One result of the Chinese publication policy of creating the mostfavorable impression possible is, as has been alluded to earlier, the tendency to withhold statistics reflecting difficulties or poor performance. The statistical drought during the 1960's is a dramatic case in point. Most of the statistics published in recent years mainly reported achievements. This can be seen from table 2 which presents the recently published data on the production indexes for various industrial products with 1965 as the base year. With a few exceptions, these indexes were higher than the average growth rates as represented by the indexes for the gross value of industrial output. Therefore, a number of industries not included in table 2 must have experienced lower than the average rates of growth. The fact that no quantitative information has been made availble for the output of these industries is indicative of the policy of withholding information on mediocre performance or failures. Such statistical concealment added to the paucity of qualitative data would make the study of certain sectors of the PRC economy utterly impossible.

The usability of Chinese statistics is limited further in several other ways. As has been mentioned many times throughout this paper, recently published statistics were frequently given in percentages without absolute figures for the base. In some cases, even the base year was not precisely specified. For example, it was merely referred to as "the previous record year," "the early liberation period," or "the last several years." Sometimes figures were shown in imprecise terms such as "approximately," "nearly," "less than," "more than," et cetera. Not infrequently, the growth rates of output were published in terms of ranges.

Occasionally discrepancies existed in the published figures. Some illustrations may be helpful. In the January 5, 1973, issue of Peking Review, it was reported that retail sales of TV sets and transistor radios in 1972 rose 50 percent and 80 percent respectively over 1971. Ten months later the same journal gave the respective figures as 56 percent and more than 100 percent.55 The Kiangsu Provincial Service broadcast from Nanking on July 4, 1973, reported that the gross value of industrial output of Kiangsu Province rose by 150 percent between 1965 and 1972,56 but the same broadcast on September 29, 1973, indicated an increase of only 140 percent.57 It is conceivable that as was frequently the case in Chinese statistics published in the 1950's the latter figures could represent an adjustment of the preliminary estimates released earlier. But there were also cases in which discrepancies existed in the figures for the same indicator released at about the same time but by different sources. One example is provided in the reports on the increase in the unit yield of early rice in Yunnan Province from 1972 to 1973. It was 23 percent according to the New China News Agency dispatch in English from Peking on August 11, 1973,58 but only 12 percent according to the Yunnan Provincial Service broadcast in mandarin from Kunming on the following day.59 The gap must have been due to errors rather than adjustments. A number

55 Peking Review, Oct. 12, 1973.

56 FBIS. No. 129, July 5, 1973, p. C1. 57 FBIS. No. 190, Oct. 1. 1973. p. C4. 58 FBIS. No. 157. Aug. 14. 1973. p. B4. 59 FBIS. No. 159, Aug. 16, 1973, p. E3.

of other similar instances may be found.60 In general, however, such reporting errors appear to be less frequent in national statistics than in local statistics.

All in all, Chinese statistics are fraught with methodological deficiencies and other pitfalls. Figures published in China, therefore, cannot be taken on face value, and should be used with great care. After a decade of statistical secrecy, economic researchers on China welcome recent increases in the flow of quantitative information. With a lot of guesswork and some bold assumptions, the bulk of the recently published data could be put to interesting and meaningful uses. Field's article on national product estimates in a recent issue of the China Quarterly is a case in point.61 Rawski's study cited above is another. With national statistics still scarce, provincial data could provide a useful source for research. Some macro-economic studies could be carried out through aggregating provincial data, as demonstrated in the Field-Lardy-Emerson study and in the work being done in the Economic Research Service of the U.S. Department of Agriculture. Some studies of regional economies also could be done on the basis of available provincial data, as exemplified by some papers in this volume.

VI. SUMMARY AND CONCLUSIONS

The sources of Chinese economic information stem from both inside and outside China. The amount of such information available has varied over time. During the 1950's qualitative and quantitative economic data were relatively abundant. Subsequently, however, they dwindled in amount until the end of the 1960's. During this period the Chinese government adopted a policy of statistical secrecy while only isolated scraps of qualitative information were published. Since 1970 the publication policy seems to have been relaxed, but the amount of statistics released is not as plentiful as that in the 1950's. Some of these released statistics are relative numbers without an absolute figure for the base, and certain types of data, which were available for the 1950's, have been withheld from publication.

The quality of Chinese data varies for different periods of time and for different economic sectors. In general, Chinese statistics are believed to be internally consistent in view of the close correlation, as demonstrated in a number of studies, between aggregate and disaggregated data, between figures published for different periods of time, and between quantitative and qualitative data.

Internal consistency of published Chinese statistics does not appear to be the result of outright fabrication. There is no evidence to suggest that the central authorities practice deliberate falsification under double bookkeeping. Several considerations indicate the contrary. It is difficult to envisage how a vast country like China with a statistical working force deficient in both quantity and quality could maintain

00 One instance involved industrial output statistics for Shanghai. It was reported in the New China News Agency Domestic Service broadcast in mandarin from Peking in July 1973 that in Shanghai between 1965 and 1972 the output of 35 horsepower tractors increased by 10 times and the output of hand-guided tractors by 3 times. But these increases were given as 11 times and 3.5 times respectively in the NCNA dispatch in English from Peking in August 1973. (FBIS, No. 135, July 13, 1973, p. C5; and No. 166, Aug. 27, 1973, p. C7).

