There also appears to be a broad consistency between quantitative and qualitative data. Perkins cited a few examples to support this contention for statistics published prior to the Great Leap Forward.25 In an attempt to ascertain the changing pattern of spatial distribution of Chinese industry, I gathered data to derive estimates for the percentage increase in the gross value of industrial output between 1965 and 1973 for various provinces, autonomous regions, and municipalities.26 A rank test for the significance of the difference in these industrial growth rates between coastal and inland provinces suggests that industrial production in the interior grew more rapidly than in the coastal areas. A second test was made to measure the degree of relationship between two sets of ranks the ranks of various provinces and their equivalents according to the gross value of industrial output in 1954 and their ranks according to the rates of industrial growth during 1965-73. The rank correlation coefficient is significant at the 5-percent level. These two tests together suggest that industrially backward provinces, which are mostly in the interior, tended to develop more rapidly than more advanced provinces, most of which are located in the coastal areas. These findings seem to accord with the available information showing that the Chinese policy is to spread industry to the interior, and that in recent years state investment has gone up sharply in some of the industrially backward provinces.27

Proof of internal consistency of published Chinese statistics does not constitute proof of lack of outright fabrication. Could not the Chinese practice statistical falsification in the sense of what Prof. Abram Bergson in his discussion of Soviet statistics has called "free invention under double bookeeping?" 28 As in the Soviet case, the answer seems to be in the negative. There is no evidence to indicate that the Chinese maintain two sets of national statistics, one for planning and another for propaganda. In fact, several considerations point. to the contrary.

To maintain two sets of statistics would require a huge administrative apparatus and a high degree of statistical sophistication. The PRC inherited a weak statistical foundation in 1949, and encountered many difficulties in the establishment of a state statistical system during the 1950's.29 These difficulties were aggravated by the poor quality of the working force in statistics.30 The quality deteriorated further in the Great Leap Forward, and probably again was affected adversely during the Cultural Revolution. Since there are literally millions of basic accounting units in China, the supply of statistical personnel has never been able to meet the demand.31 It is difficult, therefore, to envisage how a vast country like China with a statistical working

25 Dwight Perkins, op. cit., pp. 220-221.

28 Nai-Ruenn Chen. "Industrial Development in China." Trade Analysis Division. Bureau of East-West Trade. U.S. Department of Commerce. Research Note 11. October 1974.

For example, state industrial investment in Ningsia Hui Autonomous Region in the 7-year period 1966 to 1973 was 85 percent above total state investments in the region in the 17-year period 1949 to 1965. (People's Daily, Aug. 26. 1973, p. 3.)

Abram Bergson. "Reliability and Usability of Soviet Statistics: A Summary Appraisal." The American Statistician, vol. VIII, No. 3 (June-July 1953), pp. 19-23.

For an excellent discussion of China's statistical system in the 1950's, see Choh-ming Li op. cit.

As pointed out by the State Statistical Bureau in May 1954, "an overwhelming majority of the national statistical working force never had any special training in statistics. lacking any knowledge in statistical work or economic construction ***" (Cited in Li, op. cit., pp. 51-52.)

31 Leo Orleans estimated that "the number of persons with high degree in statistics was very small probably not more than a few hundred a year obtained degrees between 1961 and 1966." (Orleans, op. cit. p. 50.)

force deficient in both quality and quantity could operate a double bookkeeping system on a national scale without detection.

If the Chinese Government were to compile a separate set of statistics for propaganda, those propagandized would have included the Chinese people on the mainland.32 The statistics contained in Chinese broadcasts to the areas outside the mainland such as Taiwan, Japan, Southeast Asia, and the Soviet Union, and in foreign-language publications such as Peking Review, China Reconstructs, and China's Foreign Trade have usually appeared also in domestic broadcasts and in publications for domestic circulations such as Jen-min jih-pao and Hung-chi. The Chinese Government relies heavily on the news media Jen-min jih-pao in particular, to disseminate information and directives throughout the country. False economic data would tend to mislead cadres at various levels in making economic decisions, and therefore would have disruptive effects on the economy.

The PRS appears to have adopted a policy of withholding unfavorable information and publishing mainly statistics showing success. As noted earlier, there was and still is a concealment of most of national production statistics for the 1960's when the economy suffered difficulties and disruptions. The industrial statistics published so far for the 1970's have covered mostly those industries with high growth rates, and relatively few figures have been released for those industries having mediocre performance. As Professor Li points out:

If it is true that only statistics of achievements are published while those reflecting difficulties and problems are either withheld or not collected at all, then there are not likely to be two sets of national statistics * * *33

This contention is also borne out by a check of official claims against data gathered outside China. For example, Chinese sources indicate that foreign trade turnover (exports and imports combined) in 1972 was 48.4 percent larger than that in 1965, and that 1973 trade turnover was 466 percent larger than that in 1952 and 159 percent larger than that in 1965.34 These claims are basically consistent with independent estimates constructed on the basis of the trade statistics of China's trading partners.35 Another example is Chou En-lai's statement that 1970 cloth output was 8.5 billion meters. This figure compared with cloth output figures published for the 1950's seems to be in general agreement with the outside information about the PRC economy such as raw cotton output and imports, cotton cloth exports, per capita consumption of cotton cloth, and the state of the synthetic fiber industry during the corresponding period.36

