Another Look at the Growth of Output in China

Gregory C Chow

Princeton University

August 18, 2004

I. Introduction

This note utilizes data on outputs in physical units for a sample of important industrial products in China as reported in China Statistical Yearbook (CSY hereafter) published by the State Statistical Bureau in order to estimate the rates of growth of real GDP and of industrial output. It was prompted by Young (2003) which claims that the rate of growth of real GDP in the period 1978-1998 as reported in CSY is an overestimate by an order of 2 percent per year. Following Young (p.1221) “rather than discount the Chinese statistical record, I embrace it. Accepting all the numbers the statisticians of the People’s Republic of China produce,” I shall explain why the reported estimate of real output growth is reasonable and where Young’s alternative estimate is in error.

II A Look at the Growth of Components of Industrial Output

Indices of real GDP and output of secondary industry (including industry and construction) in constant prices as reported in CSY 1999 (Table 3-4) grew from 100 in 1978 to 638.2 and 916.8 in 1998 respectively, implying average annual exponential rates of growth of 0.09267 and 0.11079. The output indices of industry and construction (components of secondary industry) reached 938.5 and 743.6 respectively, implying average exponential growth rates of 0.11196 and 0.10032 respectively.

CSY 1999 (Table 13-12) provides data on the output of a sample of important industrial products in physical units from 1978 to 1998. From these data I have computed average annual exponential growth rates of growth as presented in Table 1 under the column headed by PRC.

Table 1

Average Exponential Rates of Growth of Industrial Output

PRC Taiwan

(a) 9 Non-durable Consumer Goods

Chemical Fiber (10000 tons) 0.14430

Yarn (10000 tons) 0.04111

Cloth (100 million m) 0.03908

Silk (10000 tons) 0.04120

Paper and Paper Boards (10,000 tons) 0.07887 0.10911

Sugar (10000 tons) 0.06458

Vegetable Oil (10000 tons) 0.06125

Beer (10000 tons) 0.19529 0.09772

Cigarettes (10000 cases) 0.05244

(b) 5 Durable Consumer Goods

Refrigerators (10000 units) 0.26884

Electric Fans (10000 units) 0.11146

Household Washing Machines (10000 units) 0.19487

Color Television Sets (10000 units) 0.45636 0.25723

Cameras (10000 units) 0.28661

(c) 8 Consumer or Producer Goods

Electricity (100 million kwh) 0.07573 0.09469

Hydropower (100 million kwh) 0.07475

Steel products (10000 tons) 0.07908

Cement (10000 tons) 0.10530 0.10094

Plate Glass (10000 weight cases) 0.11329

Plastics (10000 tons) 0.11612

Motor Vehicles (10000 units) 0.09962 0.18682

Trucks (10000 units) 0.08459

(d) Construction

Construction (floor space completed) 0.10364 ^a 0..09871

Notes: Data for PRC are found in China Statistical Yearbook 1999 (Table 13-12, and Table 14-32 for construction). The growth rates are from 1978 to 1998, except for construction which is from 1985 to 1998 because of data availability. Data for Taiwan are found in Statistical Yearbook 1990 (Table 149; Table 151 for electricity; Table 152 for construction of new building). The growth rates for Taiwan are from 1970 to 1989. Products in Table 1 are listed in the same order as in CSY. Mineral products are excluded. Items with small total values relative to the values of products already included in the same group are also excluded because each product is used in the computation of a mean growth rate; these products include salt, sulfuric acid, soda ash, caustic soda, chemical fertilizer, chemical pesticide (all of very small total values as compared with included products), metal-cutting machines and tractors (of very small values as compared with motor vehicles and trucks combined).

The first 9 items in group (a) are consumer perishable or semi-durable goods which in general grew more slowly than national income or output as they have income elasticities of demand below unity. The next 5 items are consumer durables with higher income elasticities of demand and grew faster than national income. In addition the outputs of some of these products (and beer) grew rapidly for export. The next 8 items in group (c) are consumer or producer goods. The mean exponential growth rates of the products in these groups are 0.07979 for group (a), 0.26363 for group (b) and .09356 for group (c).

The last row of Table 1 gives the average exponential growth rate (starting from 1985 to 1998 because of data availability) of 0.10364 for construction as measured by square meters of floor space completed as reported in CSY 1999 (Table 14-32). This is very close to the reported rate of growth of 0.10032 for the construction industry in the period 1978-1998. Note also that the growth rate for the output of cement in Table 1 is 0.10530, very close to the reported growth rate for construction.

