those lines divide the distribution into two halves with 50% of the companies benefiting from more subsidies per asset than the level indicated by the median line, and 50% receiving less than the level indicated by the median line.

Source: OECD MAGIC database.

Some firms’ characteristics, such as their home jurisdiction, their size or their SOE status, seem to significantly impact their receiving subsidies. It is econometrically better to use these percentage of state ownership in analyses and regressions, rather than a clear-cut “state owned or not state-owned” binary variable, as it more adequately takes into account the heterogeneity of steel firms’ situation regarding their state ownership and degree of government involvement. For example, Figure 8 below shows that the amount of government ownership drastically correlates with the amount of subsidies that a given steel firm receives relative to its size:

Figure 8. SOE tend to benefit more from subsidies than other firms compared to their total revenue

Source: OECD MAGIC database

This shows that the level of government ownership of steel firms in China drastically impacts the amount of subsidies they receive. The government's ambition to create "National Champions" favours SOEs, granting them greater access to credit, technology, and markets, leading to a distorted playing field.

Notes

← 1.  For example, cost refunds can cover a given percentage of a steel company’s investment in research and development.

← 2.  For government ownership share to be a proper IV variable, it needs to satisfy two conditions. The first is easily checked with a simple regression: government ownership share is very much correlated with total amounts of cash grants received. The second condition can never be statistically checked so requires a strong economic justification: government ownership must not impact capacity once cash grants amounts are controlled for. Because there are many ways a government can provide support to expand capacity through many other instruments, such as the one depicted in Figure 2 as well as political connectedness, etc., one could argue this is not the case. But, since the amount for grants received can reasonably be used for a proxy or signalling device for the total amount of support provided, we do believe using government ownership share is a proper IV (although the sample will probably have to be split to account for the breaking of correlations among instruments in OECD Member countries).