Ethnic Mosaic of Modern China: 

An Analysis of Fertility and Mortality Data for the Twelve Largest Ethnic Minorities

There are major variations in fertility and mortality patterns among the ethnic populations in China, mainly owing to different stages of economic development

By Farhat Yusuf and Monica Byrnes^*

^*The authors of this article are members of the Demographic Research Group, Macquarie University, Sydney, Australia. This article is a revised version of a paper presented at the International Conference on China's Minority Population organized by the State Statistical Bureau of China at Kunming, 12-15 October 1993. They would like to acknowledge with gratitude the assistance of the China State Statistical Bureau for promptly providing the 1990 census data, as well as their colleague Sheng Yuming and graduate students Tian Gang and Zheng Shao-Mei for assistance in translation, and Craig Ellem for preparing the diagrams. Any views or opinions expressed in this article are the authors' own and do not necessarily reflect those of either Macquarie University or the State Statistical Bureau of China.

Modern China consists of a population with diverse social, cultural, religious and linguistic backgrounds. The uniqueness of these groups constitutes a mosaic of people who add to the ethnic diversity of the world's most populous country.

In recent years, interest in the study of China's minority nationalities has increased substantially, particularly since the end of the Cultural Revolution in the mid-1970s (see, for example: Eberhard, 1982; Heberer, 1989; Moser, 1985). Numerous studies have been undertaken, both nationally and internationally, utilising the 1982 census data for ethnic minorities (see, for example: Poston and Jing, 1987; Zhang, 1984). Such studies were not possible previously because of the paucity and the suspect quality of data from earlier censuses and population surveys.

The census identifies 58 ethnic groups varying in population size from a few thousand to more than 15 million. It was decided to limit the scope of this article to the 12 largest groups, all of which had populations of 2 million or more persons each. Data for the majority Han people and the total population of China have also been used for comparative purposes, wherever appropriate.

Data and methods

The data presented in this article were taken from the published first volume of the 1990 census tabulations (China State Statistical Bureau, 1993). This volume contains 34 tabulations pertaining to various aspects of the demography of the ethnic minorities in China. The tables and figures included in this article are based on data derived from some of those tabulations, except for the 1982 census data which were taken from the relevant census publication (China State Statistical Bureau, 1985).

No adjustments were made to the age and sex data. The 1990 census table 3.29 provides data on births registered over three six-monthly periods ending on 30 June 1990, so it was essential to estimate the annual number of births. For this purpose, two methods were considered to overcome problems related to recall lapse, as the accuracy of data is inversely related to the length of the reference period. The first method involved calculating two-thirds of the total births over this period. In the second method, births during six months prior to the census were doubled. When comparing the two methods, it was noted that, for the majority of groups, the estimated number of births using the first method was between 2 and 8 per cent higher, except for the Zhuang, Yao and Tujia, where the figures were 13 to 15 per cent higher. The difference in the two estimates for the Man was very slight, i.e. less than 0.2 per cent. For the Weiwuer, on the other hand, the estimate, using the first method, was 7 per cent lower. Although the first method is based on a relatively longer reference period and therefore subjected to a higher incidence of recall lapse, the estimates using this method, contrary to expectations, were generally higher. All fertility analysis in this article is based on data derived using the first method.

Unlike the data on births, mortality data were available for a 12-month period only. Although the quality of mortality data for six months prior to the census is considered to be more accurate, census table 3.33, which gives a cross classification of deaths by age, sex and nationality, provided only aggregated data for mid-1989 to mid-1990, and therefore, similar analysis as for fertility was not possible. The authors realize that the estimates of mortality derived from these data are somewhat under-estimated.

Although the estimated population on 1 January 1990 should have been used as the denominator for the fertility and mortality rates, actually the reported census population as on 1 July 1990 was utilized. This means that the fertility and mortality rates presented are slightly under-estimated.

Preliminary analysis of the data revealed some differences between the age distributions of various ethnic groups. In order to minimize the effect of these differences on birth and death rates, it was deemed essential to standardize fertility and mortality rates for various ethnic groups. Two standardization procedures, namely the direct and the indirect methods, were both used (Pollard, Yusuf and Pollard, 1990). The total population of China was taken as the standard when calculating the age-adjusted rates using the direct method of standardization. Age-specific fertility and mortality rates for the total population of China were taken as the standard in the computation of standardized fertility (or mortality) ratios using the indirect method of standardization.

Provisional life tables, for both males and females, were prepared for various ethnic groups, for the total population and for the total of all ethnic groups combined. No attempt was made to smooth any random fluctuations in the age-gender specific death rates derived from the appropriate 1990 census tabulations.

Size, growth and age structure

According to the 1982 census, the minority nationalities consisted of 67 million people and accounted for 6.7 per cent of the total population of China. The recently released figures from the 1990 census show that during the last eight years the population of ethnic minorities increased to more than 91 million and now accounts for 8.7 per cent of the national population. The 12 minority groups identified in table 1 account for nearly 85 per cent of the total population of ethnic minorities enumerated in the 1990 census.

