On Famine and Measuring Famine Deaths
Probably nobody starved to death during China’s Great Famine (‘Three Difficult Years) of 1959-1961. Thanks to long experience of feeding millions of men on the march during the war, the PRC was able to provide some food to everyone every day during the drought and flooding that devastated the cropland. In this essay, Utsa Patnaik, the Indian Marxist economist who taught at the Centre for Economic Studies and Planning in the School of Social Sciences at Jawaharlal Nehru University (JNU) in New Delhi from 1973-2010. Her husband is the economist Prabhat Patnaik.–Editor
***
There has been a long discourse on famine in the history of India. The students of her colonial history between 1765 and 1947, starting from the nationalist writers, have always made a distinction between the famines of the earlier type arising from local crop failures and inadequate systems of movement of supplies from surplus areas, on the one hand, and on the other, the secular tendency to famines arising from the specific operation of colonial surplus extraction, in particular through the revenue system. This system, it is argued, imposed not only a higher bar den of land revenue than did earlier indigenous fiscal practices, but this higher demand was required to be paid at rigid intervals in cash regardless of the actual fluctuating state of the harvest. The resulting debt-enforced changes in the crop- ping pattern away from food grains towards export crops, and the enforced Marketisation of surplus even for foodgrains, left no or little buffer for tiding over dearth. This produced an endemic vulnerability of peasant producers to famine. 1
The frequency and severity of famine in the late nineteenth century was a strong confirmation of this analysis. In the first half of the twentieth century, for which period systematic all-India data are available for British India, it was found that per capita food production declined by nearly 30 per cent in the inter-War period, while per capita export crops output registered rise. This pattern for India, we find, is replicated for every colonial situation: we find an inverse relation between exports and food production for the local population whether we look at die experience of Java under the Netherlands, Ireland under England, or Korea and Formosa under Japan, to cite only a few examples (Patnaik, 2000). High rates of growth of exportables were associated with low or negative growth of food crops. In India the severest fall took place in Bengal with a 38 per cent estimated decline in per head food availability between 1911 and 1947 (see Blyn, 1966; Sivasubramanian, 1965). It has been argued elsewhere by this author that this large secular fall in per head food consumption in rural Bengal implying a weaken- ing of the disease-resisting capacity of the mass of the rural population, has to be taken into account if we are to explain the high mortality in the subsequent famine of 1943, even though the proximate cause of the famine no doubt was the impact of the highly inflationary War financing burden imposed on India (Patnaik, 1991).
The expectation of life at birth in India was estimated to have been a pathetic 24 years at the turn of the century in 1901, which further declined to a mere 20.9 ; wars by 1921. The rapid war-time inflation eroding real incomes and the great influenza epidemic played their part in this abysmal record. The infant mortality rate (IMR) too rose during the war years to 219 from an already very high 204 in ; the period 1911-15 (see Table 2.1). In the next two decades the IMR declined to 161 and the expectation of life improved to reach nearly 32 years by 1941 and both indicators remained at around that level thereafter until Independence. Dur- ing the first decade after Independence, 1951 to 1961, the crude death rate aver- aged 22.8 per thousand; its decline was very slow and it averaged 19.0 during the next decade, 1961-71, whereas China starting from a higher initial level had already brought the death rate down to 12 or so by the end of the 1960s despite the rise during the famine of 1959-61, which we discuss in more detail later. Similarly, while India reduced its IMR from 160 before Independence to 85 by 1989-90, China starting from a higher initial level had already reached this point two decades earlier, by 1970. With regard to life expectancy, there was
Table 2.1
Infant Mortality Rate and Life Expectancy (1911-93)
Period
Rural
Urban
Combined
Life Expectancy in lean Census Tear M
F
1911-15
204
1901
23.6
24.0
1916-20
219
1911
22.6
23.3
1921-25
174
1921
19.4
20.9
1926-30
178
1931
26.9
26.6
1931-35
174
1941
32.1
31.4
1936-40
161
1951
32.5
31.7
1941-45
161
1961
41.9
40.6
1946-50
134
1971
46.4
44.7
1951-61
146
1980
54.1
54.7
1971
138
82
129
* 1981-86
55.6
56.4
1976
139
80
129
*1986-91
58.1
59.1
1981
119
62
110
* 1991-96
60.6
61.7
1983
114
66
105
•1996-2001
62.8
64.2
1986
105
62
96
1987
104
61
95
1989
98
58
91
1990
86
50
80
1991
87
53
80
1992
85
53
79
1993
82
45
74
Source: Health Information of India 1994, Govt, of India, Ministry of Health and Family Welfare. – ‘, IMR estimates for 1911-15 to 1941-45 are from Kingsley Davis, Population of India and i Pakistan; for 1946-50, from S. Chandrasekhar, Mortality in India; for 1951-61 from the •. Census Actuary, for 1971-93 on the basis of a sample registration system conducted by the Office of the Registrar General. .!
