The breakneck growth rate of the Chinese economy is in large part driven by capital accumulation (and exports). As the diagram shows, the country’s investment to GDP ratio has been high and rising in recent years, exceeding 40% of GDP in 2005. This is much higher than even the high rates observed in East Asia before the 1997 Crisis.
An investment rate this high raises concerns.
Is it really possible to invest wisely four out of every ten RMB earned? This concern is especially relevant when it comes to investment by state-owned enterprises. If the investment is not efficient, then the same output could be achieved with less capital, thus freeing resources for other uses, such as raising household consumption or improving profitability and/or the balance sheets of the banks that fund them.
There are a number of a priori reasons for believing that state-owned enterprises may be investing less efficiently than domestic private firms. They are burdened with more administrative interference in terms of restrictions on hiring and firing and on switching product lines in response to changing market conditions. Despite some progress over the years in linking executive pay to performance, managers in these firms often still do not have compensation schemes that encourage efficiency, or discourage overinvestment and “empire building.” Some state-owned enterprises also have weak corporate governance.
The Chinese financial system is dominated by majority or wholly state-owned banks as the figure shows. Such banks may favour lending to state-owned enterprises despite efforts by the authorities to increase the commercial orientation of these banks. For example, while state-owned firms represent a declining share of output (it was about a third in 2005), their borrowing from domestic banks accounts for more than half of the total lending by these banks. Majority state-owned firms also take up the lion’s share of all publicly-traded companies in China’s two stock exchanges. Some of the bias may be related to the smaller size and higher risk of some private firms. But it is common to hear private firms complaining about the difficulty they face in securing funding for both short-term working capital and long-term investment needs even when they have comparable size and risk profiles as their state-owned peers.
Currently, approximately half of China’s investment is financed out of the corporate sector’s retained earnings due to favourable factors such as high world prices for commodity producers. The reliance on bank lending by state-owned enterprises is correspondingly lower than it otherwise would be the case. The high level of retained earnings in such firms may reflect an incentive structure that leads managers to prefer over-investing to paying dividends to the state.
To see whether state-owned enterprises are systematically less efficient investors, David Dollar and I analysed a data set based on a 2005 survey of 12,400 firms in 120 cities located throughout China. What we found was that ownership did indeed have an impact on the productivity of investments even after more than two decades of corporate reforms. As the chart shows, state-owned firms had the lowest marginal product of capital as measured by the marginal revenue product of capital.
For every firm in a given sector and location, the study computed the marginal revenue product of capital (MRPK) as value added minus payments to labour, divided by the stock of capital. This measure was then subject to statistical analysis aimed at determining the impact of ownership on capital productivity, controlling for other determinants.
Specifically, firm-level MRPK was regressed on a set of indicator variables representing sector-time pairs and locations as well as a set of indicator variables representing firm ownership. The sector-year indicators capture the possibility that demand or supply shocks in a given sector/year could cause MRPKs in a particular sector-year to be different from others. The ownership indicators measure the MRPKs of various ownership groups relative to domestic private firms. These ownership groups are defined in a way that are mutually exclusive: wholly state-owned, majority state-owned, minority state-owned, wholly foreign-owned, majority foreign-owned (with no state shares), minority foreign-owned (with no state shares), and collectively owned.
According to standard theory, profit-maximising managers should equate a firm’s MRPK to the sum of the market interest rate, depreciation rate, and distortions in the capital market that the firm faces. Again according to the standard theory, capital is efficiently allocated if MRPKs are equalised across all firms, regardless of sector, location or ownership.
As the chart shows, this is clearly not the case in China, according to our data. The difference in the MRPKs between two firms in the same sector reflects mostly the difference in the cost of capital. For example, if state-owned enterprises receive more favourable treatment than domestic private firms in borrowing from banks or in obtaining government approvals to be listed in the domestic stock market, then the MRPKs for state-owned enterprises would tend to be lower than those of private firms on average. Using this framework, the study assesses three types of inefficiency, or biases, in capital allocation at the level of ownership, location, and sector, respectively.
One key result is that wholly and majority state-owned firms are found to have lower marginal returns to capital than private or foreign firms. The median MRPK for private firms is 63%. In contrast, the median values for wholly and majority state-owned firms are 37% and 52 %, respectively. These numbers all appear large because they are computed before tax and depreciation, and reflect all other distortions to the cost of capital. The key point is that the returns to capital are not equalised; state-owned enterprises have substantially lower returns than private firms. (The study controls for a number of obvious sources of bias such as over-representation of state-owned firms in sectors or locations that happened to have lower returns due to temporary factors that are not related to ownership issue per se; with these corrections, the results still suggest that private firms have to forego projects of significantly higher returns - on the order of 11-54 percentage points, depending on the specifications - than their state-owned counterparts due to differential ability to meet their financing needs.)
There is also a significant locational bias in returns to capital. Coastal areas, especially the Yantze River Delta region (Shanghai, Jiangsu and Zhejiang) and Bohai Circle (encompassing Beijing and Tianjin) have higher returns to capital than Northwest and Southwest regions. Agglomeration economies are a possible interpretation though not the only one. We also find weaker evidence of differential returns to capital at the sector level, though not as significant economically and statistically as the other biases.
To get a handle on the magnitude of the cost of this inefficient financial allocation, we did a simple calculation based on the following thought-experiment. Consider a transfer of X% of capital currently employed by the state sector to the private firms, but leave the labour (and other inputs) allocation fixed where it is. Since the MRPK measures the impact on output of a marginal change in the amount of capital employed, we can use our MRPK figures to work out the notional rise in output that would occur with a more efficient allocation of capital. In particular, the amount of capital to be transferred is chosen in such a way that the MRPKs would be equalised across ownership groups.
How much capital would have to be transferred? How much gain in aggregate value added can be achieved by the change? The answers depend on the estimated current gap in the MRPKs, the form of the production functions, and some other parameters such as the capital-labour elasticity of substitution. In the benchmark case reported in our paper, two-thirds of the capital currently employed by state-owned enterprises would have to be transferred to the private sector to equalise our measure of capital productivity (MRPKs). Doing so would raise GDP by 5%, according to our calculations.
To put it differently, with a more efficient use of capital, the country could reduce its very high investment rate substantially (on the order of 8% of its capital stock) with no adverse effect on its growth rate. Such an improvement in investment efficiency could lead to a faster rise in household consumption and living standards.
These back-of-the-envelop calculations suggest that there is much to be gained from improving China’s capital allocation. How could the efficiency of capital usage be improved?
Besides curbing the amount of investment in the less efficient state-owned firms, further reforms of the incentives faced by the managers of such enterprises, including privatisation (so that they will behave like their counterparts in profit-maximising private firms) would raise efficiency. China’s accession to the World Trade Organization has given rise to significant financial sector liberalisation (in addition to greater trade openness). The scope of the liberalisation is somewhat constrained by various regulations such as a relatively large minimum capital requirement for banks and their branches and a need for foreign banks to be incorporated locally in order to accept retail RMB deposits. By increasing competition from foreign-owned financial institutions, financial sector liberalisation may steadily prod domestic banks to rid themselves of biases in capital allocation, especially if the government also moves away from a promise of unconditional bailouts of failed banks. On the other hand, if state-owned banks are slow to act, and their best customers migrate to foreign banks, then the possibility of drastically increased non-performing loans will loom large again, and the risk of a financial crisis will elevate as well.
“Das (Wasted) Kapital: Firm Ownership and Investment Efficiency in China“, David Dollar and Shang-Jin Wei, January 2007