Information about Asia and the Pacific Asia y el Pacífico

Chapter 14 Decentralized Socialism and Macroeconomic Stability: Lessons from China in the 1980s

Manuel Guitián, and Robert Mundell
Published Date:
June 1996
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details
Gang Fan, Wen Hai and Wing Thye Woo 

The economic reforms that China has implemented since 1978 have three dimensions: decentralization of the state sector, diversification of ownership patterns,1 and China’s opening to international trade and investment. The result has been an impressive acceleration of growth in the 1980s.2 National income grew at an annual average rate of 7.0 percent during 1975–79, 8.6 percent during 1980–84, and 9.2 percent during 1985–89 (see Table 1).

Table 1.China’s Economic Performance in the 1980s(In percent)
Real national income growth7.
Real GNP growth7.94.48.710.314.612.78.311.
Retail price increase1.
Current account balance (percent of GNP)−4.0−2.60.1−1.0−1.01.9
M2 growth24.119.713.119.242.417.129.324.221.018.426.2
Budget deficit (percent of GNP)
Sources: China Statistical Yearbook (various issues); and IMF staff estimates.
Sources: China Statistical Yearbook (various issues); and IMF staff estimates.

The higher growth rates came at the price of greater macroeconomic instability, however. The annual average inflation rate in 1985–89 was 12 percent compared with 4 percent in 1980–84 and 1 percent in 1975–79. In response to the fast growth and high inflation, China’s demand for imports soared, turning the current account surplus from 1.1 percent of GNP in 1980–84 to -1.7 percent in 1985–89. The Government responded to the two disconcerting trends of worsening inflation and deteriorating external position by reducing credit expansion and budget deficits, although it reversed policies whenever growth slowed down. The outcome was a rather volatile pattern of output expansion.

The classic example of this stop-go cycle of macroeconomic management is the 1984–86 episode. The money supply (M2) grew by 42 percent in 1984, and prices began rising quickly in 1985. The Government reacted by slowing money growth to 17 percent in 1985 and reducing the budget deficit to 0.5 percent of GNP from 1.5 percent in 1984. The inflation rate dropped but it was accompanied by an economic slowdown. The growth rate in 1986 was only 8 percent compared with 13 percent in 1984 and 1985. The Government then relented on its tight macroeconomic policies, allowing the money supply to grow by 29 percent and the budget deficit to widen to 2 percent of GNP. This loosening restored growth to 10.2 percent in 1987 but also raised inflation to 7.3 percent, thus laying the groundwork for the policy-induced economic crunches in 1989 and 1990.

We will argue in this paper that the observed stop-go phases of the economy were the result neither of a deliberate approach to macroeconomic management nor of incompetent macroeconomic management. Our explanation is that the institutional mechanisms within the economic system created the “go” phases and that the “stop” phases occurred because these institutional mechanisms made it difficult for contractionary economic policies to have a “soft touch.” Since 1979, China has faced a dilemma, aptly paraphrased by Chinese scholars and policymakers as follows: the economy leaps into chaos when macroeconomic control is loosened and plunges into a deep recession when macroeconomic control is tightened.3

We will trace the origin of this dilemma, along with the rise in inflation since 1978 and its acceleration after 1984, to the decentralization reforms in the state sector. Our case is based on data collected in a survey of 300 large and medium-sized state enterprises. These firms were asked to provide information on their activities in 1980 (as a reference point) and for every year during 1984–89.

This paper focuses on the macroeconomic implications of the survey, which was also used by Woo, Hai, Jin, and Fan (1994) to assess microeconomic efficiency during the period. The empirical approach adopted in this paper is based on the rationale of the representative agent, which dominates theoretical modeling in macroeconomics. The representative agent procedure permits us to rely mainly on the sample means of the different variables to support our macroeconomic analysis. We also used this procedure to study total factor productivity growth, and we confirmed the negative result found in Woo, Hai, Jin, and Fan (1994), which used the traditional microeconometric method of panel estimation.

In the paper, an economy in which most of the capital stock is owned by the state is called a public ownership economy. We will refer to nonfinancial state enterprises as state-owned enterprises (SOEs) and to financial state enterprises as state-owned banks (SOBs).

Public Ownership and Labor Compensation

In a pure public ownership economy, the national income (Y) is distributed as personal income (W, wages and other payments to individuals) and state revenue (T, taxes, profit remittances, and the retained profits of the state enterprises).4 Because of the institutional absence of a capital market and the presence of a state-run social security program, capital accumulation (I, investment) is the sole responsibility of the state. For simplicity, we will limit our theoretical discussion to the case where government consumption is small and the state budget is balanced; that is, investment (I) equals state revenue (T), and consumption (C) equals personal income (W).

One key characteristic of all existing forms of public ownership economy is that they actively seek to minimize differences in the personal income of workers in different firms. The result is that the effective rate of taxation and the absolute amount of taxation differ vastly across firms.5

Suppose that the state, acting as the agent of public ownership, solves the social welfare maximization problem and finds the optimal I to be I* and the optimal C to be C*. We define

We recognize three reasons why managers and workers in a public ownership economy will prefer an investment-consumption ratio dc that is smaller than d*. The first reason is that these individuals see investments as “public goods” from which everyone benefits but to which no one likes to contribute. This “free-rider” aspect of capital accumulation is peculiar to a public ownership economy. It is only in such a setting that an ordinary investment that generates no positive technological externalities has the broad social benefit of a public good.

The second reason comes from the inequity felt by the big contributors to state revenue. They realize that the egalitarian government will use the pooled state revenue to benefit the smaller contributors (including government officials), and they will therefore attempt to consume as much of their output as possible to minimize the profits that the state can tax.

