The primary contribution of the 1990 census to the study of
fertility in China at the national and provincial level is to
provide fertility statistics for the late 1980s. The level and
trend of fertility in China for earlier years are well known from
earlier work, including the 1982 census (State Statistical Bureau
1985) and one per thousand survey (China Population Information
Centre 1984; Coale and Chen 1987; Feeney and Yu 1987), the 1987 one
percent survey (State Statistical Bureau 1988; Feeney, Wang, Zhou
and Xiao 1989; Luther, Feeney and Zhang 1990), and the 1988 two per
thousand survey (State Family Planning commission of China 1991;
Lavely 1992; Feeney and Wang 1991).
While the period on which the census provides new information
is short, great interest attaches to the result, for the level of
fertility was rising during the mid-1980s (Luther, Feeney and Zhang
1990; Feeney and Wang 1991). Will the census show the level of
fertility continuing to rise in the last years of the 1980s, or
will fertility level off or decline? How may the observed changes
be explained?
Results of the 1990 census are available in several sources,
most importantly in the 1991 publication of 10 percent sample
tabulations (State Statistical Bureau 1991). This paper is based
both on these 10 percent tabulations and on a one per thousand
sample tape of census records.
The ten percent sample tabulations show a total fertility rate
of 2.25 children per woman for calendar year 1989 (calculated from
the single year age-specific birth rates given in State Statistical
Bureau 1991:469, Table 10-16). This represents a significant
decline from the peaks reached in the late 1990s. Birth histories
collected in the 1988 two per thousand survey showed a total
fertility rate of 2.57 children per woman for calendar year 1987
(Song and Li 1991:1). Reconstructed birth histories from the 1987
one percent survey show a rate of 2.46 for the year ending on 30
June 1987 (Luther, Feeney and Zhang 1990:350). The plot shown in
Fenq (1991:10) and the time series given in Luther, Feeney and
Zhang (1990) show that these values represent an increase over lows
reached following the 'high tide, of birth planning work in 1983
(Hardee-Cleaveland and Banister 1988: Table 2 and passim; see also
Zeng 1989).
This indication of resumed fertility decline in China during
the late 1980s is based on a direct question on births in
households during the three six month periods prior to the census
date (30 June/July 1990). Questions of this type have been used in
censuses and surveys throughout the world, and the general tendency
has been to more or less severe under reporting of births. They
have been used in China before with very good results, to be sure,
but it would nonetheless be unwise to let conclusions on so
2
important an issue be decided by reference to a single statistic
whose accuracy is even slightly in doubt.
In any case, the census data are capable of yielding far more
information on fertility change. The original own children method
yields annual age-specific fertility rates for 15 years prior to
the census (Cho, Retherford and Choe 1986). Birth history
reconstruction, an extension of the own children method developed
by Norman Y. Luther (Luther and Cho 1988), allows us to apply the
whole panoply of techniques for analyzing birth histories,
including the calculation of period parity progression ratios and
mean birth intervals (Feeney 1983; Feeney and Wang 1991).