A. A Brief Overview of SOEs in China

1. State-owned enterprises command a large share of the Chinese economy. In 2018 (latest data available), total assets of Chinese SOEs stood at 194 percent of GDP—higher than in the early 2000s, and several orders of magnitude larger than in any other country (IMF Fiscal Monitor). China’s SOEs also operate in all sectors of the economy, while in other countries their operations are usually concentrated in a few sectors (mostly transport, utilities, and finance). After consecutive waves of reforms, the number of SOEs in China has declined significantly since the 1990s (a ⅔ decline among the industrial firms alone), along with the share of urban workers in SOEs. Some of the remaining SOEs have grown into industry leaders, with many now counting among the world’s largest firms (IMF Fiscal Monitor). However, despite improvements in recent years, SOEs continue to underperform compared to private firms. As of end-2019, industrial SOEs (for which the data is readily available) remained less profitable than private firms, and a higher share of SOEs were loss making. At the same time, SOEs continued to receive a higher share of bank financing and enjoy lower interest rates on their liabilities, likely due to the existence of implicit government guarantees (Lam et al., 2017; Gatley, 2018; Zhang and Wu, 2019).

View Full Size

Total assets of the SOE sector

(Percent of GDP)

Citation: IMF Staff Country Reports 2021, 012; 10.5089/9781513566528.002.A002

Sources: CEIC, China Statistical Yearbook, OECDNote: Data for 2015 (latest available) for countries other than China.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Composition of the SOE sector in China

Citation: IMF Staff Country Reports 2021, 012; 10.5089/9781513566528.002.A002

Sources: OECD; and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

2. SOEs provide many social and other non-economic functions, complicating the reform process. Traditionally, the presence of SOEs has been justified by the need to correct market failures that prevent efficient provision of services to the population or production of essential goods or industrial inputs (OECD, 2018). SOEs have also been used by governments to implement industrial policy, stabilize employment, or to protect national security (OECD, 2018). Some of these factors are at work in China, where SOEs have played a role in supporting the economy and employment during recessions (including during the latest COVID-19 crisis), and provide health and pension services to the population (IMF Fiscal Monitor). If SOEs are not able to properly recoup costs of the non-economic functions they are asked to fulfill, their performance and profitability may suffer. A comprehensive approach to reform efforts is therefore necessary, encompassing the broad arrays of roles SOEs are asked to fulfill. For example, while SOEs are often used to maintain employment stability, improving the social security system would protect workers over jobs while reducing the burden on SOEs.

B. Productivity Gaps Between Listed SOEs and POEs

3. We use the Wind database of listed firms to analyze the relative performance of listed SOEs and POEs from 2002 to 2019.² The database covers over 3700 listed firms in the Shenzhen and Shanghai stock exchanges between 2002 and 2019. The main benefits of using Wind are the recent time span covered by the database and the broad coverage of sectors.³ We define SOEs based on the identity of the major shareholder of the firm, and separately define ‘central’ and ‘local’ SOEs as those owned by the central and local government, respectively. While listed firms are only a small subset of all registered firms, they account for a substantial share of Chinese economic activity: 6 percent of GDP and 10 percent of manufacturing value-added in 2019. Among listed firms, SOEs account for a large share of output in all industries, though this share has declined over time. In 2019, SOEs accounted for 29 percent of listed firms and 57 percent of listed firm value-added, with considerable heterogeneity across sectors. SOEs tend to be much larger than POEs, with the typical SOE employing more than twice as many workers as the typical POE (for more details see Jurzyk and Ruane, 2020).

4. We construct revenue productivity measures to compare the performance of listed SOEs and POEs and infer the extent of resource misallocation. We define revenue productivity as the average product of capital and labor (that is, value-added per unit of capital and labor). Revenue productivity varies dramatically across listed firms even within the same two-digit sector and year; the revenue productivity of firms at the 90^th percentile is more than four times larger than that of firms at the 10^th percentile. These average revenue productivity differences do not necessarily reflect differences in the technology of firms or the quality of their products. Rather, they could reflect differences in factor prices faced by firms (e.g. interest rates), differences in markups, or other taxes and subsidies. Such permanent differences imply a misallocation of resources across firms which lowers aggregate productivity.⁴ Reforms which reallocate resources more efficiently, closing the gaps in marginal products across firms, could therefore provide a substantial boost to Total Factor Productivity (TFP) growth.

5. We find that SOEs have significantly lower revenue productivity than POEs in the same sector, and that these productivity gaps remain substantial over time.

• There is a lot dispersion in the revenue productivity of both SOEs and POEs, with significant overlap between their distributions (figure). Many SOEs are clearly profitable and productive, even relative to POEs.

• However, there is a large statistically significant gap between the average productivity of SOEs and POEs, which widened to 30 percent during the GFC, though it has reduced somewhat to 20 percent as of 2019.

