Learning Center
Plans & pricing Sign in
Sign Out



									‘Fertility histories of the Great Leap Famine’

1. Introduction

While the Chinese Great Leap Famine was widely discussed and debated before the
1980s, I would argue that it only became a subject of scientific enquiry twenty years
after it had taken place. It was only upon the publication of data from the 3rd Chinese
National Census conducted in 1982, through the demographic work on the history of
the Famine conducted by people such as Judith Bannister and Ansley Coale. This
early work on the demography of the famine examined national-level data. In the
early 1990s the Shanghai-based demographer Peng Xizhe published more detailed
analysis of the demography of the famine period, including analysis of provincial
trends in mortality, fertility and migration trends during the famine. Since then, the
publication of county-level gazetteers and increasingly open access to archival data
has enabled some analysis of demographic trends through the famine at the sub-
provincial level, and here we might mention the study of excess mortality in the
Famine conducted by Cao Shuji, which utilises county-level demographic data
gleaned from local gazetteers.

I will argue today that there is still a lot that the demographers can tell us about the
Famine by utilising the censuses and large-scale demographic surveys conducted in
the 1980s, in particular the 1982 National Census and the several fertility surveys
conducted between 1982 and 1988. It would be nice if we had open access to the
household registration data held in the archives of the Public Security Bureau, which
we can assume has a reasonably complete if not perfect record of births, deaths and
migration through the Famine period. Unfortunately, I have not seen any evidence
that this may be possible in the near future.

Simple analysis of the relative size of cohorts tabulated in the 1982 National Census
can provide us with a measure of the relative severity of the Famine in administrative
precincts of various sizes. In the national data, the cohort born in 1960-61 is about
half the size of the cohort born in 1957-58, and less than half the size of the cohort
born in 1962-63, and this ratio varies significantly from province to province. In
Henan, for example, the 1960 cohort is a quarter the size of the 1963 cohort. This is
an indication that the Great Leap Famine was much more severe than the Henan
famine of 1943, for the 1943 is just one third less than the cohort born after this
famine had passed. The 1982 Census can also be used to glean information on which
age groups were particularly-severely affected by the Famine, and even some
information about migration trends.

The main advantage of using the 1982 Census data as a measure of famine is that
analysis can be conducted down at a very detailed level. With the help of county-level
tabulations of the 1982 Census data which were published in book form in the mid-
1980s, most of which are available at the Chinese University of Hong Kong, we can
conduct this type of analysis right down to the level of the commune for many
counties, or even below the commune to the level of the brigade. The main problem
of using the data from a single census to investigate demographic trends of the famine
period is that it does not distinguish clearly between different types of famine events.
There appears to be a pattern in the Chinese data that fertility in urban areas is more
sensitive to a drop in food availibility than in rural areas. For example, in Nanyang
district of Henan, which by Cao Shuji’s calculation of excess mortality was the
second most severely-affected by the Famine, the diminuition of the 1960 cohort is
remarkably similar than the average for Henan as a whole. Within the district, the
ratio for each county varies between 1/5th and 2/3rd , reflecting the hierarchy that we
would expect from the available mortality data, and if we go below the county we can
find a similar range of variation.

If we are to use the Census data to its full potential, we have to be able to distinguish
between excess mortality and fertility loss. This is the main problem in calculating the
excess mortality caused by famine between censuses conducted before and after the
event. In particular, assumptions made about infant and early childhood mortality,
information that is poorly recorded in the first two national censuses conducted in the
PRC, have a massive effect on the demographic estimates of excess mortality. This is
the main reason behind the variation in estimates published by demographers of
excess mortality in the Chinese Great Leap Famine, which range between 15 and 40
million. The fertility surveys conducted in China between 1982 and 1988 provide
with a rich source to distinguish between population loss caused by fertility loss,
excess mortality and migration.