Note: ‘Life expectancy from 1981 onwards projected on basis of the 1981 Census by taking sex ratio as 105 males to 100 females.
48 • t)tsa<PatmiHi
improvement to about 60 years by 1990 in India. This occurred despite the sub- stantive failure to reach health services to villages where seven-tenths of the popu- lation lived; had there been the implementation of a more progressive rural health policy there is little doubt that infant and under-five mortality rates would have reached their low 1990 levels much earlier.
In the period of independence from direct colonial rule, most developing coun- tries privileged food production for their own population. The attempt to follow an autonomous trajectory of development also implied sharply growing wage- goods needs of industrialisation and hence the necessity of raising foodgrains output at a commensurate rate to avoid high inflationary pressures which would have adverse income distribution effects. Since the agricultural sector is the main source of economic surplus, various ways of extracting the surplus for industrialisation through taxation and through the terms of trade were thought to be necessary in the absence of a substantial expansion of trade and capital inflows. Most planned (China) and semi-planned, dirigiste economies (India) trkd to avoid rapid inflation in foodgrain prices relative to manufactured goods prices, with varying success; while taxation of agriculture was conspicuously more successful in China after the formation of cooperatives, compared to India.
The developing countries which have a high proportion of poor in their population, are typically characterised by a high death rate as well as a high birth rate, with the birth rate substantially exceeding the death rate. The rate of natural increase which is the difference between the birth rate and the death rate, has to be further adjusted to obtain the actual increase of population by taking account of net migration. In the course of development, as health services reach a larger segment of the population, education levels improve and per head real incomes •rise, it is expected that not only will the death rate come down but so will the birth rate. The aim is to obtain an even faster decline in the birth rate than in the death rate, if the rate of natural increase is to decline from the initial high levels.
In a class-divided society, average figures do not give a true picture of how bad the situation is for the most deprived, who owing to endemic lack of adequate nourishment are more vulnerable to sickness or morbidity. We know that poverty, under-nutrition and lack of access to affordable medical services get reflected in a much higher death rate than the average, for the poorer segment of the popula- tion. By the same logic, a given decline in the average death rate over time in the overall population, may well be very unequally distributed, with a very large de- cline in the already lower-than-average death rate among those in urban areas as their access to health services improves further, and a very small or non-existent decline in the death rate among the poorest in rural areas with initial high levels of deaths. A decline in the average death rate may be quite compatible even with a rise in the death rate for some segments of the population. In this context it is significant that with the inception of economic reforms and sharp macroeco- nomic contraction during 1991-93, there was a recorded rise in the crude death rate by 1993 in a number of states in India as well as a rise in the infant mortality rate (see Table 2.2).
On Famine and Measuring ‘Famine (Deaths’ • 49
Table 2.2
India: Statewise 1MB. in 1991 and 1993 and Vital Rates 1993
State
Infant Mortality Rate
Pet* cent
Death
Life
R
V
Combined
Change
Rate
Expectancy
1991
1991
1991
1993
1993/1991
1993
Tears
Andhra Pradesh
77
56
73
66
9.6
8.3
60.6
Assam
83
42
81
77
4.9
9.6
54.9
Bihar
71
46
69
73
5.8
10.5
58.5
Gujarat
73
57
69
62
10.1
7.6
60.1
Haryana
73
57
69
68
1.4
8.0
62.9
Himachal Pradesh
76
38
75
n.a.
n.a.
n.a.
n.a.
Karnataka
87
47
77
62
19.5
7.6
61.9
Kerala
17
16
16
16
0.0
6.0
72.0
Madhya Pradesh
125
74
117
99
15.4
11.1
54.0
Maharashtra
69
38
60
55
8.3
7.4
64.2
Orissa
129
71
124
103
16.9
10.8
55.5
Punjab
58
40
53
54
1.9
7.3
66.4
Rajasthan
84
50
79
85
7.6
9.1
58.0
Tamilnadu
65
42
57
56
1.8
7.9
62.4
Uttar Pradesh
102
74
97
86
11.3
10.4
55.9
West Bengal
76
47
71
59
15.2
7.7
61.5
All-India
87
53
80
74
7.5
9.0
59.4
Source: Health Information of India 1994 for first three columns, and Economic Survey 1996-97, Tabic 10.3.