The third reason for the proclivity to overconsume comes from the special institutional arrangements of a public ownership economy. In practice, a portion of the personal income received by the individuals in such an economy comes from the return on capital.6 This capital income, however, is not given to an individual as “returns on capital” (profits or dividends), but is lumped in as part of “returns on labor.” As a result, in the eyes of individuals, the return on labor is higher and the return on capital is lower than they actually are. In other words, individuals perceive a production function that differs from the true one. This kind of distorted perception leads individuals to discount the contribution of capital accumulation to growth and therefore to prefer a lower investment-consumption ratio.

In a centralized public ownership economy, all decisions on the distribution of expenditure come from the state. Individuals’ requests for higher personal income and consumption may have some influence on decision making, but by definition the central planner’s choice will dominate. Although in some situations the planner’s choice will differ from the optimal ratio, generally speaking, the planner will choose a ratio higher than dc because he or she is responsible for capital accumulation and has more complete information about the role capital plays in economic growth. In fact, the history of socialist economies suggests that the actual d under the central planning system was usually higher than d* despite the inherent desire to overconsume.

In a decentralized public ownership economy, where local governments and SOEs have autonomous decision-making power, individuals’ requests for higher personal income and consumption become much more influential than in a centralized one. The result is that the investment-consumption ratio will be closer to dc, and in general lower than the d in a centralized public ownership economy. Our prediction is that the greater the autonomy, the higher the proportion of personal income to total enterprise revenue.

Decentralization Reforms in China

“Decentralization reforms” refers to the devolution of decision-making powers from the Central Government to various levels of the local governments, the SOBs, and the SOEs. These changes affect the managerial system and not the ownership structure.

The Chinese state industrial sector was decentralized in two stages during 1979–89. In the first stage, 1979 to 1984, the “fiscal responsibility system” was adopted to reshape fiscal ties between the central and local governments. Local governments were given greater autonomy in making decisions concerning local economic development (see Wong, Heady, and Woo, 1995). A bonus system was introduced to improve labor productivity in SOEs. Other than this, little discretionary power was given to the enterprises.

The second stage of the decentralization program occurred from 1984 to 1989. SOEs started operating under the “contract responsibility system,” whereby an SOE would enter into a contract with the state (the owner), which would specify the amounts (or rates) of taxes and profits that it would pay each period. In return, the SOE would have much greater autonomy in determining production mix, production level, labor compensation, investments, and use of the retained profits.

The local banks began playing a more important role in 1983 when the state stopped providing circulating capital to the SOEs and gave this function to the SOBs. At the same time, SOBs were allowed to make long-term investment loans. Although the state continued to provide investment funds, budget funds became a less important source of investment financing over time. A major institutional change occurred in 1984 when the local banks were granted greater autonomy in their loan decisions.

However, the administrative structure of the financial system remained unchanged. A local bank was still required to promote the development of the local economy and subordinate itself to the local government’s guidance. It was common for a local bank to have various administrative ties to the local government and for the manager of the local bank to be appointed on the basis of the local government’s recommendation or approval.

Macroeconomic Consequences of the Decentralization

To understand the relationship between inflation and the decentralized system, we need to understand the behavior of the local governments, the SOEs, and the SOBs.

When an SOE is considered only from the supply side, expanding its autonomy could result in some improvements in its economic efficiency. But an SOE is also in a position to use public resources, including borrowed financial resources, to claim a larger share of the national income. With the expansion of enterprise autonomy, an SOE perceives that it can increase both present and future consumption without any trade-off between them. It recognizes that present investment spending, by expanding future output, is actually a guarantee that future consumption will increase. An SOE can raise consumption and investment spending simultaneously under decentralization reforms because not only is it allowed to retain more of its profits, but also it is more likely to be able to increase its bank liabilities.

This second source of funds for the SOEs comes from the greater autonomy of the local governments and local banks. Although the credit quotas set by the central financial authority were essentially unchanged by the 1978–89 reforms, the local banks after 1984 had greater incentive and faced greater pressure to expand credit beyond their quotas. The incentive stemmed from the fact that the personal incomes of the local banks had become dependent on the volume of their lending and the overall prosperity of the local economy. The pressure to exceed the credit quota came from the local governments, which were often co-investors in the local SOEs and also oversaw the management of the local SOBs.

This confluence of self-interest and external pressure led many SOBs not only to ignore the credit quotas when they had excess reserves but also to find ways to “squeeze” reserves from the central bank. A common method was to lend to local enterprises the funds designated for projects in the central plan. When a centrally directed project began to draw on its centrally allocated credits, the local bank would present the central bank with the dilemma of supporting or stopping the fulfillment of the central plan. The usual central bank response was to provide more credit to enable the completion of the central plan projects.

Another common way of squeezing the central bank for funds was to present the central bank with the fait accompli that the local bank had extended credits beyond its reserve base. Because many banks were overextended, the central bank normally opted for accommodation rather than censure.7 It is this combination of actions by the local governments, the SOEs, and local banks that has raised aggregate demand continuously and caused inflation to accelerate since 1984.

In the survey of 300 enterprises, we easily found evidence consistent with the enterprises’ inherent tendency to overconsume. Enterprise managers were asked the following question: “How much pressure are you under from the workers to increase their income in various forms?” Thirty percent said that they were under very high pressure, 43 percent under quite high pressure, 21 percent under moderate pressure, 5 percent under little pressure, and 1 percent felt no pressure.

This finding, of course, cannot be taken as unique to China’s decentralized socialism. Economics assumes as a matter of course that utility-maximizing agents everywhere desire ever-increasing wage compensation and make their desire clear to their supervisors in a manner that is consistent with continued employment. However, wishing for higher labor compensation is not the same as receiving it. We shall show that SOE personnel in China have been extremely successful in realizing their desire. This may explain why the Government responded with many regulations in the 1980s to control wage compensation. Our evidence will suggest that SOEs have been able to skirt these regulations by increasing other forms of personal income. We define personal income as cash income and income in-kind, or direct income and indirect income, respectively.