The fertility surveys carried out in China were the largest-scale fertility studies ever
conducted. The largest of these, the 1988 two-per-thousand fertility survey, collected
detailed birth histories and information on family planning, household structure,
migration and economic background from over 600,000 women. Simply filling in the
8 part questionnaire provided full-time employment for 12,000 professional staff for a
month. The reason that the Chinese Communist state devoted such massive resources
to understand fertility patterns was to provide support for the One Child Policy,
arguably the number one policy objective of the Chinese state in the 1980s. The
survey has primarily been used by demographers to shed light on the process of
fertility transition that has accompanied the implementation of family planning in
China since the 1970s. With a number of notable exceptions, such as work by Peng
Xizhe mentioned above and William Lavely, very little use has been made of these
surveys to shed light on the demographic and political history of the early period of
Communist rule, including the famine period. I will talk about some of the trends
through the Famine that can be found in one of the fertility surveys, and introduce the
type of micro-analysis that the surveys make possible.

2. Fertility surveys of the 1980s
There were three main fertility surveys conducted in China in the 1980s, and they all
have their advantages when it comes to analysing demographic trends through the
Famine. The first was the One-Per-Thousand Fertility Survey conducted in
conjunction with the 1982 Census. A sample of 300,000 women from all provinces of
China between the ages of 15 and 67 were interviewed. The questionnaire for this
survey was relatively simple, recording time of birth of each child but not details on
the timing of death of infants. The main advantage of this survey is that it was
conducted early and included older women in the sample, allowing us to reconstruct
provincial fertility trends going right back to the mid-1930s, whereas the later surveys
only allow analysis of trends from the early to mid 1950s. Other advantages of this
survey is that the administrative boundaries used to record the location of women in
the survey are the same as those of the 1982 Census, which is convenient when it
comes to mapping trends from the survey and comparing them with other trends
shown in the census data. The design and questionnaire was based largely on the
World Fertility Survey, which means that the One-Per-Thousand data can be readily
compared with trends from forty other countries around the world.

The second fertility survey was the Chinese In-Depth Fertility Survey conducted in
two stages in 1985 and 1987. This was conducted in a third of China’s provinces with
a sample size about half that of the One-Per-Thousand survey for each province.
Phase II of this survey, conducted in 1987, which is what I have used for this paper,
interviewed samples from one city and five provinces, namely Beijing, Liaoning,
Guangdong, Shandong, Gansu and Guizhou. It so happens that the severity of the
Famine in this selection of provinces is evenly spread, with Beijing and Liaoning
lightly affected and Gansu and Guizhou amongst the worst-affected provinces. I’ll
talk below about two of these, Liaoning as a place were the famine came light and
even and late, and Gansu as one of the worst-affected provinces with a wide
divergence in severity from district to district. The age of the women interviewed in
this survey ranged between 15 and 49, which means that for the surveys conducted in
1985 the oldest women interviewed were born in 1936. The questionnaire was much
more detailed than that used in the One-per-thousand survey, including a range of
information on the socio-economic and household context of each women
interviewed, as well as more detail on their individual birth histories, such as the age
of death for any children who had died. Only the individual birth histories are
available from Princeton. As I will show below, the main problem in using the In-
Depth Fertility Survey to look at trends during the famine is that there were very few
women included in the survey who had established a personal fertility pattern prior to
the onset of famine. The information on the famine recorded in the survey comes
mainly from women who had their first child during or immediately after the famine,
and it is clear that the effect of the famine on this young cohort was not representative
of its effect on the general population of women.

The In-Depth Fertility Survey was conducted as a pilot study for the Two-Per-
Thousand survey conducted in 1988, the largest and most elaborate of the surveys and
the one most often used by demographers. The Two-per-thousand survey interviewed
women aged between 15 and 57, meaning that the oldest women included in the
survey were born in 1931 and were reaching the high fertility stage at the time of the
establishment of the PRC. The method of sampling used in the two-per-thousand also
makes it more interesting as a source for historical analysis. While in the One-per-
thousand and In-Depth surveys a random sample of women was taken in a random
sample of brigades in a random sample of communes in a random sample of counties,
the basic sampling unit of the Two-per-thousand survey was not the individual but an
entire village sub-committee consisting of 30-odd households. We therefore have in
each of these village sub-committees the fertility history of suburban neighbourhoods
and natural sub-units of the village, and can study more complicated social questions
such as marriage patterns through the famine and the intergenerational effects of the
famine on fertility within a or within a village.