Note: Bold type indicates states where combined (rural plus urban) mortality rate (the IMR, or the crude death rate) has risen compared to 1991. Data for 1991 death rate is given in Economic Survey 1995-96, Table 10.4.
The difference between endemic high death rate among the (mainly rural) poor, and what is identified by most academics as ‘famine deaths’, seems to be the fact that the first involves the poor dying at a rate higher than the average for the population, but slowly, unobtrusively and over a longer period of time owing to being chronically under-nourished and therefore being subject to higher morbid- ity; this higher than average death rate being considered nothing ‘out of the way or abnormal, given the existing unequal distribution of incomes which is taken as given. The second, which is considered not normal or usual and is termed ‘famine deaths’, involves a sudden rise in nutrition-deprivation and hence sudden rise in morbidity and death rate, usually among segments of the very same classes which are poor as a ‘normal’ state of affairs. In short, a sudden upward deviation from the prevalent death rate is thought of as ‘famine death’. The cause of a sudden and unpredicted rise may be various — a sudden decline in output, which affects the already poor more than others, or a sudden rise in the price of the basic food staple relative to the income of the consumers. The second cause is not necessarily predicated on the first; a sudden rise in food price may take place, not because output is less than usual, but because government follows policies of suddenly increased expenditures, which rapidly expand incomes in the hands of one seg- ment of the population and hence their demand for food, while leaving untouched the nominal purchasing power of another segment of the population, as happened
SO • IksaiPatnaili
during the 1943 Bengal famine when deficit-financed defence-related expendi- tures were suddenly raised.
‘The Question of 30 9A.iCCion t( Famine (Deaths during the Great Leap (period in China
China has performed much better than has India in reducing inequalities and reaching health care to its vast rural population and thereby reducing infant, child and maternal mortality rates as well as the overall death rate much faster than India has, over a comparable period (see Byres and Nolan, 1976; Liu Zheng, 1984). It is widely believed however that China had a more severe famine than India ever had, during the ‘Great Leap’ period, especially 1959-61 in which millions died; the figure of a total of 27 to 30 million famine deaths is frequently quoted. Over time, the lower estimate has dropped out and the figure appears to have become firmly fixed at ’30 million’ in the academic and in the general public mind. The main-source of this figure is Amartya K. Sen’s many writings and speeches which are more widely known and reported, certainly in India, than are the basic source which he uses, viz., the findings, embodied in a brief report in 1984, by the committee of demographers chaired by Ansley J. Coale (1984) with the aim of studying China’s population trends following the 1982 Census in China, and the later more detailed study by Judith Banister (1987). The argument made by Amartya K. Sen is that the absence of democratic freedoms including press freedom in China explains the fact that public policy did not respond to social crisis and the world did not have any inkling that such a massive famine ‘following the failure of the Great Leap Forward’ had taken place at the time. In a recent book, Develop- mint as Freedom, the proposition is repeated by Sen in categorical terms. 2
In general, the thrust of the argument is that collectivisation produces famine and the absence of a ‘free 5 press’ as in capitalist countries, prevents anyone outside these countries from knowing about it until much later — when Western scholars painstakingly uncover the facts through their research. Since collective ownership and production is the very essence of socialist production relations, this appears to constitute a damning indictment of socialism. The issue is complicated by the fact that in China itself, some who were always opposed to egalitarian principles of contribution and wanted to dismantle the rural communes (which were indeed disbanded from 1980 onwards), have acquiesced in the argument of famine as an ex post justification for doing so, regardless of the fact that they themselves, despite their active involvement in political life were apparently quite ignorant at that time that such a massive famine with 27 to 30 million deaths, had taken place in their own country.
China has had population censuses at irregular intervals unlike the regular de- cennial census in India. For the first four decades, up to 1990, population census was carried out in China in the years 1953, 1964 and 1985. What we have done in Table 2.4 is to take the 1964 Census population total (see Table 2.3) as the refer- ence point, and then calculated each preceding years’ population by using the
On famine and Measuring ‘famine <Death 6s’» 51
official annual rates of natural increase, on the assumption that net migration was negligible. The resulting series which emerges gives a terminal figure for 1955 which is broadly consistent with the 1953 Census total as well. We have also reproduced the grain output series and grain procurement series from Riskin ( 1981) . It is clear enough from this that, owing to a run of three bad harvest years, grain output in China did fall substantially, much more than it did in India during its food crisis of 1965-67, and the death rate also rose. But the order of the rise is nowhere close to giving the 30 million figure over the three years 1959-61; in fact, a quick mental calculation of the excess death rate compared to the pre- famine year, applied to the population in each year, would give a number between one-third and two-fifths of that figure.