• These revenue productivity gaps are of similar magnitudes for central and local SOEs, suggesting that they are not driven by specific demands placed on central SOEs by the central government.

• They are also pervasive in almost every sector (figure). They are particularly large in sectors such as Utilities, Transportation and Steel Manufacturing, though smaller in more high-tech sectors such as IT Services, Manufacturing of Pharmaceuticals, and Medical Equipment. In addition, sectors with larger revenue productivity gaps are also those where SOEs account for a larger share of output and inputs, amplifying the distortionary impact of these gaps.

• There is also evidence that the low productivity of SOEs reflects a particularly low average product of capital: value-added per unit of fixed assets is almost 40 percent lower for SOEs than POEs, while value-added per unit of labor of SOEs is similar to that of POEs (for more details see Jurzyk and Ruane, 2020). The low revenue productivity of SOEs is therefore explained to a large extent by an inefficiently high capital intensity of SOEs.

View Full Size

SOE and POE Revenue Productivity Distributions

(Share of firms)

Citation: IMF Staff Country Reports 2021, 012; 10.5089/9781513566528.002.A002

Sources: Wind and authors’ calculations. See Jurzyk and Ruane (2020) for more details.Note: Revenue productivity defined as value-added divided by a geometric average of capital and labor.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

SOE/POE Productivity Gaps by Sector

(productivity gap, POE = 1)

Citation: IMF Staff Country Reports 2021, 012; 10.5089/9781513566528.002.A002

Source: Wind, authors’ calculations. See Jurzykand Ruane(2Q20) for more details.

• Download Figure
• Download figure as PowerPoint slide

6. A quantitative model of misallocation suggests that closing SOE-POE revenue productivity gaps could increase the productivity of listed firms by around 6 percent. Large revenue productivity gaps between SOEs and POEs suggest that resource misallocation is an important drag on aggregate productivity. We evaluate these gains using a quantitative macroeconomic model of resource misallocation (Hsieh & Klenow, 2009). While we only have data for listed firms, these are the largest firms in the Chinese economy and therefore the most macro-important. We find that a policy which reduces the average SOE-POE revenue productivity gap in every sector could increase aggregate productivity among listed firms by between 5 and 6 percent.⁵ A more ambitious policy aimed at reducing both the average productivity gap and the distorted capital-intensity of SOEs could yield gains of over 6 percent. This suggests that measures which equalize the playing field between SOEs and POEs, in particular equalizing the effective rental rate of capital, is an important potential source of growth.

View Full Size

SOE-POE Productivity Gaps and Reform Gains

(LHS, productivity gap_r POE – 1_r RHS, in percent)

Citation: IMF Staff Country Reports 2021, 012; 10.5089/9781513566528.002.A002

Source: Wind, authors’ calculations. See Jurzykand Ruane (2020) for more details.

• Download Figure
• Download figure as PowerPoint slide

7. Extrapolating our results to the broader Chinese economy, our findings suggest that SOE reforms could double the rate of TFP growth for five years. The existing literature also finds large capital productivity gaps between SOEs and POEs for a much larger sample of industrial firms before 2012 (see Hsieh and Klenow (2009), Hsieh and Song (2015) or Wu (2018)). In addition, David and Venkateswaran (2019) document similar patterns of capital productivity dispersion between listed Chinese firms and a larger sample of industrial firms.⁶ We therefore assume that our estimates of SOE-POE misallocation for listed firms apply also to non-listed firms, though we scale them down to around 4 percent based on the fact that the SOE share of assets for the whole economy is smaller than for listed firms.⁷ Our findings of gains from reallocation in the order of 4 percent are similar to Hsieh and Klenow (2009), who find 5.3 percent gains on average between 1998 and 2005 from reducing the average revenue productivity gap between state and private firms.⁸ These gains are somewhat smaller than the 10 percent gains Brandt et al. (2013) estimate for state vs. non-state capital reallocation, however they account for between-sector capital misallocation while we restrict our attention to within-sector misallocation. Factoring in such across sector differences in SOE intensity would imply larger gains. Given that aggregate TFP growth has averaged 0.6 percent between 2012 and 2017 (Penn World Tables), our results suggest that SOE reform could more than double the rate of TFP growth for five years—or likely more, if sectoral reallocation would be considered as well.