61 Robert Michael Field, "Some Overlooked National Product Data," The China Quarterly, No. 59, July/September 1974, pp. 583-86.

two sets of national statistics, one for planning and the other for propaganda, without outside detection. A double bookkeeping system would certainly cause confusion among cadres at lower levels and mislead them in making economic decisions. The Chinese Government seems to have adopted a policy of selective publication, withholding information on unfavorable developments and failures, and, therefore, deliberate falsification under double bookkeeping would not be necessary. Also, some of the published Chinese statistics appear to be consistent with the information derived from non-Chinese sources.

Statistical fabrication, however, was and probably still is, practiced by lower administrative echelons. This has affected the quality of Chinese statistics even with corrective measures taken by the central authorities. The general usability of Chinese statistics is limited further by a number of methodological deficiencies and certain peculiar practices.

One of the major difficulties with Chinese statistics has to do with the definition of concept, coverage, and classification. They are often left unexplained. To the extent that they were explained, they were sometimes defined in a peculiar way. Frequently, definitional changes were made without corresponding corrections to the data. In some of the time series data, widely heterogeneous components were grouped together and treated as if they were homogeneous.

Chinese official indexes suffer from many defects. Major defects of the industrial production index, for example, relate to, among other things, the use of gross value output, the choice of price weights, and the treatment of new products. These methodological defects are aggravated further by the tendency to use a low base in order to provide the most favorable impression possible. For the same purpose, statistics reflecting poor performance are usually withheld. For example, the growth rates of various industrial products, for which data were published in the last few years, were as a rule higher than the growth rates of total industrial output.

The usability of Chinese statistics is also hampered by the imprecise way in which the figures were released. Terms such as "nearly," "less than" and "more than" are frequently shown in connection with the published figures, and the base year for indexes is sometimes not precisely specified. Occasionally, more than two different figures were published for the same indicator, and usually no explanation is given as to the discrepancies.

In sum, Chinese statistics in terms of their availability, reliability, and usability are fraught with problems and difficulties. From the vantage point of 1975, the First Five-Year Plan (1953-57), particularly the latter half of it, may be viewed as a golden period for Chinese statistics, although in those years published Chinese data were by no means plentiful or of high quality by the standards of advanced countries. Being kept in statistical darkness for a decade, economic researchers on China in the 1960's were nearly desperate as they reached the point of no return. The gradual resumption of some statistical outflows from China since 1970 has opened new research possibilities. To be sure, there are many deficiencies and pitfalls in Chinese statistics, particularly those published in recent years. With patience, care, and ingenuity one could construct certain meaningful estimates on the basis of the statistics that China has so far made public.

CHINA'S POPULATION: CAN THE CONTRADICTIONS BE RESOLVED?*

By LEO A. ORLEANS

At the opening session of the United Nations World Population Conference in Bucharest in 1974, Huang Shu-tse, leader of the Chinese delegation, stated that "in the 20-odd years since the founding of the People's Republic, China's population has increased nearly 60 percent from about 500 million to nearly 800 million." This is not the first time Peking spokesmen have referred to "nearly 800 million," but because there is something magical about a figure presented at such a prestigious international conference, it has, for all practical purposes, been accepted by the world community as the official figure for China's population. There are, after all, few alternatives and the general assumption is, of course, that "they" must know and would not make such a statement casually. To those of us who have been more intimately concerned with studying China's population, however, the "nearly 800 million" is no more significant than were the 650, 700, or 750 million figure previously and sometimes interchangeably used by the Chinese. It adds little to our knowledge and is worthless in helping us either make or select alternative projections and estimates. In the light (or obfuscation) of some of the recent Chinese statements regarding population, this paper will attempt to address some of the contradictions and respond to the query: "Where are we now?"

SOME INDISCREET QUESTIONS

If we eliminate the modifiers-"about," "nearly," "close to❞—that precede Huang's figures, we do indeed get a growth rate for China's population of exactly 60 percent. But where did Huang get the figure of 500 million? In Ten Great Years (and elsewhere) Peking reported the end of 1949 population of some 540 milion for China proper.1 Could "about 500 million" refer to 540 million? Not likely, since a 60-percent increase would then produce a total population of 864 million, rather than "nearly 800 million." Is it not possible that the Chinese revised their 1950 population downward and now consider 500 million more accurate than 540 million? This is again unlikely. It would mean that they have discarded the 1953 census figure of 582.6 million and all other population data published since then. One is therefore led to conclude that either the population is not "nearly 800 million" or that it did not increase by 60 percent since 1950.

*My since gratiutude to John Aird for his helpful comments; this in no sense implies his concurrence with the analysis in this paper.

1 Ten Great Years, State Statistical Bureau (Peking: Foreign Language Press, 1960). This source reports a population of 548,770,000 including Taiwan. The figure of 540 million is derived by subtracting the population of Taiwan (7,591,298 in 1953) and rounding.