SPECIAL ISSUE ON CHINA

Prewar Manufacturing Production

Fumio Makino


D.K. Lieu and P.S. Ou. There is probably not a single scholar among modern Chinese economic historians who does not know these names. Embarrassing though it is, however, I have to admit that I, having previously studied Japan, did not know their names at the start of the Asian Historical Statistics Project. My level of understanding of Chinese economic history was such that when I was asked to take charge of "Manufacturing Output" for the Statistics Project, I hadn't a clue on how to begin or how to proceed.

To start with, I conferred with my partner, Prof. Guan Quan of Tokyo Metropolitan University, to decide whether to put our main efforts in compiling statistics on the period from the establishment of the People's Republic of China in 1949, or the period prior to 1949. We decided to choose the latter. Almost all reliable statistics from 1949 are monopolized by the State Statistical Bureau, meaning that we would have to purchase materials from the Bureau in order to compile the data, so that our contribution would probably be limited to refining the statistics/data to some degree. On the other hand, it seemed much more likely that we researchers in Japan could make an original contribution by focusing on the prewar era (the Republic of China era), an approach which would enable us to start by gathering the materials, difficult though that might be.

Having decided to focus on the prewar era, we faced the problem that we had no idea what statistics existed or where they were stored. But at that difficult moment, a providential event intervened. Prof. Kubo Toru, an expert in modern and contemporary Chinese economic history at Shinshu University, joined our group. To recap, the Manufacturing Statistics Group actually set to work in June 1996 when he presented the report on the survey of manufacturing statistics and estimates for the China Republic period. I came to know of the scholars whose names led off this paper when their names were introduced to me at the start of the project work.

Now let us examine the statistics themselves. There are seven sets of nationwide industrial statistics (Table 1) and three representative sets of estimates (Table 2) for industrial production output for the China Republic era.

Both Ou and Liu & Yeh estimated manufacturing output value and value added as part of estimates of GDP. The main source of raw data for these works was the Report on the Survey of Manufacturing in China (hereafter, RSMC) listed in Table 1. The estimates by Ou were produced at the Central Research Institute, a research agency directly attached to the republican government which worked hard on compiling estimates of national income, leading to the compilation of National Income of China, 1933. Considering that it was produced in the mid-1940s, this is a remarkably complete work. Anyone compiling statistics on the Republic of China era must consult this work, whose coverage is not limited to manufacturing industry.

As can be understood from the above, the three-volume RSMC, a study of the manufacturing statistics of the Republic of China era compiled and edited primarily by D.K. Lieu, is the best source of statistics in terms of reliability and coverage. The study was conducted by trained investigators who actually visited the factories and conducted interviews with the factory managers. Apart from plants which were foreign-owned, or located in remote provinces or the northwest regions under Japanese control, the survey seems to have covered nearly all the factories in China.

Thus, it was natural that we also make 1933 our benchmark year. However, having only one year as a benchmark is not reliable. So we obtained the time series data for the 1910s and used Statistical Tables on Agriculture and Commerce (hereafter STAC), which encompasses all manufacturing industries. In this manner we used 1933 and 1912-21 as benchmarks, and formed a basic plan for linking them to other materials and extending the estimates.

In order to conduct the work, we created the following division of labor. Makino verified the 1933 manufacturing output statistics, principally by scrutinizing and revising RSMC while Prof. Guan scrutinized and revised STAC; and Prof. Kubo gathered and investigated other documents and materials. The tentative results of their revisions (which should be regarded as preliminary versions) came out in winter 1997 in two Japanese-language discussion papers. These are (with titles translated) Fumio Makino and Toru Kubo, "Estimates of Chinese Manufacturing Output in 1933" and Guan Quan, "An Estimate of Production Value in China's Manufacturing Industry in the 1910s: An Evaluation and Revision of Statistical Tables on China's Agriculture and Commerce."