To examine the rate of growth of industrial outputs based on our sample, we first estimate the growth rate of the value of the products in groups (a) and (b) evaluated by their prices in 1987, a year close to the midpoint of the sample period 1978-1998. Prices in 1987 rather than in 1988 are employed mainly because 1988 was a year of serious inflation (at an annual rate of 18.5 percent by the general retail price index) with many relative prices different from their normal values. Selecting a year earlier than 1987 would give more weights to the consumer durables which had higher relative prices in the earlier years. Since the consumer durables had higher growth rates this would produce a higher growth rate for the combination of groups (a) and (b). The quantities of these outputs in 1978 and 1998 are given in SYC 1999 (Table 13-12). The prices in 1987 are given in China Commodity Prices Statistical Yearbook 1989 (primarily on pp. 175-6, supplemented by pp.172-174) as follows. Chemical fiber, missing; yarn, missing; cloth 1.881 (yuan)/meter; silk, missing; paper, 2450.6/ton; sugar, 1537.6/ton; vegetable oil, 2396.2/ton; beer, 1.024/bottle converted to 2.048 per kilogram; cigarettes, 1332.8/case; refrigerator, 1307.7 per unit; electric fan, 157.2 per unit; washing machine, 370.2 per unit; color TV set, 1476.4 per unit; camera, 330.04 per unit. Of the three items with prices missing, chemical fiber has a high growth rate of 0.1443 that offsets the low growth rate of 0.0411 for yarn, and silk has small total value as compared with the first three products of the same category already included in group (a). Therefore the exclusion of these three products from our calculation would not lead to serious errors. Calculated by the above 1987 prices, the total value of the remaining 11 products was 56.94 billion (1987) yuan in 1978 and 353.46 billion (1987) yuan in 1998, implying an exponential growth rate of 0.09129. To make sure that the weights given to the 5 consumer durables are not too large we observe that their total value in 1998 was 86.28, or 24.41 percent of the total value 353.46. Thus our combined growth rate is not unduly influenced by the included consumer durables.

To summarize the information available from our sample, groups (a) and (b) combined provides an exponential growth rate of 0.09129; group (c) provides an mean growth rate of 0.09356 and construction (d) grew in terms of completed floor space at a rate of 0.10364. How should these growth rates be used to evaluate the reported growth rates for GDP, for industrial output and for construction reported in CSY? To arrive at an exponential growth rate for industrial output comparable to the official rate of 0.11196 we have to make an upward adjustment of the above rate 0.09129 due to the following factors: (1) We used the prices of 1987 to weight the physical outputs whereas prices of earlier years than 1987, including 1978, were used for the calculation of real output in CSY which should yield a slightly higher growth rate than the one obtained above. (2) A second source of downward bias of our method as compared with the CSY method is the absence of new products, notably computers, in our sample. The appearance of new products which grew rapidly can result in a much higher growth rate than the one based only on products that existed in 1978. (3) We have not accounted for quality improvement in the products, counting one television color set in 1998 as the same output as one set in 1978. This is in addition to the first bias that results from the same color TV set being cheaper (relative to other goods) in 1987 than in earlier years. As far as construction is concerned, our estimated growth rate 0.10364 confirms the official rate of 0.10032. The above calculations strongly contradict the suggestion of Young (2003) to reduce the reported growth rates of China’s real GDP and its non-agricultural components by approximately two and three percentage points per year respectively.

For another check I have selected a period in the economic development of Taiwan with similar growth rates as the reported growth rates for PRC during the period 1978-1998 in order to find out whether the outputs of the same products grew at similar rates. The period selected is from 1970 to 1989. National income data beginning in 1970 are reported in Statistical Yearbook of the Republic of China 1990 (Table 47). In the period of 1970-89, real GDP increased at an annual exponential rate of 0.091427 and manufacturing product increased at an annual rate of 0.10143. These rates are very similar to the corresponding rates for PRC of 0.09267 and 0.11079 mentioned in the first paragraph of section II.