Based on the above considerations it may be concluded that Chinese statistics are not deliberately falsified by the central authorities. This conclusion, however, does not imply that deliberate falsification is not practiced by lower administrative echelons. On the contrary, falsifica

32 This point was raised in Perkins, op. cit., p. 216.

33 Choh-ming Li, op. cit. p. 149.

34 Li Chiang, op. cit.; Peking Review, Nov. 30. 1973; and Shu Hsuan. "Develop Foreign Trade by Maintaining Independence Keeping the Initiative in Our Own Hands and Relying on Our Own Efforts." People's Daily, Oct. 15. 1974, p. 6.

See Nai-Ruenn Chen, "China's Foreign Trade. 1950-74," in this volume; and Alexander Eckstein, "China's Economic Growth and Foreign Trade," U.S.-China Business Review, vol. 1, No. 4 (July-August 1974), pp. 15-20.

30 Rawski, op. cit.

tion at lower levels was quite extensive in the 1950's." The problem became so serious in the Great Leap years of 1958 and 1959 that agricultural output figures initially published for these years had later to be scaled down considerably.

The central authorities were fully aware of the imperfect reliability of the data submitted to them, and of the necessity to compile accurate statistics for economic planning. The State Statistical Bureau was seriously concerned over the problem, and soon after its establishment in 1952 steps were taken to improve the quality of statistical work. This effort was disrupted by the Great Leap campaign and then probably again by the Cultural Revolution. In view of the restoration of the economic planning mechanism in the last few years, a system of statistical quality control semblance to that of the mid-1950's may now be in operation. However, given the intense pressures on industrial and rural cadres to fulfill and over fulfill production targets, one may suppose that varying degrees of falsification probably still exist in the statistical reporting to the center by the cadres of state enterprises and

communes.

Since little is known about Chinese statistical system since 1960, one is not certain how much, if any, improvement has been made in statistical quality in the last 15 years. If the statistics published in the mid1950's can be used as a guide, some generalizations may be made of the comparative reliability of different types of available statistics on the PRC economy. Various economic sectors may be ranked according to a decreasing degree of data reliability: Industry, transportation and communications, trade, finance, construction, and agriculture. Data for state enterpises are more reliable than those for local enterprises and rural communes. Physical output data tend to be better than aggregative value data.

V. USABILITY OF CHINESE STATISTICS

To the extent that the central authorities fully realize the degree of possible falsification by lower echelons in collecting and reporting statistics and take corrective measures to improve them, and that the published statistics are in fact used by Chinese economic planners, these statistics may serve as first approximations in studying the PRC economy. However, the general usability of Chinese statistics is subject further to a number of other limitations, the most important of which are discussed in this section.

One of the most frustrating problems facing users of Chinese statisties is to find out what a given statistic means. One cannot use Chinese statistics meaningfully without knowing their definition, coverage, and classification. And yet the meaning of Chinese data is often left unexplained, especially for those published in recent years. Two examples may be cited.

37 Commenting on the reporting of agricultural statistics in the 1950's, Professor Li writes. "Reports were invariably readjusted by party and political leaders at every level of local government, where the conflict of interests had to be reconciled between the production bureau which tended to inflate the figures and the bureau of tax collection and internal trade which tended to do the opposite. Collectivization in 1956 and 1957 did not contribute any improvement to the mechanism of statistical reporting, but rather produced a predominantly upward bias on the part of local leadership in editing the figures." (Chohming Li, op. cit. p. 58.)

The Chinese claimed that the gross value of light industrial output in 1973 was up 8 percent over 1972.38 A number of claims were also made for the growth of light industry in various provinces and localities in the last few years. But no explanation is given as to the definition and coverage of light industry. In the 1950's the PRC published statistics on the gross value of output for consumer goods industry rather than light industry.39 It is not clear to what extent in Chinese statistics light industry is defined differently from consumer goods industry, or whether the two terms are simply used interchangeably. A similar vagueness may be found in a recent statement by Foreign Trade Minister Li Chiang. It is stated that "in export, the percentage of industrial and mining products has risen steadily, accounting for 65 percent in 1973." 40 Yet no explanation is available as to how "industrial and mining products" are defined in Chinese foreign trade statistics."1

Confusion and misunderstanding also result from changes in definition or coverage usually without appropriate adjustments to published figures or adequate explanations. For example, according to official statistics the yearend number of workers and employees in Chinese industry rose from 7,907,000 in 1957 to 25,600,000 in 1958.42 The more than tripling of the size of China's industrial labor force in 1958 was not solely due to rapid industrial expansion in this year of the Great Leap, but also to a significant degree attributable to a reorganization of handicraft workers into state factories. To take another example, the gross value of industrial output at 1957 prices was reported to have increased from 70.4 billion yuan in 1957 to 117 billion yuan in 1958.43 A large part of such increase, however, was statistical, and some of the statistical increase was due to a change in the treatment of the output of subsidiary agricultural activities. During the First Five-Year Plan period, the processing by peasants of agricultural products, such as grinding grains and slaughtering animals, and handicrafts produced by peasants for their own use, such as tailoring and making shoes and socks, were classified as a part of agricultural production under the heading of subsidiary agricultural output. Beginning in 1958 these subsidiary activities of peasants were shifted from agriculture to industry, and became part of the gross value of industrial output."