C. The Role of Credit Misallocation⁹

8. Firms with low revenue productivity have higher leverage, suggesting an important role for credit misallocation. We find that there is a negative correlation between firm productivity and the leverage ratio (measured by the debt-to-asset ratio), suggesting that credit is allocated to the least efficient firms in the economy. This likely reflects the distortions from the implicit guarantees that make SOEs more credit-worthy (IMF Country Report No.19/274). Indeed, the data show that on average SOEs have higher leverage ratios than POEs (figure). Moreover, although both low-productivity SOEs and low-productivity POEs have deleveraged since 2016, the former remain the most leveraged. In fact, the leverage ratio of low-productivity SOEs is still more than 5 percentage points higher than that of high-productivity POEs, indicating that credit could be better channeled to more productive firms. In addition, credit misallocation seems to also exist within the SOE universe. The leverage ratio of high-productivity SOEs was 10 percentage points lower than the low-productivity SOEs at the end of 2016, although the gap has somewhat narrowed since then.

View Full Size

Less profitable SOE have higher leverage

(In percent)

Citation: IMF Staff Country Reports 2021, 012; 10.5089/9781513566528.002.A002

Sources: Wind and IMF calculations

• Download Figure
• Download figure as PowerPoint slide

9. A reallocation of credit from highly-leveraged SOEs to POEs could increase aggregate investment and productivity. Using the threshold effect model (Hansen, 2002), we find that there is a non-linear correlation between leverage and investment. In particular, the impact of a 1-ppt increase in a firm’s leverage ratio on investment is 0.1 ppt but falls dramatically to 0.01 ppt if the leverage ratio exceeds a threshold of about 35 percent. Both SOEs and POEs have similar leverage thresholds and the impact of leverage change on investment is of similar magnitudes. As a higher share of SOEs are highly indebted than POEs (figure), there is a clear benefit from deleveraging SOEs without having a large impact on investment. The resources or credit from such SOEs can then be allocated to the less-indebted POEs to boost productive investment. Simulations find that reallocating credit from highly-leveraged SOEs to POEs can increase aggregate investment as more credit is allocated to the generally more productive POEs, thereby boosting overall productivity and enhancing growth. Growth would increase by 0.3–0.4 ppts annually when highly-indebted SOEs deleverage by 2 ppts and the freed-up credit is channeled to POEs.

View Full Size

Share of highly-indebted firms by ownership

(Share of high-indebted firms)

Citation: IMF Staff Country Reports 2021, 012; 10.5089/9781513566528.002.A002

Source: IMF staff calculations.Note: Highly-indebted firms are defined as those with debt-to-asset ratio above the estimated threshold of about 35 percent.

• Download Figure
• Download figure as PowerPoint slide

D. Policies to Reform Chinese SOEs

10. Continue the process of identifying non-viable SOEs and opening non-strategic sectors to private/foreign competition to improve market competition among firms, supported by reform of the social safety net to relieve SOEs of social functions. Non-viable firms should be allowed to default and exit the market rather than being merged with more profitable/efficient SOEs. That would protect good performers and enhance market competition. Low productivity of SOEs in key sectors can create bottlenecks in the supply-chain, especially if they are in upstream sectors supplying intermediate inputs. Such distortionary bottlenecks can have large impacts on aggregate productivity (Jones, 2011). Reform of the social safety net should proceed in parallel to relieve SOEs of their role to stabilize employment and provide social security benefits for workers, transferring these obligations to the state. That would also allow more productive private firms to hire workers. This will be particularly important in light of recent hukou reforms, which could lead to a significant increase in demand for jobs in urban areas.

11. Ensure equal access to credit and capital by private firms and allow SOEs to deleverage. This would require recognizing and removing the implicit government guarantees that allow SOEs to access financing from banks and financial markets at lower rates. To ensure that the financial sector is prepared for the removal of implicit guarantees, banks could be required to carry higher risk weights on SOE loans, build liquidity buffers, reduce reliance on short-term funding, and increase capital. As highly leveraged firms invest less, deleveraging highly indebted SOEs could mitigate the impact on investment while improving credit allocation. Moreover, concerted efforts are needed to ensure that market-based policies are in place to allow credit to flow to its most productive use. Previous papers have discussed policy options including establishing competitive neutrality among firms (Jahan and Kang, 2019). Strengthening the credit culture would also help improve lending decisions, with reforms targeted to improve credit ratings, strengthening credit registries, ensuring adequate capitalization of banks and promoting more risk-based vs. collateral-based lending (Jahan et al., 2019).

12. Improve SOE governance. Most non-financial SOEs operate as large business groups organized under one parent holding company owned directly by the State-Owned Assets Supervision and Administration Commission (SASAC). While current policies prevent officials from holding part-time roles on corporate boards, there is a tradition of annual exchange of management staff between SASAC and the central SOEs, and some of the top managers of national SOEs are given seats in important party bodies (Milhaupt, 2019), thus blurring the distinction between the company and its supervisors. Corporate boards—a standard feature elsewhere and recommended by the 2013 Third Party Plenum—continue to be missing in many mid-sized and smaller SOEs. It is therefore important to allow for the appointment of company managers with international/private sector experience, to increase the transparency of SOE group structures and activities, and to clarify the role of the Party in decision making.