The list of outputs of individual manufacturing products is unfortunately much smaller in the ROC statistical yearbook than in CSY. For all the seven products in both lists I have provided the rates of output growth in Taiwan in the column under Taiwan in Table 1. Of these seven products, PRC had higher growth rates in four, namely beer, color TV sets, cement and construction, and lower growth rates in three, namely paper and paper boards, electricity, and motor vehicles. Ignoring the production of beer for which PRC had a much higher growth rate, we find that PRC had a higher growth rate in the production of TV sets (and probably of other household consumer durables in view of their extremely high rates of growth reported earlier) but a lower growth rate in the production of automobiles. It had a slightly higher rate of growth in construction and a slower rate in the production of electricity. The rate of growth of output of automobiles in PRC, though lower than that in Taiwan, is close to the reported rate of growth for industrial output as a whole. Since construction has a much larger weight than electricity production in the computation of national output, the above limited comparison suggests that industrial output growth rates in PRC and in Taiwan were similar in the corresponding periods.

III. An Examination of the Estimates of Output Growth by Young (2003)

Young (2003) has raised questions regarding the accuracy of aggregate output data reported in CSY. He writes (pp.1227-9), “enterprises are called on to report the value of output in current and constant (base year) prices. The difference between the two series produces an implicit deflator, which is then used to deflate nominal value added (also reported by the enterprises)…Ruoen and Woo argue that many firms assume that the constant price value of output equals the nominal value, that is, the implicit deflator always equals one. This simplifying assumption has been taken by statisticians in other countries, and it seems likely that Chinese firms would find it to be a time-saving approach. More generally, enterprises might assume that the inflation rate has some constant value. Regardless, the assumption of a constant rate, be it zero or positive, will make the GDP deflators insufficiently sensitive in the underlying rate of inflation, such as has occurred during the reform period.“

His main point is that if we deflate nominal output by its implicit deflator to obtain real output, the estimated increase in real output is too large because the deflator underestimates the true inflation rate. By replacing the output deflator by another official price index for each of the three sectors as given in his Table 3 (using the farm and sideline products purchasing price index for the primary sector, using the ex-factory industrial price index for the secondary sector and using the service price component of the consumer price index for the tertiary sector), the official growth rates of real GDP and of the its non-agricultural components from 1978 to 1998 were reduced respectively from .091 and .106 to .074 and .081. The annual growth rates of real GDP according to the CSY estimate and alternative estimate are given in Figure 2 of Young (2003).

The most dramatic difference between the two estimates of GDP growth occurs in 1989 when the adjusted estimate shows a negative 5.2 percent and the CSY estimate shows a positive 4 percent. The negative growth is attributed (p. 1232) to “the forces that precipitated the political unrest of that year.” This estimate does not seem plausible for at least two reasons. First, there was no sign of significant economic disruption in the first five months of the year. Peaceful demonstrations did not start until April and the Tiananmen Incident occurred on June 4, 1989. According to CSY China’s real GDP was growing at 11.3 percent in 1988 and a reduction to the official rate of 4.1 percent in 1989 was a very large reduction. We can reasonably assume that for the first six months of 1989, real GDP was increasing at least at 7 percent annually as compared with 11.3 percent the year before. To get a negative growth of -5.2 percent for the entire year would require an annual rate of decline in output of -17.4 percent ([7-17.4]/2=-5.2) in the second half of 1989, which is highly unlikely. Even if we accept the adjusted rate of growth of 6.3 percent provided by Young for 1988 (as read from his Figure 2) and assume a 4 percent annual growth in the first half of 1989, to arrive at a negative growth of -5.2 percent for the entire year would require an annual rate of decline of -14.4 percent in the second half of 1989, which is also highly unlikely.

Second, to provide direct evidence of positive growth of industrial output and construction, we compute the ratios of outputs in 1989 to outputs in 1988 for the same list of 23 products as given in Table 1. The ratios are respectively (a) 1.139, 1.024, 1.007, 1.025, 1.050, 1.087, 1.031, 0.980, 1.032 (b) 0.885, 1.110, 0.789, 0.906, 0.785, (c) 1.073, 1.083, 1.036, 1.001, 1.158, 1.081, 0.905, 0.901 and (d) 1.035. As in section II we estimate the combined growth rate for groups (a) and (b) by the ratio of the total value of the 11 products (omitting three products lacking in price data) in 1989 to the total value in 1988, both evaluated at the 1987 prices given earlier. The respective values are 167.57 billion and 165.47 billion, giving a growth rate of 1.013. This growth rate for groups (a) and (b) combined, the mean growth rate of 1.030 for the third group (c) and the rate 1.035 for construction (which accounted for 12.3 percent of the secondary industry in 1988 as shown in Table 2 below) are consistent with the reported positive growth rate of 0.04 for GDP given by CSY. They strongly contradict the claim of a negative rate of growth of GDP by as much as -5.2 percent according to the adjusted estimate of Young.