A third example is the shift in 1960 from biological yield to barn yield as the basis for measuring crop production.45 The established practice in China, at least up through 1958, was to measure crop production by biological yield. The discrepancy between biological out

38 Peking Review. Jan. 4, 1974.

Statistics on the gross value output of consumer goods industry were published for 1949 through 1959. See Nai-Ruenn Chen, Chinese Economic Statistics, table 4.37. 40 Li Chiang, op. cit., p. 1.

41 According to the Ten Great Years (p. 176), exports are classified into three categories: Industrial and mining products; processed products of agriculture and side occupations; and products of agriculture and side occupations. No explanation is given for the definition and coverage of these export categories. In 1958. industrial and mining products accounted for 27.5 percent of total exports. It is not known whether the rapid rise in the ratio to 65 percent in 1973 was purely the result of economic development or might be in part due to changes in definition and coverage.

42 Nai-Ruenn Chen, Chinese Economic Statistics, table 11.2.

43 Ibid., table 4.37.

Wuhan op cit.

Department of Statistics and Planning. Hupeh University. Industrial Statistics, Hupeh jen-min chu-pan-she, 1960, pp. 9-10; and Field, Lardy, and Emerson, 45 Huang Chien-t'o, "Major Tasks in the Agricultural Statistical Work Program for 1960," Chi-hua yu tung-chi (Planning and Statistics), No. 2, February 1960, pp. 18-21.

put and barn yield was not important prior to 1958, but became significant during the initial years of the commune movement when losses in the process of reaping, threshing, gathering, and storing were reportedly large.46 It is not known whether or not the practice has reverted to biological yield, and if not, whether the post-1960 crop output series has been adjusted to become comparable to those published for the 1950's.

Another aspect of the definitional problem has to do with the Chinese practice of grouping heterogeneous products into one and the same output series. The most infamous example is the inclusion in steel output statistics for 1958 and 1959 of a large amount of steel produced with indigenous methods and generally of unusable quality. Presumably such distortion, perhaps to a much lesser degree than the steel example, also exists in the production statistics published in recent years for chemical fertilizers, cement and other industries where the small-plant program has expanded rapidly. In physical terms, for example, small plants accounted for 54 percent of total nitrogenous fertilizer output 48 and 50 percent of total cement output in 1973.49 Yet the products of small plants are as a rule inferior in quality. Aggregating the products of both modern and small plants into the same physical output series as if they were homogeneous tends to inflate the output data and yield upward bias over time due to a gradual increase in the relative importance of small plants.50

Another limitation of published Chinese data arises from the way in which some of the statistics are presented. The Chinese usually show economic achievements in terms of indexes. The index number problem was well investigated by students of the Soviet economy. The major deficiencies inherent in the Chinese formula for estimating industrial production, which was adopted from Russian usage, has been discussed by Kang Chao 51 and by Field, Lardy, and Emerson.52 Among them are the following:

(1) In China, the index of industrial production for a particular period was derived as the ratio of the gross value of industrial output at 1952 (and later, 1957) constant prices of the given period to that of the base period. The use of gross value of output as the standard of measurement was likely to result in increased double counting as the complexity of manufacturing processes grew. The use of 1952 or 1957 constant prices would impart an upward bias to the production index since these prices tended to overstate the real prices in later. periods when output expanded.

(2) The Chinese practice of pricing new products at starting up prices and keeping their prices at the same level tended to inflate yearto-year increases in industrial production. The problem was particularly serious for those industries in which many new products were produced.

46 Choh-ming Li, op cit., pp. 94-96.

47 The Ten Great Years shows that crude steel output increased from 5,936,000 tons in 1957 to 13.690,000 tons in 1958. But 1958 output included 4,160,000 tons produced by native methods during the hectic backyard furnace campaign.

48 Peking Review, Jan. 18, 1974, p. 22.

49 Peking Review, Jan. 11. 1974, p. 23.

50 The contribution of small industry to the total output of nitrogeneous fertilizers increased from 12 percent in 1965 to 33 percent in 1968, 43 percent in 1971, and 54 percent in 1973. The share of small plants in cement output rose from 24 percent in 1965 to 27 percent in 1968, 40 percent in 1971, and 50 percent in 1973. (Peking Review, Dec. 8, 1972; Dec. 15, 1972; Jan. 19, 1973; Jan. 11, 1974; and Jan. 18, 1974).

61 Kang Chao, op cit. chap. 2.

52 Field, Lardy and Emerson, op cit.