Effects on Direct Income

Cash income is the sum of the basic wage and the different types of bonuses. Table 2 shows that the direct income of workers increased by 117 percent during 1984–88, while net output increased by only 59 percent. Cash income rose because of the near doubling of the basic wage and the near tripling of the bonus. These increases caused a significant rise in the real level of direct labor compensation because the rise in direct income exceeded the rise in the retail price level by 7–15 percentage points every year. The rise in net output value, in contrast, exceeded the inflation rate by only 1–5 percentage points.8

Table 2.Direct Labor Compensation, Output, Profits, and Taxes(In nominal terms)
Direct Labor Compensation




Year(Normalized by

1984 direct income = 100)
(Rate of increase;

Source: Survey of 300 medium and large state-owned enterprises.Note: Net value output is the sum of value added; it is identical to national income at the aggregate level.





Taxes and



Taxes and




Indices, 1984 = 100
Rate of change; percent
Source: Survey of 300 medium and large state-owned enterprises.Note: Net value output is the sum of value added; it is identical to national income at the aggregate level.
Source: Survey of 300 medium and large state-owned enterprises.Note: Net value output is the sum of value added; it is identical to national income at the aggregate level.

Direct labor compensation continued to grow in real terms when real profits-fell in 1987 and 1988 (Table 2). Real direct income of SOE personnel rose by 8 percent and 10 percent in the two years, respectively, while real total profits fell by 38 percent and 8 percent, respectively.

The large increases in direct labor compensation occurred not only because of the introduction of bonuses but also because of the financial decentralization. Bonuses were introduced in 1979, and they were dispensed so generously that a tax was introduced in 1984 to discourage SOEs from giving big bonuses. An annual bonus of up to four months of basic wages was exempted from the bonus tax; but the fifth-month bonus would require the SOE to pay a 100 percent bonus tax on it, the sixth-month bonus would be subject to a 200 percent bonus tax, the seventh-month bonus to a 300 percent bonus tax, and so on.

In retrospect, the 1979–84 bonuses granted in the absence of the bonus tax were still quite restrained. The share of direct income in net output value in 1984 was virtually identical to that in 1980 (Table 3). When wide financial autonomy was given to the SOEs in 1984, the share of direct income rose from 12 percent in 1984 to 16 percent in 1988 despite the presence of the steeply progressive bonus tax.

Table 3.Distribution of Net Output Value (Value Added)



Taxes and


Pretax Debt






Share of net output value; percent
Normalized by 1984 direct income =100
Rate of change; percent
Source: Survey of 300 medium and large state-owned enterprises.
Source: Survey of 300 medium and large state-owned enterprises.

Effects on Indirect Income

The indirect income received by SOE personnel comprised three components: collective consumption, distribution of private consumer goods, and housing. The first two items were treated as production costs and the third as investment expenditure.

Collective consumption came from the myriad services supplied by the SOEs to their employees. The range of services was so broad that an SOE resembled a self-contained social community. They included kindergartens, hospitals, transportation, recreational facilities, dining facilities, funeral facilities, and relief work, all of which were paid by the welfare fund. The size of the welfare fund allowed by state regulations was proportional to the amount of cash income (wages plus bonuses) and retained profits. Our survey sample showed a 240 percent increase in the size of the welfare fund during 1984–88, a finding consistent with the large increases in cash income and retained profits documented in Tables 2 and 3.

Consumer goods were distributed at two levels: the worker level and the management level. Many SOEs bought grain, fruits, meat, eggs, fish, clothes, furniture, and housewares and distributed them to their employees to enable the employees to avoid the bonus tax. The costs of these items were charged mostly to “material costs” and “nonproductive expenditure.” This system partly explains why material costs rose 82 percent during 1984–88, and nonproductive expenditure rose 234 percent (see Table 4). Studies by Zhang (1990) and Zhao (1989) suggest that the distribution of consumer goods accounted for 25–33 percent of the total personal income of SOE employees.

Table 4.Sources of Indirect Income
Indices of Nonwage Cost, 1984 = 100









Net Non-


Note: Average of 300 enterprises.

Expenditure on raw materials, intermediate goods, and energy.

Total management cost minus cost for new product research minus technological research cost minus training cost.

Total nonproductive expenditure minus labor insurance minus expenditure on enterprise-run schools.

Composition of Capital Stock
Total Value

at Original

Book Value
Total Value

of Productive

Fixed Assets
Total Value

of Nonproductive

Fixed Assets
Proportion of

Assets to

Total Assets

(thousands of renminbi)
Note: Average of 300 enterprises.

Expenditure on raw materials, intermediate goods, and energy.

Total management cost minus cost for new product research minus technological research cost minus training cost.

Total nonproductive expenditure minus labor insurance minus expenditure on enterprise-run schools.

Note: Average of 300 enterprises.

Expenditure on raw materials, intermediate goods, and energy.

Total management cost minus cost for new product research minus technological research cost minus training cost.

Total nonproductive expenditure minus labor insurance minus expenditure on enterprise-run schools.

Consumer goods—for example, luxurious official cars and expensive office furniture—were also distributed in the management office and the sales department. Consumption also took the form of lavish banquets, tourist travel, and high-class hotels. The survey of 300 enterprises showed that net management costs rose 165 percent in 1984–88, and sales costs rose 300 percent (Table 4). Management costs expressed as a proportion of the “GNP”9 rose steadily as well: 2.8 percent in 1980, 3.9 percent in 1984, 4.4 percent in 1985, 4.8 percent in 1986, 5.4 percent in 1987, and 5.9 percent in 1988.