The next two sections of this paper provide a commentary on the trends shown in the
In-Depth Fertility Survey sample collected in Liaoning and Gansu provinces in 1987.
2. Fertility trends reflected in the Liaoning In-Depth

Monthly trends
The In-Depth survey recorded the date of birth of the women interviewed and their
children, living and deceased. It also recorded the age of death of deceased children.
From this we can construct a noisy chart of month by month trends in fertility and
infant mortality in Liaoning.

[Chart 1. Monthly trends, Liaoning]
The yellow line in the chart above shows the annualised monthly fertility rate of the
women sampled in Liaoning. The noise element is greatest to wards the left-hand side
of the chart, were our sample of women of child-bearing age in 1956 is very small. It
does not show an unambiguous seasonal pattern, but the fertility rate does regularly
peak towards the end of the year, which is the main seasonal pattern we would expect.
I don’t know that any clear trend is shown between 1957 and 1959, but for 1960 the
fertility rate is clearly lower, and falls to a sustained low level throughout 1961. The
fertility level then jumps five in the summer of 1962, reflecting the effect of the 1961
autumn harvest as it worked through the natality cycle.

The black columns show a very spotty infant mortality pattern. In many cases this
represents a single infant death. While the bars are highest during the famine period,
there are also a lot of gaps between the bars were the monthly infant mortality rate of
the sample was zero, so I don’t think we can say anything about the monthly infant
mortality trend from the Liaoning sample.

[Chart 2. Partial fertility rate of different age brackets, Liaoning]
In the 1950s, women reached peak fertility in their late 20s, and our sample starts
with women born in 1937, so eldest cohort of the sample was only 21 at the beginning
of the Great Leap Forward. This means, sadly, that we don’t have very good trends
for the 1950s. Luckily women started having children young in those days, and we
can see that the number of children had by women aged 14-20 fell by more than half
between 1957 and 1960, and returned back up to the 1957 level by 1963. It also
shows more clearly the effect on fertility of the mini-Great Leap of 1966, and most
clearly what the survey was designed to show, which is the drop in fertility with the
introduction of family planning in the early 1970s.

If we strip out the older cohorts, and look just at what 14-23 year old women were
doing during the famine, we can also say something more concrete about infant

[Chart 3. PFR for young women and infant mortality, Liaoning]

Firstly, the basic trend is as we would like it to be, showing that peaks in infant
mortality corresponded to troughs in fertility, as in 1961. The sample has a
disproportionately great number of women for the early years, and as infant mortality
is higher for younger (and therefore more likely first time) mothers, the data here
overestimates infant mortality for the 1950s. This sample shows a rising infant
mortality rate from 1956, but I would not state that this was the case for the province
overall as our sample of births in the mid-50s is very small. Once again, the data is
much more robust by 1966, when it shows a substantial spike in infant mortality
before dropping down in 1968 to a level that remained more or less constant through
to 1978.

[Chart 4. Cumulative fertility by cohort, Liaoning]
If we look at cumulative fertility, we can see that the Famine had different lasting
effects on different cohorts.

We see the famine drove a wedge between women who were 18 and women who were
19 at the onset of the Great Leap Forward. The elder group widened their lead over
the younger group in total number of births through the first half of the 1960s,
suggesting that they were in a better position to make use of the improved nutritional
environment after the Famine. Put another way, the chart suggests that the Famine
had a permanent debilitating effect on the reproductive health of women who had or
tried to have their first child during the famine. Perhaps in a different political world
the younger group might have tried hard and have caught up with elder group in their
old age, but the implementation of family planning in the 1970s prevented them from
doing so.

And one last chart for Liaoning. If we look at the fertility trend through the 1950s and
1960s of women of a given age year rather than following the histories of cohorts
born in particular years, we get the clearest indicatio of the effect of the Famine on

[Chart 5. Fertility of women of a given age, Liaoning]

If we start down the bottom of the graph, we can see that one in a hundred 15 year-
olds gave birth in the 1950s, and again for a short time immediately after the Famine.
Again its a pity that our data only kicks in at the time of the Famine, for it would have
been nice to know what women at the peak of their reproductive powers were doing in
the leap up to the Famine. We can see clearly from what was happening to 19 year
olds that the famine caused a massive fall in fertility in the three consecutive years
after the Great Leap harvests. To the right of the graph we can see a trend of
childbirth being pushed back later in life, as fewer 21 yos give birth while a consistent
number of 23 yos give birth.