Table 2.3 Key Statistics on China’s Population
Census
Estimated
Growth
Sex Ratio
Urban Population/
Date
Population
Rate per
Males per
Total (%)
Annum ‘000 females
1.7.1953
580.6
1075
12.9
1.7.1964
694.6
1.6
1055
19.1
1.7.1982
1008.2
2.1
1063
20.9
31.12.1985
1045.3
1.0
1069
36.6
Source: Visaria (1988), Liu Zheng, Song Jian and Others (1981).
Table 2.4
Vital Rates, Estimated Population of China, Grain Output’ and Procurement 1
Tear
Birth
Death
Rate of
Estimated
Grain
Grain
Rate
Rate
Natural
Population
Output
Procurement
Increase
(Million)
(Million ton 1 ) (Million ton)
1955
32.60
12.28
20.32
594.05
183.9
19.7
1956
31.90
11.40
20.50
606.23
192.7
14.9
1957
34.03
10.80
23.23
620.32
195.0
17.4
1958
29.22
11.98
17.24
631.01
200.0
20.9
1959
24.78
14.59
10.19
637.44
170.0
28.0
1960
20.86
25.43
-4.57
634.53
143.5
21.5
1961
18.02
14.24
3.78
636.93
147.5
17.5
1962
37.01
10.02
26.99
654.12
160.0
16.1
1963
43.37
10.04
33.33
675.92
170.0
17.0
1964
39.14
11.50
27.64
694.60′
187.5
17.0
1965
37.88
9.50
28.38
714.31
194.5
17.3
Source: Liu Zheng, Song Jian and Others (1981), C. Riskin (1987: Table 6.5, p. 137).
Notes: “Vital rates are per thousand population.
2 Grain output always includes tubers in China, converted to grain at the rate of 5 kg = 1 kg. •’1964 Census population figure for China, excluding Taiwan. This has been taken as the fixed point and the annual rates of natural increase, taken from C Riskin (1981), have been applied to obtain the population series. Riskin’s figures for birth rates, death rates and hence rates of natural increase reproduced above, are found to be identical with rounded values for these variables presented, not as continuous time series but in several separate tables, in Liu Zheng, Song Jian and Others (1981). The specific years 1960 and ioai ,n- hmwwr cvrliirfpH frnm nil tahles in the latter book.
52 . Visa (Pattmili
It would be instructive, therefore, to investigate how exactly the estimate of massive ‘famine deaths’ totalling 30 million, has been arrived at. But to anticipate, when we look carefully at the data we find that in fact, over 18 million out of the 30 million Chinese alleged to have died owing to famine, were not born in the first place. The Chinese are an extremely talented people and have moved moun- tains; but even their ingenuity would be stretched, we believe, to achieve the feat of dying_ without first being born. There is a basic responsibility that everyone, but more particularly academics have, to be clear and precise in their proposi- tions. To say or write that ’30 million people died in the famine in China’ conveys to the listener or reader the picture of people who were actually alive, starving to death. This would represent an extremely large order of mortality, unprecedented in the history of developing countries even in the colonial period in the twentieth century.
The fact is, however, that it did not happen: 30 million people actually alive in China in 1958 did not die over the next three years. The proposition that they did, is a false one. There was indeed famine in the sense of substantial excess mortality compared to 1958 (which was a good agricultural output year), during the three bad agricultural years 1959 to 1961 when grain output declined by nearly one- third; we estimate on Table 2.6 that the cumulative extent of excess mortality owing to rise in the death rate compared to the immediately preceding normal year, was under two-fifths of the figure quoted, viz., is likely to have been of the order of 1 1 .6 million over the three years, provided the death rates are not overes- timates. (The fact that even the official death rates may be overestimated is sug- gested by the fairly adequate level of per capita food grain output even in the worst year, 1960 and even after procurement is deducted, a question discussed at the end of this paper.)
« Even if we consider A.J. Coale’s much higher death rates, constructed as we shall see on dubious assumptions, the excess mortality over the three years with 1958 as the benchmark works out to 13.7 million (Table 2.7), less than half the quoted figure. Accepting the official death rates for the time being, a famine involving 11.6 million excess deaths of actually living people is quite bad enough; academics have a responsibility not to give currency to sensationalised estimates of more than two-and-a-half times higher deaths as has been done, most unfortu- nately, and done repeatedly. It is like saying that 8 million people died in the Bengal famine of 1943 , when all data show actual excess mortality of the order of 3 million or so.