I have attempted to reproduce the estimates of real GDP for 1988 and 1989 by the method of Young and to pinpoint where they might lead to inaccurate results. Table 2 summarizes the data and my reproduction.

[Insert Table 2.]

The resulting 5.15 percent decline in real GDP is made up of a decline of 4.14 percent in the primary industry, 6.84 percent in the secondary industry and 3.31 percent in the tertiary industry. I have pointed out in the last paragraph that a negative growth of 6.84 percent for secondary industry is a very unreasonable estimate. Furthermore, if we deflate the nominal value of construction 794.0 in 1989 by the ex-factory industrial price index of 1.186 (1988=1.0) to obtain its value 669.5 in 1988 prices, we obtain a rate of change of 1-0.8265 =-0.1735 for construction as given by Young’s method. This alleged negative 17 percent rate of decline in construction is highly inconsistent with the reported rate of increase in completed floor space by 3.5 percent. The calculation for the growth rate in 1989 pinpoints one specific large error resulting from using Young’s method. Therefore the method of Young (2003) for estimating the rate of change of real output can be very unreliable. In fact the large discrepancy between Young’s estimate of negative 5.2 percent and the official estimate of positive 4 percent for the year 1989 alone contributes to almost half a percentage point in the difference between the two estimates of the average annual growth rate for the entire sample period 1978-1998.

Table 2

Outputs and their Rates of Change in 1988 and 1989 by Young’s Method

GDP Primary Secondary Construction Tertiary

1988

Nominal value 14928.3 3831.0 6587.2 810.0 4510.1

1989

Nominal Value 16909.2 4228.0 7278.0 794.0 5403.2

Price index (1988=1.0) 1.1501 1.186 1.186 1.239

Output in 1988 prices 14173.7 3676.2 6136.6 669.5 4360.9

Output Change,1988=1 0.94945 0.95959 0.93159 0.82654 0.96692

Notes: Data on nominal outputs are found in CSY 1997 (Table 2-9). Price index for primary industry is purchasing price index for farm products (1978=100) given in CSY 1997 (Table 8-11) converted to (1988=1.0). Price index for secondary industry is ex-factory price index of industrial products given in CSY 1997 (Table 8-12). Price index for tertiary industry is the price index of the service component of the consumer price index (CSY 1990, Table 7-17). 1989 Output in 1988 prices for each of the three components of GDP is obtained by deflating nominal output by the corresponding price index. GDP in 1988 prices is the sum of the above three components.

____________________

VI. Conclusion

In this note I have utilized data on outputs of a sample of important industrial products and of construction to show that the growth rates of real GDP and output of the secondary industry during the period 1978-1998 as reported in CSY are reasonable and consistent with estimates obtained by examining the output growth rates of four groups of products included in the sample. The same statement applies to the reported growth rates in 1988-1989. These findings contradict the alternative estimates of the rates of growth for the periods 1978-1998 and 1988-1989 provided by Young (2003). I have also pinpointed places where Young’s method has led to serious errors.

In Chow (1985; 2002, pp. 90-91, 152-3; 2004, pp. 59-63) I have suggested that statistics reported in CSY are by and large fairly accurate. This note is a confirmation of that view. It shows that the performance of the Chinese economy in terms of an average growth rate of about 9.5 percent for GDP and about 11 percent for industrial output in the period 1978-1998 as reported in China Statistical Yearbook is reliable although the Yearbook may contain errors in reporting some other aspects of the Chinese economy.

References

Chow, Gregory C. “Chinese Statistics.” The American Statistician. 40 (August, 1985): 191-196..

Chow, Gregory C. China’s Economic Transformation. Oxford: Blackwell Publishers, 2002.

Chow, Gregory C. Knowing China. Singapore: World Scientific Publishing Co., 2004.

People’s Republic of China. State Statistical Bureau. China Statistical Yearbook. Beijing: China Statis. Pub. House, annual issues.

People’s Republic of China. State Statistical Bureau. China Commodity Prices Statistical Yearbook (in Chinese, Zhongguo Wujia Tongji Nianjian). Beijing: China Statis. Pub. House, annual issues.

Republic of China. Statistical Yearbook of the Republic of China. Taipei: Directorate- General of Budget, Accounting and Statistics, annual issues.

Young, Alwyn, “Gold into Base Metals: Productivity Growth in the People’s Republic of China during the Reform Period,” J.P.E. 111 (December 2003): 1220-1261.