3. Fertility trends reflected in the Gansu In-Depth Fertility Survey

[Chart 6. Monthly trends, Gansu]

The yellow line in the graph above, showing the annualised monthly birth rate of the
women interviewed in the sample, shows a much more severe fertility drop in the
famine year than in Liaoning. The monthly fertility rate for the sample even drops to
zero for December 1960, showing that none of the 6000 women in the Gansu sample
gave birth in this month, nine months after the worst spring hunger of the famine
period. Interestingly, there are very few infant deaths recorded for 1960 and 1961, a
reflection partly of the very low birth rate during this period.
[Chart 7. Partial fertility rate of different age brackets, Gansu]

The partial birth rate of the Gansu sample is much higher than for the Liaoning
sample, over 50% higher for women under 35 in the early 1970s, and double for the
under 23 age group for which we have data in the early 1960s. For this age group of
women, the fertility rate dropped by half during the famine

[Chart 8. PFR for young women and infant mortality, Gansu]

Whereas in the Liaoning sample, the calculated year-to-year change in the infant
mortality rate moved consistently in the opposite direction to the change in the
fertility rate, this is not the case for the Gansu sample. We can see that for the famine
years the infant mortality rate does trace out a mirror image to that the fertility rate,
as reflected by the youngest age bracket of women, the infant mortality rate did not go
down as the fertility rate jumped back up immediately after the famine. The infant
mortality rate is higher than that recorded for Liaoning but still surprisingly low, less
than 150 per 1000 throughout the period of the famine.

[Chart 9. Fertility of women of a given age, Gansu]

If we look at what women of a given age where doing in each year, we can see that
the fertility rate of 19 yo fell by a third during the famine and never recovered, a sign
the famine corresponded with (or caused?) the onset of demographic transition in
Gansu. This chart shows clearly that fertility levels remained very low for three years,
whereas in Liaoning the fall was sharper but for only one year.

[Chart 10. Fertility of women born in a given year. Gansu]
Most of the charts we have looked at so far have shown the fertility of women who in
a certain age in a given year. If we instead track the fertility history of the same group
of women born in a given year, in a given year, we get a slightly different story. Here
we can see that the famine was experienced quite differently for different cohorts of
mothers. Those born in 1937 reached a high fertility level prior to the onset of famine,
and saw their fertility level when they were 22 drop back to where it had been when
they were 18. For women born in 1943, motherhood came late in life and with a bang,
with only one in twenty having a child when their eighteenth year, and one in three
having a child in their nineteenth year. It is also notable that this older cohort who
had not experienced motherhood during the Famine were also better able to
withstand later shocks, while the fertility of the older cohorts shown here also
experienced a drastic drop in fertility in the mini-Great Leap of 1966.

[Chart 11. Infant mortality by mother’s cohort, Gansu]
The information on infant mortality in the Phase II survey is very patchy. The group
of interviewees who were of birthing age in the 1950s is very small, and of these
women the number who lost a child born in a given year in the 1950s is only a

The main information revealed in this graph is that the sample of women interviewed
in the 1987 survey was too small and too young to know with any confidence. Infant
mortality is much higher for first-time than second-time mothers. As young mothers
are disproportionately first time mothers, and so we have a general trend that infant
mortality falls rapidly for women born in a certain year as they get older, which can
be seen with the grey and black lines. Our number for 1954 is entirely reliant on 17
yo mothers, and as it turns out the infant mortality rate shown here of 111 is derived
from a single infant death out of 9 born. For the Famine years we have a sample size
of between one and two hundred births and a dozen deaths, rising to a sample of a
few dozen deaths after the Famine. Given this small sample size, it is surprising that
the Gansu survey data still does show a consistent trend across women born in
different age brackets. Younger mothers experienced a consistently higher incidence
of infant mortality, with the oldest age bracket of women experiencing an infant
mortality rate about a third lower than those five years younger than them in the late
1950s and early 1960s. However, the overall trend is unclear. Did infant mortality in
Gansu really rise between 1954 and 1955, echoing the trend shown in official
mortality data for the period? Or is this just a reflection of the high sampling error?
Does infant mortality respond in the same way as fertility during a famine? This
might not necessarily be the case, if the famine were to have a disproportionate effect
on malnourished women who already had a high incidence of infant mortality,
preventing them from giving birth and shifting the burden of motherhood towards
relatively well off mothers.