Next let us discuss the process of compiling the estimates, starting with the gathering of materials. The first and second (of three) volumes of RSMC and Ou's National Income of China, 1933 are both held in the Hitotsubashi University Institute of Economic Research library, albeit on microfilm. The library holds about half of the yearly editions of STAC, and since Tokyo Metropolitan University and the Toyo Bunko also hold copies, it was possible to assemble a complete set within Japan. The Toyo Bunko has a microfilmed copy of the third volume of RSMC, obtained from the Library of Congress in the US, but we found that two pages were missing. We eventually got a copy of the pages from the library of the Institute of Economics of the Chinese Academy of Social Sciences. We obtained about 20 other items regarding economic statistics on trips to the Institute of Economics of the Chinese Academy of Social Sciences, the Institute of Modern History, also of the Chinese Academy of Social Sciences, the Nankai University Institute of Economics, the Tiangjin Library, the library of the Shanghai Social Science Institute, and the Liaoning Province Library (we benefited from the energetic assistance of Luo Huanzhen and Hao Renping in gathering the materials).

This is an aside, but I had a couple of interesting experiences during a visit to a library in Beijing in October 1997. First, I noticed in the Foreign Visitors' Register in the front office that almost all of the signatures were those of members of the Statistics Project's China Group. I had the illusion of having somehow come to Hitotsubashi University. And the other experience. In China, a librarian has strong discretionary power, and it is vital, especially for foreigners, to get along with him (or her) in order to be able to gather materials. For that reason, I gave the librarian an "omiyage," or gift (the nature of which I leave to the reader's imagination). The splendid manner in which he/she received the gift, never making eye contact, left a deep impression upon me. Needless to say, the miraculous effect of the omiyage was just as described.

Back to the main topic. In order to examine the data from RSMC, we adopted the approach of dividing workplaces, apart from industrial and regional characteristics, into those with 30 or more employees and those with fewer than 30 employees, and then to compare production output on a per capita basis. In more than a few cases, it could be seen that output per worker in the workplaces with fewer than 30 employees exceeded that of workplaces with 30 or more. We assumed that the production output for the under-30 workplace group was too large and decided to correct it. The production output that we obtained in this manner became the industrial production output for the modern factory sector. Since hand manufactures, foreign-owned plants, and factories in the Manchurian region are not included in RSMC, we used different estimates (especially Ou's) or obtained different materials and produced our own estimates (for details, see the discussion paper by Makino and Kubo).

The most difficult task was Prof. Guan's revision of STAC. If you look at the contents of the volume the statistics seem to resemble those of Japan's Statistics Tables of the Ministry of Agriculture and Commerce (Noshomusho Tokei Hyo) of the prewar era, but in fact the data consists of answers to written questionnaires which the central government in Beijing had prefectural governments gather, and they are no more than simple aggregates of the findings. Clearly there were many cases when the factories, concerned that the government would learn about managerial conditions or demand higher taxes, did not provide accurate information. Thus, there are, as we will explain below, many doubts regarding the reliability of the surveys. Further, the area controlled by the central government was limited, in addition to which the area covered by the surveys gradually contracted. In 1921, the final year of the survey, only six out of the country's 22 provinces were covered.

One feature of the reported figures is quite surprising. Taking as an example one column from the relevant data sources, output of cooking oil in 1912 was 2 billion yuan, but the following year it was 16 million yuan, and in the year after that 1.8 million yuan. This illustrates part of the reason why STAC could not be used up to now. We would like everyone to understand what a difficult job Prof. Guan undertook in revising the STAC's figures.

Finally, let us briefly discuss our next research projects. First, we need to upgrade the statistical estimates we have collected so far. It is necessary to render the research results for the various classifications for 1933 compatible, especially for agriculture, population, prices, and trade. Judging from the scale and complexity of the problems in STAC, it will not be possible to finish the corrections this time. We would also like to announce versions 2 and 3 in order to test the multiple revisions made using the ratio indices for productivity, unit price, and other factors. Further, since STAC includes agricultural as well as industrial production output figures, the Agriculture Group will probably use the same statistics. Therefore, the Manufacturing and Agriculture Groups will need to exchange information on conducting revisions.

Second, there are the estimates for the 1920s and from the period after 1933. Since there are very few materials for the 1920s, it seems that we will have to rely heavily on trade statistics and on Chang's estimates in Table 2. For the period from the mid-1930s, we can use the materials from sections 5., 7., and 8. in Table 1. Similarly, for the 1920s, we can probably use trade statistics and Chang's estimates.

The third task is the compilation of value added estimates. In order to estimate value added, materials on costs are necessary, and these do, fortunately, appear in RSMC. Further, by using 7. we can find out the costs of raw materials. We believe that value added can be estimated by using these sources.

Tokyo Gakugei University, Faculty of Education