The third component of the indirect income paid by SOEs was housing. SOEs used a significant portion of the retained profits to build apartments (and physical structures for kindergartens and employees’ clubs), renting them to their workers at rates that would cover only maintenance costs. This diversion of investment funds is clearly seen in Table 4. The value of nonproductive fixed assets increased by 160 percent during 1980–88, while the value of productive fixed assets increased by only 92 percent. As a result, the proportion of nonproductive fixed assets to total fixed assets rose from 15 percent in 1980 to 19 percent in 1988.

Effects on Labor Productivity, Retained Profits, and State Revenue

Table 2 shows that there was an 8.7 percent increase in the real net output value during 1984–88. Although this rate of growth is not impressive, it is still tempting to argue that the generous increases in personal income were necessary to induce this positive output growth. But this argument is hard to make because there were extremely large investments over this period. The real stock of productive capital in 1988 was 35 percent larger than in 1984.10 Because there is no reason to believe that this additional capital stock was left unused, we think that this massive infusion of capital, rather than increases in labor efficiency, was responsible for the output growth.

A simple growth-accounting exercise supports our hypothesis of the extensive nature of SOE output growth. Assuming a capital share of income of 30 percent, we expect a net output growth of 10.5 percent. But since the actual net output growth was only 8.7 percent, it suggests either negative growth in total factor productivity or a decline in labor efficiency during 1984–88. We note that the 30 percent capital share of income is typical only for advanced industrial economies; the capital share of income in low-income countries is typically much larger, say, 50–60 percent. So, if we use the more realistic finding of Chen and others (1988) that the share of capital income is 54 percent under the Cobb-Douglas specification of the production function and 72 percent under the translog specification, then the decline in the efficiency of SOEs would be even worse.

Our conclusion is that the 45 percent increase in real cash income in 1984–88 did not induce any noticeable gains in labor efficiency. We also conclude, in light of the survey finding that over 70 percent of managers faced strong worker pressure for higher pay, that the explanation for the large real personal income gains is that the managers caved in to the consumption demands of the workers. The decentralization reforms gave the managers the financial room to cave in.

Table 3 shows that the total taxes paid and profits remitted to the state by SOEs fell as a proportion of net output value: from 78 percent in 1980, to 66 percent in 1984 and 56 percent in 1988. This fall in proportion was expected; it is the essence of decentralization. What might not have been expected, however, was the drop in the amount of real resources transferred from the SOEs to the state. Inflation exceeded the increase in taxes and profits received by the state by 5 percentage points in 1986, 6 percentage points in 1987, and 2 percentage points in 1988 (Table 2).

There are five reasons for this decline in real revenue from the SOEs. First, the economic slowdown in 1986 helped decrease nominal profits by 31 percent in 1987. Second, in 1985, centrally allocated investment grants to SOEs were replaced by investment loans; this change reduced the amount of profits to be taxed because SOEs were allowed to deduct the debt services as cost. The debt service rose from 1 percent of net output value in 1980 to 6 percent in 1987 before falling to 4 percent in 1988 (see Table 3).

The third reason for the decline in revenue from SOEs was the large increases in direct labor compensation, which further decreased total profits. Direct income absorbed 16 percent of net output value in 1988 compared with 12 percent in 1980. Fourth, the large increases in indirect labor compensation raised production costs and reduced profits. The misdirection of investment funds into housing meant that output and profits would not increase in the future.

Finally, the SOEs were retaining an increasingly large proportion of profits. They retained 5 percent of profits in 1980, 11 percent in 1984, and 16 percent in 1988 (Table 3). As a result, the amount of retained profits increased 126 percent from 1984 to 1988.

Our findings so far indicate that decentralization has weakened the fiscal base of the Central Government without necessarily achieving the desired goals of rationalizing the allocation of investment spending and increasing production efficiency. A significant portion of what would have been state revenue was diverted into consumption and nonproductive investments. Furthermore, the decline in real state revenue from the SOEs exacerbated the budget deficit, and, consequently, increased the printing of money.

Effects on Investment

Ever since China embarked on decentralization in early 1980, a new term, “hunger for investment,” has appeared in discussions on the economy.11 The term, which refers to the surge in demand for investment funds in every economic sector in every province, is the product of three factors, of which the first two are mutually reinforcing. First, investment was the vehicle through which the SOEs could increase their levels of future consumption. Second, SOEs realized that, being a part of the Government, they would not be closed even if an imprudent investment were to result in bankruptcy. Because SOEs could count on being bailed out during adversity, they perceived negative rates of return as zero rates of return. An SOE’s expected rate of return on an investment was therefore higher than the actual expected rate of return, biasing the SOE toward investment activities. This second factor has been called the soft budget (Kornai, 1980). The Chinese have aptly summed up these first two causes of the hunger for investment: “The losses of SOEs are socialized but the profits of SOEs are privatized.”

The third factor behind the hunger for investment is the career considerations of SOE managers. They realize that their prospects for promotion to a larger enterprise are improved if they have a record of engineering a large expansion in output. Adding production capacity is one way of meeting that criterion. As long as credit is available, the additional investment will not occur at the expense of employee consumption. In fact, it would be optimal if managers could get a loan larger than the investment. They could then use the extra funds for consumption, which would solidify their popularity with employees and, according to Walder (1988), stand them in good stead during promotion deliberations. Career considerations bias the SOE manager toward overinvestment and overconsumption.

Table 5 presents evidence of the hunger for investment, showing that the proportion of loan applications by SOEs that was approved fell steadily during 1986–88.12 In 1988, only 20 percent of firms had 75–100 percent of their loan applications approved compared with 65 percent in 1986. The main reason for the tougher acceptance criterion was that the number of loan applications multiplied even faster than the explosion of bank credit.