William Lavely calculated infant mortality in Sichuan based on the 1988 two-per-
thousand survey, which had a sample of women who were of child-bearing age in the
1950s ten times the size of the Gansu Phase II survey. This sample data showed a
clear trend of rising infant mortality during the famine, but again with a lot of noise,
and the trend in infant mortality in the lead up to the famine is not at all clear.

Even if the fertility survey data cannot provide us with an unambiguous pattern of
infant mortality through the 1950s, there is still a lot we can learn from them about the
famine if we search not for representative trends but instead take the 6000
interviewees not as a representative sample but as 6000 case histories. I will do now
have a look at the case histories of the women from one County of Gansu that appears
to have been particularly badly-affected by famine who first gave birth in the 1950s.

4. Case histories from one commune

I will now give a quick summary of what happened in one county during the famine,
as recorded in the Gansu sample of the Phase II Fertility Survey conducted in 1987. I
will call this county as County 7, as it is referred to in the digitalised portion of the
survey questionnaires available from Princeton. It may be possible to locate where
this county was on a map at a later date, with the help of the full codebook of the
survey kept by the Family Planning Office. I have chosen to look at County 7 because
the sample includes areas classified in 1987 as both town and countryside,1 and
because the fertility and infant mortality trends evident in the data show that it was
severely affected by famine.

The fertility survey in County 7 interviewed between 30-40 women from 2 urban
street committees and two rural communes. I will call these Street Committee 1 and 2,
and Commune 1 and 2. The age distribution of interviewees is relatively consistent
across the four precincts, suggesting that the interviewers were controlling for age.
The size, age and basic fertility data of the County 7 sample is given below:

      Table 1. In-Depth Fertility Survey Phase II: sample of County 7 in Gansu

                                       SC 1    SC 2    C1      C2
Number of interviewees                 38      33      36      33
Average age of mother                  33.1    35.7    36.7    37.5
Interviewees with at least 1 birth     32      33      34      33
Total no. of births                    70      90      125     120
Average no. of birth events            1.84    2.73    3.47    3.64

The average age of the interviewees in the city precincts is slightly younger than in
the village precincts. The average age of Street Committee 1 three years younger than
Street Committee 2, which when taken alongside the lower average number of births
of the interviewees in Street Committee 1 suggests that it was the more intensely
urbanised of the two urban precincts.2 We can also see that the rural interviewees had
a substantially higher average number of births than the urban interviewees.

Only 1 of the interviewees in the two urban precincts had a child before or during the
Famine period. This reflects the fact only 1 of the urban interviewees had reached the
age of 15 by 1958, the older child-bearing age of urban women and perhaps the
suppressing effect of the Famine on fertility of young women in particular.

The first baby born to an urban interviewee was a boy born to a 17 yo mother in
October 1959, who weighed a low 5 pound and died the day after birth. The second
urban baby was another boy born to the same women in February 1962, who weighed
6 pound and survived through to the time of the survey. Two further babies were born
later in 1962, both girls weighing 5 pound who also survived. With these few lines we
have said about all we can say about the experience of the Famine in the urban
precincts of County 7.

One further feature of the urban survey worth noting is the story of the first two
women to give birth of Street Committee 1. These women gave birth to a boy in late
1959 and a girl in early 1963, both of whom weighed only 5 pound and died within a
week of being born. These two women subsequently had five and four further
children respectively, all of whom survived, making them the two most fertile women

  The fertility survey data classifies rural and urban areas into three categories of city,
township and village. The number of interviewees from ‘townships’ is very small.
  Further evidence for this is that 3 of the 38 women interviewed in Street Committee
1 spent their childhood in a rural area, as against 7 of the sample of 33 in Street
Committee 2.
in Street Committee 1. While this sample of 2 might not be enough to establish a
general pattern, it does show one possible response to the loss of a first child during
the Famine.

