Prewar Population Statistics

Luo Huanzhen

Is it possible to reestimate the prewar population?

The objective in compiling prewar population statistics is to estimate the population of China in the period prior to the Liberation in 1949 and, if possible, compile separate estimates for men and women, and for regions. It is desirable for the purposes of the Statistics Project and the framework of the Project's China Group to go back to population statistics of the 1870s or 1880s (when the modern industrialization movement ("Westernization movement") was beginning). However, as explained below, limitations of data mean that we had to begin our work with the Republican population.

China has a 5000-year history, and the history of population surveys extends back to the Qin dynasty in the years 221-206 B.C. However, those surveys did not cover the entire population, but only the so-called ding (in general, adult males). In 1712, the Qing government enacted the policy of shengshi tianding, yong bu jiafu (no taxes on newborn infants) and expanded the survey to the entire population. It is said that reliable population statistical surveys were conducted from that time until 1850. In 1851, the Taiping Rebellion erupted and the Qing government lost control of many provinces. Consequently, the system ceased to function, and surveys and reports on population were discontinued.

In 1908, the Qing government formulated a policy for conducting a population survey in six years as part of its preparations for becoming a constitutional body, but the outbreak of the Republican Revolution of 1911 meant that the complete population survey project could not be carried out. Later, demographer Chen Changheng collected the survey results from the end of the Qing era and estimated the total number of households in China at 71,264,000 and the total population at 368,245,000 persons. The survey of the late Qing Dynasty thereby became the first population estimate for China. This estimate forms the starting point for our reestimate of the population of modern China.

The Republic of China was established in 1912. For 38 years, until its collapse in 1949, the Republican government regarded population statistics as important, and conducted three nation-wide surveys, in 1912, 1928, and 1936. In addition, many researchers and private research organizations also conducted surveys and estimates of the Chinese population. As a result, a great deal of data on the Republic era population still remains (Table 1). A number of population censuses were planned but, for a variety of reasons, never executed.

However, an abundance of data does not necessarily mean that population estimates are easy to compile. Two of our experiences help to explain why. In beginning the task, we, the members of the China Group, including Prof. Minami, asked for advice from a large number of specialists, including Prof. Liu Ts'ui-jung of Academia Sinica, Taiwan, Prof. Ge Jianxiong of Fudan University, and Prof. Wu Chuang ping of Renmin University, but the unexpected reply was "It is best not to estimate the population of the Republic of China." The reason they gave was that proper surveys had not been conducted because of political and social unrest and warfare, so that while a considerable amount of data had been compiled, its reliability was low. Further, agreed the specialists, the data was mutually contradictory, so while reestimating it was not without merit, it would be too difficult. Upon starting to work on the gathering the surveys, we contacted the authority (self-styled) on Republican population, and conducted various negotiations on working together to compile estimates, but we were turned down in indirect fashion.

Upon learning this, some sentiment for abandoning the task emerged within the China Group, but population is the key data for measuring economic growth in modern China. We judged that it was essential, and decided to continue with the planned research.

Postwar Republican population research

Population data is indispensable for studying modern China's social and economic development. Many population estimates have been compiled by both Chinese and non-Chinese researchers, notably including the United Nations estimates of the 1950s. The most important efforts, in addition to the United Nations, include Ta-chung Liu and Kung-chia Yeh, Etsuzo Onoe, and Zhang Youyi. .

The above population estimates can be classified into two groups: the first group used statistics and estimates on the Chinese population from before the Liberation, and the second did not. Researchers in the first group used the 1953 census, the first conducted by the People's Republic of China, as a base, then revised and supplemented earlier population materials to produce population estimates. This method is called direct estimation. It is the approach used by Liu and Yeh and by Zhang.

The second group, on the other hand, did not trust the past statistical estimates and did not use them at all. This group, consisting of the UN and Onoe, used the 1953 census results as their base to estimate China's population and the rate of population growth from the beginning of modernization. Taking the 1953 census results as their base, the UN and Onoe used some of the surveys of birth and death rates to hypothesize birth and death rates for the period from 1910 through 1949, and to estimate China's population until 1911.

There are numerous problems concerning these estimates. The groups making direct estimates used and revised statistics compiled and published by government agencies, but the basis of the revisions is by no means clear. This problem is serious in the case of Zhang's estimates, and is worst of all in the estimates on the natural rate of increase of the population. First, it is pointless to hypothesize a uniform rate of population increase from the 1910s through the 1950s, as the UN does. Onoe's estimates hypothesize population fluctuations in different periods, but the basis of the hypothesis is not adequately shown. Further, while the method for estimating the rate of population increase is concise, there are many materials which they neglect to use. For example, materials on sex ratios and ratios of infants in the total population and life tables estimated from regional surveys are very useful for estimating the Chinese population.

New approximation methods

While there have been various problems in the research conducted on the Republican era population, as explained above, it is not impossible to compile estimates. While using the Republican era materials, a new method for compiling estimates is needed. We want to propose the following two methods. One means, which appears like the direct estimation methods, is to use the Republican government statistics together with the results of regional surveys and local population censuses. We can then search for omissions (of infants or of women) in population data which are believed to exist in the government data, and compensate for these. There are many regional surveys which can be very helpful in this respect.

The second method is based on the so-called Princeton life table. In the 1970s, demographers at Princeton University (hereafter called the Princeton group) adapted an indirect method for estimating population developed from the 1950s, and used it to conduct reestimates of the data collected by J. Buck in the 1930s. Their work confirmed that China's agricultural and village population prior to the Liberation was quite stable. The group also compiled a life table for the agricultural and village population.

The Princeton group did not estimate the Republican era population, but it is possible to reestimate it by using the Princeton life table. (Prof. Osamu Saito informed us about the table.)

Finally, let us briefly discuss the current state of our work and the main issues. In the fall of 1997, I experimented with using the two methods to reestimate the Republican population. However, there was a large differential in the two outcomes, indicating that the methods need further examination and adjustment. This is the first matter to be dealt with for now. The second matters concerns estimates of the Republican labor force based on population estimates. Because of data limitations, it is impossible to compile detailed estimates by industry, but I do plan to compile separate estimates for the agricultural and non-agricultural sectors.

Nihon University, College of International Relations


Minami Ryoshin, chief editor, Luo Huanzhen, "Republican Population: A Review." Hitotsubashi University, COE Discussion Paper No. D97-9, August 1997 (in Japanese).