The intercensal growth rate of the minorities was more than two and a half times that of the national average and about three times that of the Han people. Half of the selected minority groups had growth rates in excess of 4 per cent per annum. The highest rate observed was among the Man people, whose rate, if correct, indicates a doubling of their population once in every seven years. Some of these extraordinarily high rates could possibly emanate from the increasing tendency among the minority groups to reveal their ethnicity in the recent censuses. This phenomenon has been referred to as the "explosion of ethnicity", whereby people are now more ready to identify with their ethnic groups than they were in the past. With the increasing recognition of the role and importance of ethnic minorities by the people and Government of China, this is a plausible explanation, at least for part of the relatively high growth among ethnic groups other than the Han.

Figure 1: Proportionate age distributions for the total population of China and of 12 selected ethnic groups Source: 1990 Census of China

The higher growth rates among ethnic minorities are also substantiated by figure 1, which shows that in percentage terms there were many more younger people (0-24 years old) in the ethnic groups than in the total population, whereas the reverse was true for those aged 25 and older. This is also corroborated by the age data presented in table 1 .

The age-specific sex ratios for all of China and for the 12 ethnic groups combined are shown in figure 2. It appears that up to age 60 sex ratios were very similar and that there was an excess of males over females. At age 60 and older, the reverse was true. This is usually because of the increasing gap between the longevity of males and females. Although the quality of age reporting among the Han people is known to be generally good, this may not necessarily be true for the ethnic minorities. For example, substantial misreporting of ages among the older people in Xinjiang has been noted (Coale and Shaomin, 1991).

Figure 2: Age-specific sex ratios for the total population of China and 12 selected ethnic groups Source: 1990 Census of China

The overall sex ratios of the Han people and of the ethnic minorities, as a whole, were quite similar (table 1 ). There were, however, some noticeable differences in sex ratios among the 12 ethnic groups. Five of these groups showed significantly higher sex ratios while two groups (Mongol and Zang) revealed a slight excess of females over males. It is not clear whether the excess of females in these groups is because of substantially higher male mortality or the exodus of males to other parts of China. These fluctuations in sex ratios could also be due, at least partly, to the sex differentials in the completeness of census enumerations among various ethnic groups.

Minority nationalities in China live in border or mountainous provinces (Poston and Jing, 1987). Data presented in table 2 affirm this except for three ethnic groups, namely the Hui, Tujia and Zang, which are quite scattered. For the remaining nine ethnic groups, 50 per cent or more of their population is concentrated in one province. The Weiwuer is the least segregated of the groups. Almost all of them are situated in Xinjiang Autonomous Region. They are followed by the Zhuang, with 91 per cent in Guangxi Province. The most geographically separated are the Hui, with only 31 per cent in Ningxia Autonomous Region and Gansu Province; the rest are scattered all over China. Some authors have argued that, unlike many other minority nationalities, this is probably because the main distinguishing characteristic of the Hui is their religion, Islam, otherwise in terms of their physical and other attributes they are similar to the majority Han people (Gladney, 1991; Heberer, 1989; Moser, 1985).

In 1990, the crude birth rate (CBR) among the Han was about 20 per thousand and the average rate for all ethnic minorities was nearly 24 births per thousand population per annum. Thus, overall the CBR of the latter was 20 per cent higher than that of the former. Table 3 shows that the CBR of various ethnic groups ranged between 19 and 34. The lowest rate was observed among the Man people, which was even lower than that of the Han, and the highest rate was among the Weiwuer: a CBR in excess of 34.

Because of the inter-group differences in the age distributions, two standardized measures of fertility were calculated: the standardized fertility ratio (SFR) and the age-adjusted birth rate. The SFR is defined as the ratio of the actual to expected number of births, the expected number being estimated under the assumption that the age-specific fertility rate (ASFR) for the total population of China (used as the standard) would prevail in each ethnic group. On the other hand, the age-adjusted birth rate is the expected CBR in the standard population if it experienced the ASFR of a particular ethnic group.

The SFRs for the Tujia and Mongol people were statistically not different from those of the total population (figure 3). The Man people had an SFR which was lower than that of the standard population. The highest SFR was recorded for the Weiwuer, whose fertility level was twice that of the total population.

Figure 3: Standardized fertility ratios by ethnic group

The age-adjusted birth rates given in table 3 corroborate the above-mentioned findings. It may be noted that these rates were generally higher than the crude birth rates, which is mainly because of the effect of younger age structures of the ethnic groups vis-a-vis the standard. Exceptions were the Man and Mongol people whose age-adjusted rates were lower than their crude birth rates.

The total fertility rate (TFR) for women of the Han majority was 2.1, which declined from a figure of 2.6 recorded in the 1982 census (Zha and Ji, 1984). This decline may largely reflect the effectiveness of the national family planning programme. For the minority nationalities overall, the TFR was significantly higher at nearly three children, with the Weiwuer recording the highest at about five. Each of the other groups, except the Man, had a higher TFR than that for the whole country, thus pointing to a relatively much higher potential for population growth among the non-Han segments of China's population.