Table 5.Increased Demand for Investment(Ratio of approvals to proposals; percent)
Distribution of SOEs198619871988
Below 254.97.918.6

The managers of the 300 enterprises surveyed were asked the following question: “How much would you reduce your demand for loans if the interest rate were to increase by 5 to 10 percentage points?” The average interest rate for loans in this period was about 12 percent.

The managers’ responses indicate that the interest rate elasticity of loans was very low. Eighty-five percent of managers would not change the amount of investment they would undertake if the interest rate were to increase by 5 percentage points, and 66 percent would still not change their investment plans if the increase were 10 percentage points instead.

Table 6 shows the sources of investment funds. The amount of external funds increased more rapidly than the amount of internal funds, rising by a factor of 4 and 2, respectively, during 1984–88. The amount of external funds exploded in 1988, raising the ratio of external to internal funds from 1 in 1987 to 2. Another indicator that easy bank credit was responsible for the threefold jump in investment spending during 1984–88 was the rise in the ratio of bank lending to total investment, particularly in 1988. The result of this easy bank credit was the boom in productive investment and nonproductive (housing) investment. The evidence is that the proportion of investment in each category that was covered by internal funds fell steadily over time.

Table 6.Sources of Investment Funds
Index of total investment100.00141.42186.25277.68322.10
Index of internal funds available for investment100.00158.24184.64233.68229.11
Internal funds as percent of total investment36.7241.0936.4030.90726.12
Index of total external funds available for investment1100.00142.51170.55239.67370.01
External funds as percent of total investment241.0741.3837.6035.4447.18
Bank borrowing3 as percent of total investment24.9928.3824.6519.1738.90
Percent of total productive investment covered by internal funds37.5135.7333.0731.0326.68
Percent of total nonproductive investment covered by internal funds61.8571.7168.3763.4052.80
Note: The results are an average of 300 state-owned enterprises.

“External funds” are the sum of state grants from governments at all levels and bank loans and do not include funds from unspecified “other sources.” The “state grants” include those grants distributed through banks as loans.

The sum of shares of “internal funds” and “external funds” does not equal 100 percent because it does not include the share of funds from unspecified “other sources.”

“Borrowing” here refers to true banking loans and does not include state grants distributed in the form of loans.

Note: The results are an average of 300 state-owned enterprises.

“External funds” are the sum of state grants from governments at all levels and bank loans and do not include funds from unspecified “other sources.” The “state grants” include those grants distributed through banks as loans.

The sum of shares of “internal funds” and “external funds” does not equal 100 percent because it does not include the share of funds from unspecified “other sources.”

“Borrowing” here refers to true banking loans and does not include state grants distributed in the form of loans.

Effects at the Aggregate Level

Two ways in which the individual optimizing behavior of the SOEs has caused the money supply to expand have been discussed. First, the increases in labor compensation led to the widening of the budget deficit, a large proportion of which was immediately monetized. Our finding that the amount of real resources transferred to the state fell during 1986–88 has its parallel at the aggregate level: total real government revenue fell during 1986–89 (see Table 7).

Table 7.Government Revenue and Expenditure
Billions of yuan
Total revenue124.81126.4130.1130.2140.9160.7184.3229.7244.7257.6280.4324.5
Price subsidies1.17.911.815.917.219.721.826.225.729.531.737.0
Enterprise subsidies11.511.714.112.619.724.020.025.932.537.644.760.0
Total expenditure123.7147.0147.1140.1152.2173.0196.5236.6265.6282.5315.3361.4
Memorandum item: Budget deficit−1.020.617.09.911.312.312.26.820.924.934.937.0
Proportion of total expenditure; percent
Total revenue100.886.088.493.092.692.993.897.
Price subsidies0.
Enterprise subsidies9.
Growth rate; percent
Total revenue1.
Price subsidies613.548.635.48.014.610.619.9−1.814.87.516.7
Enterprise subsidies1.320.5−10.255.821.8−16.729.525.515.918.634.3
Total expenditure18.80.1−4.88.713.613.620.412.36.411.614.6
Memorandum item: Inflation (retail)
Notes: The Chinese definition of total revenue was adjusted to exclude borrowing and include price subsidies (until 1985) and enterprise subsidies. The Chinese definition of total expenditure was adjusted to include price subsidies (until 1985) and enterprise subsidies. We ignored principal repayments and extrabudgetary capital construction because they are small items. These adjustments made the concepts closer to the IMF definitions.
Notes: The Chinese definition of total revenue was adjusted to exclude borrowing and include price subsidies (until 1985) and enterprise subsidies. The Chinese definition of total expenditure was adjusted to include price subsidies (until 1985) and enterprise subsidies. We ignored principal repayments and extrabudgetary capital construction because they are small items. These adjustments made the concepts closer to the IMF definitions.

What widened the budget deficit was that total expenditure growth did not slow down in line with the slower revenue growth. The proportion of expenditure that could be covered by revenue fell from 94 percent in 1984 to 90 percent in 1989. One major reason that expenditure growth could not be reduced was that large increases in subsidies went to enterprises to cover their mounting losses after the decentralization. Enterprise subsidies soared from Y 20 billion in 1984 to Y 60 billion in 1989, raising subsidies as a share of total expenditure to 17 percent from 10 percent.

In a nutshell, the large personal income increases of SOE personnel enlarged the budget deficit by denying the state (owner) its revenue, and, in some cases, bankrupting the firms. The second action obligated the state to disburse subsidies to maintain employment. This first route through which SOEs destabilized the economy can be described as the “overconsumption/money creation” mechanism.