We can now look at the rural precincts, where the older and more fertile interviewees
have left us with considerably more to talk about.

Commune 2 had no women over the age of 18 in 1958 and only 2 over the age of 14.
It is therefore not surprising that only one of the 33 interviewees in the same gave
birth prior to the end of 1961. This interviewee, whom I will call Xiao Wei, had a
seven pound boy in August 1958, the very month that the Peoples Communes were
established. The child survived through to the time that the survey was conducted.
Xiao Wei had a second child in June 1960, a 5 pound boy who died aged 9 months,
and in July 1962 a third child, another 5 pound boy who this time survived. There is
little else to tell of the Famine in Commune 2

Commune 1 has a larger number of older women, with 7 women aged 14 or over in
1958 and 3 women aged 18 or over. Perhaps mainly because of this, the survey
recorded 6 women gave birth to at least one child before the end of 1961.

The first interviewee in Commune 1 to give birth, whom I shall call Xiao Ding, was
born in 1939 to an illiterate peasant family. She completed two years of primary
schooling, and just after turning 14 in the autumn of 1953 a peasant of the same
schooling seven years her senior. If the interview record is correct, this was a shotgun
wedding held a month after giving birth to her first baby, a 7 pound girl. The girl died
aged two and a half, in the same month that she gave birth to her second child, also a
7 pound girl who subsequently died aged six months. Two years later, in the spring of
1958, she gave birth to another 6 pound girl whom survived, 22 months later (in the
winter of 1959) to a 5 pound girl who survived, 10 months later (in the autumn of
1960) to another five pound girl who died aged 5 months, 10 months later (in the
summer of 1961) to yet another 5 pound girl. 17 months later, at the end of 1962,
Xiao Ding gave birth to her seventh child and first boy, who weighed five pounds.
The pervious girl died when the boy was just over a year old, while she was five
months pregnant with her eighth child, a six pound boy who died in his first month.
She went on to have four further boys spaced at two year intervals, two of whom died
in the first few months and one of whom died at age 10. After giving birth to twelve
children, she was left with a 3 yo boy, a teenage boy, and two marriage-age girls. Her
fertility history reflects the declining nutrient condition in the Famine years, with her
first baby conceived in 1953 weighing 7 pounds, her next two conceived prior to 1958
weighing 6 pounds, her next three conceived through to 1962 weighing to five
pounds, and her subsequent babies weighing 6 pounds up to her last baby conceived
in 1973 which weighed 5 pounds as her first.

Xiao Ding’s tale of poverty and woe also hides a miracle, for she was able to give
birth consistently with birth intervals of less than two years right through the famine
period. Neither her husband or parents were Party members, nor does her family
appear to have done particularly well within the system, for both her and her husband
remained peasants working for the collective in 1987. The relevant information I can
offer to explain this miracle is that her profile of a woman with limited primary
schooling who married someone of a similar background at the onset of
collectivisation in the winter of 1953 fits the profile of the poor peasant whose
relative status within the village improved during the 1950s.

7 further babies were born to other women in Commune 1 prior to the end of 1961. 4
survived, and the other 3 died after 1962, 2 at age 5 and 1 at age 10. The fact that
these deaths are recorded for round numbers over the age of 1 suggests that the
information on age of death of children recorded in the fertility survey should be
treated with caution. Similarly, routine responses appear to be recorded for
information on breastfeeding, the standard length of time spent breastfeeding girls
being given at 9 months, and 24 months for boys. While this may reflect important
culturally-specific information, it also suggests that we should be cautious in using the
information on breastfeeding recorded in the fertility surveys as a measure of

This hopefully gives you some sense of the type of data that recorded in the fertility
surveys. Now we will be shown some of the national and provincial trends shown in
the much larger sample of women interviewed for in the two-per-thousand survey.

To top