Initially, minority groups were, for the most part, exempt from the population policy in force for the rest of the population. However, in 1982 authorities agreed that they should be involved in the national family planning programme and by 1984 stated that the one-child family should be encouraged for minorities of one million or more people, while other groups were allowed to have two children per family. Moreover, according to a directive issued in 1985, minorities with one million or more people were requested to conform to the one-child policy (Gladney, 1991). In fact, in 1986, only two-thirds of the minorities adhered to family planning policies compared with nearly 100 per cent of the Han (Heberer, 1989).

Apart from the knowledge and use of contraception, many other factors could account for the observed fertility differentials between the Han and various minorities. At a macro level it is relative social and economic development which is important, while at an individual level, factors such as age at marriage, literacy and religion are crucial.

In his analysis of marriage patterns from the 1987 One-per-cent Survey, Zhang (1990) noted that the age at first marriage among a number of minority groups was substantially lower than that of the Han. This obviously has implications for fertility and mortality patterns, as well as the growth rate of the populations being studied. Moreover, Xiong and Yang (1989) also noted substantial ethnic differentials in fertility in the 1982 census, particularly among the younger Yi women. They also noted that remarriage rates of ethnic minorities were higher.

When comparing the proportion of women over 15 years who were literate, major differences were found between the Han population and the minority groups (table 3). Just over two-thirds of Han women were literate, whereas for women of all ethnic minorities this figure was 58 per cent. The Man and Mongol women represented a very high literacy rate of 84 and 77 per cent, respectively. In contrast, low rates were noted for Miao at 41 per cent, Yi at 31 per cent and a very low 18 per cent for Zang women. The literacy rate among Weiwuer seems unusually high and needs further investigation. A comparison of 1990 figures with those of the 1982 census would be useful to assess any changes in literacy levels.

The census does not provide information about religion. The fact that some minorities can be identified with certain religious groups means it is possible to study the relationship between religion and demographic behaviour. For example, although both the Hui and Weiwuer nationalities are followers of Islam, in terms of their social and economic characteristics the Hui are probably closer to the Han than to the Weiwuer. Similarly, the fertility levels prevalent among the Hui were not substantially different from those for the Han in spite of the fact it has been claimed that "the Hui are often allowed to have at least 1 child more than their Han neighbours" (Gladney, 1991). On the other hand, the fertility levels observed among the Weiwuer were significantly higher compared with the Han and all other ethnic groups. This negates the hypothesis that religion is an important determinant of fertility levels prevalent among ethnic minorities in China.

The overall levels of mortality, expressed in terms of the crude death rates, were substantially higher among ethnic minorities compared with the Han (table 4). Significant mortality differences were also evident within the various ethnic groups. For example, the Man experienced the lowest mortality rates, even lower than those of the Han, while the highest mortality was noted among the Zang and Buyi people. The age-adjusted death rates were, in most ethnic groups, higher than the crude death rates, reflecting the effect of differences between the age-sex distribution of the standard population and those of various ethnic groups.

The gender-specific standardized mortality ratios (SMRs) reveal two very interesting points (figure 4). Firstly, they show that the mortality levels among the Zhuang and Hui were not significantly different, while those of the Man were significantly lower and those for other groups were significantly higher compared with the standard. Secondly, there were practically no gender differences in SMRs for three ethnic groups, the Zhuang, Man and Hui, which were characterized by mortality rates similar to or lower than those of the standard, while for the majority of the remaining groups, female mortality was substantially higher. This may have been due to higher levels of maternal mortality resulting from unhygienic conditions and lack of modern child-birth facilities (Xiong, 1989).

Figure 4: Standardized mortality ratios by gender and ethnic group

The infant mortality rate (IMR) for the whole of China in 1990 was 28.7 per thousand live births. The corresponding figure for 1981 was 38.6 per thousand (Wang, 1984), which means a decline of nearly 26 per cent over the intercensal period. Among the ethnic groups studied (figure 5) the level of infant mortality was substantially higher, except for the Man whose IMR was lower and the Hui and Mongols who experienced rates similar to the national average. The highest rate was found among the Zang, followed by the Buyi, Weiwuer, Yi, Dong and Miao & Aring; all showing rates nearly double those of the Han people.

Figure 5: Infant mortality rates by ethnic group: China

Infant mortality is a good indicator of the level of social and economic development, and is affected by factors such as the availability of public health facilities, particularly maternal and child health services; adequacy of nutrition; education; and cultural practices regarding the value of children. Because of the close relationship between fertility and infant mortality observed in many developing countries (CICRED, 1975), it is possible that the higher fertility levels among ethnic groups in China could be explained at least partly by the relatively high infant mortality experienced by these groups.

The comparison of life tables for the total population, based on the 1982 census (Jiang, Zhang and Zhu, 1984) and the provisional life tables prepared by the authors for the 1990 census (table 5), reveals an increase in life expectancies at birth of 2.7 years for females and 2.2 years for males. Provisional life tables for all ethnic minorities combined, based on 1990 census data (table 6), show that the average difference in life expectancy at birth between ethnic groups and China as a whole was 4.1 years for females and 3.4 years for males. In the absence of corresponding life tables for 1982, it is not possible to comment on improvements in the mortality of ethnic groups over the intercensal period.