The second route of destabilization can be described as the “overinvestment/money creation” mechanism. The survey of 300 enterprises revealed the great hunger for investment, which was satisfied by external funds. The frequent expansion of central bank credit to accommodate this demand for investment credit is confirmed by the balance sheet of the People’s Bank of China (see Table 8).13 “Loans to SOBs” is the biggest contributor to the growth rate of high-powered money whenever the latter was greater than 20 percent. It was hence natural that the authorities reined in “loans to SOEs” whenever they needed to cool down the economy. The primary importance of “loans to SOBs” in the periods of overexpansion of credit supports our theory that macroeconomic instability originates in local banks helping local firms in their quest for capital formation.

Table 8.Contributions to the Growth of High-Powered Money(In percentage points)
Foreign reserves−
Loans to SOBs19.52.919.020.23.9
Loans to SOEs1.
Loans to government6.75.43.1−1.42.4
High-powered money23.714.025.923.714.4
Source: Balance sheet of the People’s Bank of China.
Source: Balance sheet of the People’s Bank of China.

The decomposition of GDP growth during 1981–89 confirms this analysis (see Table 9). With the decentralization in 1984, consumption spending leapt to 11 percent, and investment spending to 21 percent. However, with a given output capacity at any point in time, the tendencies to overconsume and over-invest cannot both be realized on a prolonged basis unless massive external borrowing occurs. Which tendency dominates, and can be seen in the data, depends on the stance of monetary policy.

Table 9.Components of Aggregate Demand
Constant 1987 prices; billions of yuan
Private consumption391.3412.8429.9470.5522.2570.4585.0621.9685.4746.2
Government consumption34.045.448.152.558.161.264.370.078.485.6
Domestic investment191.4182.1200.5225.6272.4368.5404.2445.0491.1477.3
Net exports−14.8−−2.0−43.7−21.5−1.1−7.0−21.8
Contribution in percentage points
Private consumption3.
Government consumption1.
Domestic investment−−1.1
Net exports1.02.3−0.5−0.5−−0.5−1.2
Source: World Bank staff estimates.
Source: World Bank staff estimates.

The 1984 change in the incentives to the financial system is the reason investment exceeded consumption as the leading expenditure category during 1985–87. The burst of bank credit in 1985 allowed domestic investment spending to contribute 11 percentage points to the overall growth rate. The credit expansion allowed investment to crowd out consumption in the competition for the use of resources. This is why consumption grew only 3 percent in 1986.

However, with the reining in of credit (and hence investment) in late 1988, the overconsumption tendency asserted itself in the data. Consumption grew 9 percent in 1989 despite the slowdown in GDP growth to 3 percent.


We have provided what we deem to be reasonable evidence in support of the “overconsumption/money creation” and the “overinvestment/money creation” mechanisms. We argued that the decentralization reforms have allowed the innate tendencies to overconsume and overinvest to realize themselves. Managers now have the financial autonomy to accommodate the demands for higher personal income and they have no incentive to resist. Furthermore, it is in the interests of the local governments and local banks to help the SOEs to obtain the credit they need to increase local output.

We found two results that are at odds with several recent papers. Our evidence of an increasing proportion of output being paid to employees (as direct and indirect income) suggests that the increasing budget deficits were caused by upward “wage drift” at the enterprise level. We therefore dispute Naughton’s (1991) claim that the lower profits of the SOEs were primarily caused by the competition provided by the emergence of non-SOEs. By the logic of Naughton’s analysis, the budget deficit should be reduced by policing the non-SOEs more strictly to prevent tax evasion and by imposing a consumption tax to regain the transfer given to the consumer through the lower prices fostered by competition.

We have three problems with Naughton’s explanation. The first is that many heavy industries experienced rapidly falling profits even though there was little or no entry by non-SOEs.

Second, there are very few examples of prices having fallen because of the emergence of competition from non-SOEs; that is, there is little evidence of the mechanism that allegedly reduced the profits of the SOEs. Admittedly, it is hard to find instances of falling prices in the mixed situation of general inflation and of price liberalization to ease persistent shortages. There have been some significant shifts in relative prices, but it is not clear whether they were due to competitive pressures provided by the non-SOEs or to the natural corrections that came with the removal of price controls.

The third problem with Naughton’s view is that available evidence suggests that the SOEs and not the non-SOEs have been the biggest tax evaders.14 The non-SOEs have not snatched away the pre-decentralization profits of the SOEs; the decentralization reforms have allowed the SOEs to simply appropriate them.

Our finding that there has been a decline in total factor productivity during 1984–88 implies that the increase in the production efficiency of SOEs found by Chen and others (1988), Dollar (1990), Granick (1990), and Jefferson (1990) immediately after the initiation of reforms in 1978 was only a temporary phenomenon.15 The deterioration in the production efficiency of the SOE sector in the second half of the 1980s is supported by the research of Wang (1990)16 and Stepanek (1992).

Our explanation for why the productivity increase was not sustained is that the managers bowed to egalitarian pressures and discontinued the link between worker performance and bonus. The workers realized over time that collective action was more cost-effective than mutual competition in securing a higher general level of consumption. In fact, even the local industrial bureau could not impose a link between firm performance and bonus. This is why Jefferson and Xu (1992) found, in their sample of 20 SOEs in Wuhan during 1984–87, that there was “a convergence of returns to labor … [despite] the relatively unreformed nature of labor markets” (p. 65). The result of the discontinuation of the link between performance and reward was that labor efficiency growth sputtered out over time.

