SPECIAL ISSUE ON CHINA

Estimating National Income Statistics for Post-Revolution China

Masaaki Kuboniwa


The most important objective of my subgroup in the Asian Historical Statistics Project's China Group is estimating China's gross domestic product (GDP) for the post-Revolution era. Estimating GDP for this period is relatively easy compared to doing so for the prior period. This is because socialist China, seeking to execute its nation-wide planned economy, fundamentally redesigned the entire statistical environment and continuously prepared national income statistics. This is without doubt a decisive advantage in compiling statistics compared with the pre-Revolution era when no official national income statistics existed. Nevertheless, even with the existence of the post-Revolution GDP statistics, compiling the estimates necessitates far more effort than originally foreseen. If there were real time series we could manage, but had there been only nominal time series and just we Japanese researchers would have given up in despair. Still, that is not to say that we needed only to join up with some Chinese scientific researchers to compile the estimates. To complete work of any significance, it was essential to enlist the cooperation of China's State Statistics Bureau, which holds a monopoly on all statistics, not to mention many any other statistics related to them. For that reason, we believed that it would be best to avoid a roundabout approach by compiling the estimates in direct cooperation with the State Statistics Bureau.

Fortunately, our idea nicely matched the thinking of the financially strapped Chinese side in a timely bit of luck which smoothed the way to the two sides conducting joint research. Underlying the idea of conducting joint research were ongoing social and economic changes. China is encouraging high-speed marketization of the national economy, whose integration into the global economy is proceeding rapidly as well. As a result, not only has there been international and technical cooperation in investment areas, but steady progress is also being made in upgrading statistics to match international standards. Moreover, China has been the fastest of the countries in transition, past or present, to complete its current statistical estimates, and has done so for the period stretching back to 1978; further, both the government and the State Statistics Bureau have even had some leeway for compiling estimates back to 1952. As Japan belongs to the kanji-influenced cultural region, Japanese-Chinese technical cooperation is easier than might be imagined, and our case was no exception. For example, I speak almost no Chinese and in school my comprehension of classic Chinese literature was poor, yet I can understand almost everything I read in this specialized area. The same holds for Chinese reading Japanese. When reading reports in Chinese on recent input-output tables, for instance, it is easy to understand that a considerable number of Japanese documents have been utilized.

Underlying the smooth realization of cooperative research between the Chinese and Japanese sides was another factor. Many Chinese have come to study in Japanese graduate schools, and the number of Chinese employed in Japan as faculty members is not small. Our research group organized a number of these scholars to assist us from the start of the project. To begin with, Dr. Jinjun Xue, who taught at Hitotsubashi University and is presently professor at Oita University, served as an intermediary between myself and Mr. Qiang Li, the head of the Department of National Accounts of the State Statistics Bureau, helping make it possible to quickly reach accord on the joint research. In addition, Ximing Yue, a young scholar who had just completed his Ph.D. at Hitotsubashi and another valuable addition to our research team, applied himself tirelessly to resolve issues between the two sides. During the course of this work, I have seen reaffirmed more than once the substantial depth and importance of cultural exchange between Japan and China.

The results of our joint research were published just over half a year ago in Tokyo in an attractively produced English-language monograph, The Department of National Accounts of the SSBC and The Institute of Economic Research of Hitotsubashi University, The Historical National Accounts of the People's Republic of China, 1952-1995 (Tokyo, September 1997). In addition, the entire text and data set contained in the monograph have been made available, free of charge, on the internet. Thus the results of our research have become part of the public domain. As the results of our joint estimates are now public data, the basic estimation data series have been published in Chinese as well, while descriptions of the methodology for compiling the estimates and the supplementary data used for the estimates can be accessed only through the public domain. An article on the monograph appeared two issues ago in the Newsletter (No. 7, 1997). In addition, we have already received comments on our research from many persons, beginning with Professor T. J. Rawski. In the next section, I will outline the research and briefly explain its differences and distinctive features compared to previous research.

The main features of the estimating process

(1) The basic approach to compiling GDP estimates was to use the old material products system (MPS), reconstituted and supplemented, to first compile nominal base national income (the "Nationale Einkommen" of Marx's Das Kapital was translated into Russian as "natsuonalnyi dokhod" and then became Kukuzei-Syuunyuu (national income) in the Chinese translation). Thus national income can be rendered in several terms in Chinese. These include, in addition to "national income," "national income produced," "production approach national income," "net material product" (NMP), and "NMP produced" (alternatively rendered as "NMP by origin"). The material sectors include mining and manufacturing; construction; agriculture and forestry; transportation of goods; production-related communications; commerce (domestic commerce and foreign trade), agricultural requisition, and dining establishments; and a number of other material sectors (such as hunting; printing and DPE; and so on). The non-material sectors include administration, finance, insurance, and real estate; education, culture, science, medicine, insurance, and health; passenger transportation; individual communications; and so on. NMP is derived only from net value added created by material sectors.

(2) The first step in converting NMP into GDP was to supplement the statistics of the material sector's fixed capital consumption (amortization or depreciation of fixed capital). As the State Statistics Bureau had not previously released these statistics, we had already decided that we needed to have them made available. This data is now published in the monograph described above. The negotiations with the Chinese side over opening up the data had reached an impasse, but Professor Yoshiro Matsuda and I had both logged considerable experience in researching MPS and are deeply interested in this area, so we were persistent in breaking down opposition and bringing the talks to a successful conclusion. During the Cultural Revolution, it came to be forbidden to think in terms of depreciation. The position of the State Statistical Bureau, taking into account the point on depreciation, was that revisions to the NMP statistics for the Cultural Revolution era were completed. We tentatively agreed and pressed on with the research. Fixed capital consumption as listed in SNA 68 and SNA 93 is listed in the account books not as simple amortization but as reproduced value. This is the same as in Japan, so we did not concern ourselves with this point.

(3) The second step in converting NMP into GDP was to supplement gross value added for the non-material sector. The necessary data for this task was monopolized by the Chinese side so we used it.

(4) The third step is estimating inputs of non-material services into the material sector, and then deducting these from NMP. This task necessitates subtracting as well as adding, and it is easy to understand if one thinks of it in the following manner. Whether MPS or SNA, the gross domestic output (gross social output) of the material sector is the same (total production of electric power or machinery are also used to supplement MPS).

      MPS/gross social output

      = SNA/gross domestic output (for the material sector)

    By definition:

      MPS/gross social output = intermediate material input + depreciation + NMP (for the material sector)

      SNA/gross domestic output = intermediate material input + non-material intermediate input + GDP (for the material sector)

    From the three above equations we derive

      GDP = NMP + depreciation - non-material intermediate input (for the material sector)

which becomes the guiding equation.

(5) Estimating nominal base gross domestic expenditure (GDE). The Chinese call GDE "GDP on expenditure side" according to "SNA 1993". The central steps in this task are estimating final consumption expenditures of non-material service expenditures (for adding) and switching over from non-material intermediate inputs to MPS final demand portions (for subtracting). In either case, compiling estimates without the cooperation of the State Statistics Bureau would be quite difficult. At first, the Chinese side was obstinate about compiling nominal base and real base GDE, but as we had extensive experience in compiling long-term economic statistics for Japan, we were determined to prevail on this point, and finally did so. In this instance, some hard work performed by Assistant Section Chief Xianchun Xu of the State Statistics Bureau proved very useful. Thanks to the improvements he had made in GDE estimate formulae since 1990, the Chinese side was aware of the importance of GDE estimates, and as a result came to appreciate the importance of past statistics as well.

(6) To estimate real base GDP and GDE, we basically continued to use the compilation formula for real value NMP series (a conveniently available chain of five constant price series) (this was because we were unable to use such information as detailed sectoral data on the mining and manufacturing industry). However, we combined a number of approaches, such as using physical indexes (people-kilometers, ton-kilometers), as I proposed, for the transportation sector. This was not the ideal method, but it was made necessary by the limitations of the data.

(7) Whether derived from nominal base or real base, there are limitations to the compatibility of the GDP series and GDE series obtained from the processes described above. Further, time series can provide adequate supplemental information on the role and importance of the service sector in the economic growth process. However, numerous problems remain. First of all, we have compiled estimates for financial intermediation services indirectly measured (FISIM) according to SNA, but from a diachronic viewpoint this is not satisfactory (some of the results are included in the monograph). Second, the work was hindered by some continuing problems with the NMP statistics. For example, it does not take account of exaggerated industrial output figures which resulted from the incentive system of the state-owned companies (the former Soviet Union was afflicted with the same problem). I have written some papers on under-estimates of industrial production and GDP in new Russia since its birth, and, using Rawski's works as a reference, would like to conduct research on the overestimates which occur in China. We are not all satisfied with the revised statistics of the Cultural Revolution era, stunted as they are by the revised NMP. This is because our estimates are no more than a provision of baseline series with various revisions made. However, it is a fact that with regard to clarifying the meaning of the substitute estimates, no progress could be made in the talks without first having the baseline. In the case of China, the differentiated data for exports and imports under SNA present another problem. In this area as well, we were unable to achieve satisfactory results. However, the very strange formula used in the former Soviet Union for computing foreign trade revenue was not adopted by the Chinese, so we do not expect that they will have much effect on the results of the estimates for GDP and GDE.

Our estimates compared with two previous works

Among previous work on estimating Chinese GDP, two in particular merit attention. These are Angus Maddison, Chinese Economic Performance in the Long Run (OECD, forthcoming) and Harry X. Wu, "Reconstructing Chinese GDP According to the National Accounts Concept of Value Added" in The Industrial Sector, Comparative Output, Productivity and Purchasing Power in Australia and Asia (COPPAA) Series No. 4 (Center for the Study of Australia-Asia Relations, Griffith University, 1997). While Maddison estimates the entirety of GDP, Wu covers only the mining and manufacturing sectors. Maddison compiles separate estimates for different industrial sectors, but for agricultural GDP estimates he relies mainly on FOA data on production output and mid-term inputs, and for mining and manufacturing GDP he uses Wu's data. Wu calculates mining and manufacturing GDP by multiplying the 1987 value added ratio by total mining and manufacturing output for every year according to the 1987 price indicators. He obtains his basic data from 1987 industrial input-output tables and the Yearly Industrial Economic Statistics (in Chinese) (both compiled and edited by the State Statistics Bureau and published by China Statistics Publishing Company). As statistical research, these works have considerable significance, but the quality of their basic data is greatly inferior to ours. The absence of joint research imposes limitations on one's results. Whatever substitute estimates are employed, there is an upward bias in mining and manufacturing statistics, but the substitute estimates appeared useful as references when we rechecked the joint baseline estimates. We did at any rate succeed in presenting baseline time series. For the remaining work, when detailed internal components of the mining and manufacturing sector and the agricultural sector are more critically researched, the extent to which GDP series deviate from baseline series will be quantitatively verifiable. This means using more detailed sector-specific data for factors such as taxes and subsidies. By means of such research, we will be able to achieve a deeper understanding of China's economy. For this reason, we have to maintain a little distance from joint research with Chinese authorities. However, essential detailed data for the years before 1978 are held not by the National Accounts Department but by the Industrial Statistics Department or by the financial authorities. New strategies will be needed to clear such hurdles.

Hitotsubashi University, Institute of Economic Research