Our back-of-envelope calculation of total factor productivity growth in SOEs is corroborated by the panel estimation of the data in Woo, Hai, Jin, and Fan (1994). In that article, we found that positive growth of total factor productivity could be generated only if we used the intermediate input deflator constructed by Jefferson, Rawski, and Zheng (1992), but the use of that input deflator generated secularly declining value-added deflators for their data set and our data set. The anomalousness of secularly declining value-added deflators in a period of secularly rising output prices makes the positive total factor productivity results in Jefferson, Rawski, and Zheng (1992) and in Groves and others (forthcoming) highly suspect. (This issue is debated in Woo, Fan, Hai, and Jin (1993, 1994) and Jefferson, Rawski, and Zheng (1994).)

Our finding that SOE productivity performance during 1984–88 was poor is also at odds with the claims of Jefferson and Xu (1991) and Groves and others (1991). They found that labor productivity was positively correlated with various indicators of reforms (for example, bonus payments and proportion of employees on contract). The problem with the regressions in these two studies is that they do not control for changes in the capital stock. We do not find it plausible to assume that labor productivity was unaffected by the large amount of investment undertaken after 1984.

Because tolerance of the inflation generated by overconsumption and overinvestment cannot be justified by efficiency gains in the SOEs, the way that monetary policy has been used periodically to combat inflation is damning. One of the long-recognized favorable side effects of an economic slowdown is the sloughing off of a layer of accumulated inefficient producers.17 Yet, during the last two policy-induced economic slowdowns, the reduced amount of credit available was channeled primarily to the SOEs, forcing many efficient non-SOEs to close.

The chief lesson that can be drawn is that the SOEs must be restructured if macroeconomic instability is to be reduced and economic growth enhanced. An important question is whether any state in which the predominant mode of ownership is public can have the political will to enforce tight budget constraints on SOEs. Kornai (1990) is pessimistic on this question, but Roemer (1991) has recently proposed several novel mechanisms that might induce SOEs to imitate the management practices of private firms.18


    BowlesPaul andGordonWhite1989“Contradictions in China’s Financial Reforms: The Relationship Between Banks and Enterprises,”Cambridge Journal of EconomicsVol. 13 (December) pp. 48195.

    • Search Google Scholar
    • Export Citation

    ChenKuan and others1988“Productivity Change in Chinese Industry: 1953–1985,”Journal of Comparative EconomicsVol. 12 (December) pp. 57091.

    • Search Google Scholar
    • Export Citation

    DollarDavid1990“Economic Reform and Allocative Efficiency in China’s State-Owned Industry,”Economic Development and Cultural ChangeVol. 39 (October) pp. 89105.

    • Search Google Scholar
    • Export Citation

    FanGang1990Macroeconomic Analysis of Public Ownership (in Chinese) (Shanghai, China: United Press).

    FanGang1994“Incremental Changes and Dual-Track Transition: Understanding the Case of China,”Economic Policy: A European ForumVol. 9supplement (December) pp. 99122.

    • Search Google Scholar
    • Export Citation

    GranickDavid1990Chinese State Enterprises: A Regional Property Rights Analysis (Chicago: University of Chicago Press).

    GrovesTheodore and others1991“Autonomy and Incentives in Chinese State Enterprises” (unpublished; San Diego: University of California).

    • Search Google Scholar
    • Export Citation

    GrovesTheodoreforthcoming“Productivity Growth in Chinese State-Run Industry,”in Reform of China’s State-Owned Enterprisesed. by FurengDongCyrilLin andBarryNaughton (London: Macmillan).

    • Search Google Scholar
    • Export Citation

    JeffersonGary1990“China’s Iron and Steel Industry: Sources of Enterprise Efficiency and the Impact of Reform,”Journal of Development EconomicsVol. 33 (October) pp. 32955.

    • Search Google Scholar
    • Export Citation

    JeffersonGary andWenyiXu1991“The Impact of Reform on Socialist Enterprises in Transition: Structure, Conduct, and Performance in Chinese Industry,”Journal of Comparative EconomicsVol. 15 (March) pp. 4564.

    • Search Google Scholar
    • Export Citation

    JeffersonGaryThomasRawski andYuxinZheng1992“Growth, Efficiency, and Convergence in China’s State and Collective Industry,”Economic Development and Cultural ChangeVol. 40 (January) pp. 23966.

    • Search Google Scholar
    • Export Citation

    JeffersonGary1994“Productivity Change in Chinese Industry: A Comment,”China Economic ReviewVol. 5 pp. 23548.

    KornaiJanos1980Economics of Shortage (Amsterdam; New York: North-Holland).

    KornaiJanos1990The Road to a Free Economy: Shifting from a Socialist System: The Example of Hungary (New York: Norton).

    NaughtonBarry1991“Implications of the State Monopoly over Industry and Its Relaxation” (unpublished; San Diego: University of California).

    • Search Google Scholar
    • Export Citation

    ReynoldsBrucemed.1987Reform in China: Challenges and Choices (Armonk, New York: M.E. Sharpe).

    RoemerJohn1991“Market Socialism: A Blueprint,”Dissent (Fall) pp. 56269.

    SachsJeffrey andWing ThyeWoo1994“Structural Factors in the Economic Reforms of China, Eastern Europe, and the Former Soviet Union,”Economic Policy: A European ForumVol. 9 (April) pp. 10145.

    • Search Google Scholar
    • Export Citation

    SachsJeffrey andWing ThyeWoo1996“Understanding China’s Economic Performance” (unpublished; Davis: University of California).

    StepanekJames1992“China’s Enduring State Factories: Why Ten Years of Reform Have Left China’s Big State Factories Unchanged,”in China’s Economic Dilemmas in the 1990s: The Problem of Reforms Modernization and Interdependenceed. by the Joint Economic Committee, Congress of the United States (Armonk, New York: M.E. Sharpe).

    • Search Google Scholar
    • Export Citation

    WalderAndrew1988“Factory and Manager in an Era of Reform” (unpublished; Cambridge: Harvard University).

    WangXuefeng1990“The Decline of Budget Revenues as a Proportion of National Income,”Jingji Yanjiu (Economic Research), No. 4.

    WongChristineChristopherHeady andWing ThyeWoo1995Fiscal Management and Economic Reform in the People’s Republic of China (New York: Oxford University Press).

    • Search Google Scholar
    • Export Citation

    WooWing Thye1994“The Art of Reforming Centrally Planned Economies: Comparing China, Poland, and Russia,”Journal of Comparative EconomicsVol. 18 (June) pp. 276308.

    • Search Google Scholar
    • Export Citation

    WooWing Thyeforthcoming“Improving the Performance of State-Owned Enterprises,”in Issues and National Experiences in Economic Transitioned. by Wing ThyeWooStephenParker andJeffreySachs (Cambridge, Massachusetts: MIT Press).

    • Search Google Scholar
    • Export Citation

    WooWing ThyeGangFanWenHai andYinbiaoJin1993“The Efficiency and Macroeconomic Consequences of Chinese Enterprise Reform,”China Economic ReviewVol. 4 pp. 15368.

    • Search Google Scholar
    • Export Citation

    WooWing Thye1994“Productivity Change in Chinese Industry: Reply to Jefferson, Rawski and Zheng,”China Economic ReviewVol. 5.

    WooWing ThyeWenHaiYinbiaoJin andGangFan1994“How Successful Has Chinese Enterprise Reform Been? Pitfalls in Opposite Biases and Focus,”Journal of Comparative EconomicsVol. 18 (June) pp. 1037.

    • Search Google Scholar
    • Export Citation

    ZhangShuguang1990“Personal Income in Kind” (in Chinese) Jingji Yanjiu (Economic research) No. 12.

    ZhaoRenwei1989“Income-in-kind in the Process of Economic Reform” (in Chinese) Jingji Yanjiu (Economic research) No. 4.

Note: We are grateful to the Ford Foundation, the International Center for Economic Growth, and the United Nations’ World Institute for Development Economics Research for supporting our research; to Charles Adams, Eduardo Borensztein, Alan Gelb, David Lipton, Robert Mundell, Dwight Perkins, John Roemer, Jeffrey Sachs, Inderjit Singh, and Wu Jing Lian for helpful conversations; to Barry Naughton, Thomas Rawski, and the participants of the International Conference on China’s Growth and Inflation (held in May 1995 in Beijing) for comments; to Elisa Diehl for improving our exposition; and to Hong Chen, Rasa Dale, and Sam Lee for excellent research assistance.
1This refers primarily to the development of nonstate enterprises such as private and cooperative businesses in urban areas, and village and township enterprises in rural areas.
2For an overview of the nature and consequences of China’s reforms, see Fan (1994), and Sachs and Woo (1996); and for a comparative analysis of Chinese and Eastern European reforms, see Sachs and Woo (1994) and Woo (1994).
3“Yi zhua jiu si, yi fang jiu luan”; that is, “the economy, once gripped, is dead, and, once loosened, is unstable.”
4The theory outlined here is discussed more fully in Fan (1990).
5Even after a decade of decentralization, the Chinese planning agency still set the wage standard for the whole country, and many regulations are in place to implement it.
6This is entirely appropriate because all individuals are equal co-owners of the capital stock.
7Our conclusion that the banks have become less prudent is shared by Bowles and White (1989), who wrote that “after eight years of reform, …the credit constraint may have become softer” (pp. 487–88).
8It was negative in 1986 and 1988 if the cost of living index is used instead.
9“GNP” equals net output value plus capital depreciation.
10This is calculated by adding the deflated incremental changes in the productive capital stock (see Table 4 for nominal stock figures). The deflator used was the retail price level.
11See, for example, the enterprise survey report of the Chinese Economic System Reform Research Institute, which has been translated into English (Reynolds, 1987).
12The original draft of the questionnaire asked for the amount of investment funds applied for in each year. On the advice of the State Statistical Bureau (SSB), this question was changed to ask the number of projects proposed and the number funded. The SSB felt that managers would not remember the value of the projects denied.
13Data are available only from June 1985 because the People’s Bank of China assumed traditional bank functions in 1984.
14The article “Who Are the Biggest Tax Evaders?” Economic Daily, January 1, 1992, summarized the data from several studies on taxation.
15Chen and others (1988) found that aggregate multifactor productivity growth went from 0.4 percent during 1957–78 to 1.9 percent during 1978–83. Jefferson (1990) found that the multifactor productivity growth of the largest 120 iron and steel enterprises (of a total of 1,318) went up from -1.6 percent in 1957–80 to 1.8 percent in 1980–85. Dollar (1990), examining a sample of 20 firms, found suggestive evidence of total factor productivity being 0.2 percentage points higher in 1979–82 than in 1975–78. Granick (1990), using the same data set as Dollar, found that the frequency with which Chinese firms achieved or exceeded the various quotas increased after the reforms. The Dollar-Granick sample is biased toward the best-performing SOEs: three of their firms were among the ten that were selected in 1983 to receive “national awards for distinction in management” (Granick, p. 13); and, “the sample enterprises were, throughout [1976 to 1982]…, 40 to 120 percent more profitable than were Chinese state-owned industrial enterprises as a whole” (Granick, p. 180).
16But we do not agree with Wang that this is the most important reason that SOE profits fell and hence widened the budget deficit.
17Hence, Schumpeter’s famous characterization of the cyclical downturn in the capitalist economy as a much-needed douching.
18See Woo (forthcoming) for a review of the international experiences in reforming SOEs.

    Other Resources Citing This Publication