The way that the inflated estimate for 27 to 30 million deaths in China has been arrived at, is through two alternative routes: first, to take not only the actual excess mortality on the basis of the available information on the rise of the death rate, but to add on to it the estimated numbers of babies not born at all, owing to a fall in i 1 the birth rate which was steep during this period. What is being done here is to / estimate the number of ‘missing millions’ in the population pyramid for these years, but this is misleadingly designated as ‘famine deaths’. Whether we take the 1 official figures or Coale’s constructed figures, this procedure gives the total figure j of a 27 million deficit in the population by 1961 compared to 1958, but over three-fifths of the deficit in both cases arises from fall in the birth rate and does not \ represent actually living people dying during the famine (see Tables 2.6 and 2.9) .
On Famine and Measuring ‘Famine <Deaths’ • S3\
It cannot be validly assumed either that the entire fall in the birth rate was neces- •: sarily food-shortage induced, for the birth rate fall was already marked in a good) grain output year, 1958 compared to 1957, and there were important institutional; changes during 1958 to 1961 which are likely to have affected reproductive behaviour and tended to lower fertility, over and above the usual adverse effects on . . fecundity of falling nutrition during the subsequent output decline during 1959-61. ,
The second route yielding the ’27 to 30 million famine deaths* claim is also highly dubious at the methodological level in involving statistical procedures which are opaque or inadequately justified. This route has been followed by the US demog- raphers and it consists of statistically reconstructing entirely new series of birth j rates and death rates for the inter-censal period 1953 to 1964 on the basis of the j fertility survey in the much later 1982 Census. This statistical reconstruction in- j volves assumptions carrying a substantial degrees of arbitrariness, and the results, j while jacking up all vital rates before 1971, jacks up to a proportionately higher j degree the death rates in the very early 1950s and in 1960, to levels much higher than seem to be consistent with either grain output figures in those years, or with the vital rates survey of 1953. Deaths from these dubiously reconstructed death rates are then subjected to linear trend fitting (despite the fact that death rate behaviour is necessarily non : Iinear) and the excess of all assumedly higher deaths during the shortage years above this fitted linear trend is called ‘excess deaths’ (which is 27 million according to Coale and 30 million according to Banister) and assigned to the ‘famine.’
We will look at this procedure again in more detail later, suffice it to point out that while unlike in the first route, unborn babies during 1959-61 are not being included here, what is being included is hypothetical steeply higher death rates for 1959-61, and an estimate of ‘excess deaths’ which is effectively (owing to linear trend fitting) based on the assumption that the death rate should have continued to decline, at a rate influenced by the hypothetical steeper rate of decline assigned to it for the 1950s. This procedure thus involves not only reconstructing death rates but also an implicit and methodologically dubious definition of ‘excess’ deaths by projecting falling death rates, which is a definition which does not seem to have i been ever applied by demographers and economists before, and never applied in / contexts other than China.’
What is the alternative measure of excess deaths that we can use, in place of the non-transparent and methodologically dubious procedure summarised above? A clue is provided by the fact that, even after a period of decline in death rate, people are not inclined to interpret a stable death rate for a few succeeding years as ‘excess deaths’ compared to the earlier period. In short, we consider that an ‘ex- cess’ cannot be defined in relation to some hypothetical decline which might have taken place under different circumstances. An ‘excess’ can only be defined as a rise . in relation to a benchmark level. India, for example, has had high, very slowly declining death rates (its normal 1960 death rate of 24.6 was little different from China’s abnormal official famine death rate of 25.4 in that year) without this being designated as ‘excess deaths’ leave alone ‘famine deaths’ in relation to the hypothetical much lower death rate it could have achieved with alternative poli- cies. The same argument applies to other relevant variables: for example the rate
of decline of the IMR has markedly slowed down in the economic reforms period in die 1990s in India compared to the 1980s, and while this is to be noted and is a cause for concern, we do not designate as ‘excess deaths’ the difference between actual infant mortality in the 1990s and the trend inclusive of earlier faster decline. ‘Excess deaths’ would apply to an actual rise in mortality in any sensible defini- tion — a rise such as is observed for some states in India between 1991 and 1993 (see Table 2.2).
When there is an upward spurt in the death rate induced by output shortfall or by other causes in a particular year or years, the conceptually simple and transpar- ent measure, understood by all, is to estimate the excess mortality by applying the normal mortality rate in a preceding normal, benchmark output year and look at the difference made to the population by the actual, higher mortality. This is the procedure we have adopted in Tables 2.6 and 2.9.
Conceptually, it is the excess of actual mortality in a particular year over the benchmark of a normal year, which is thus the transparent and correct measure of ‘famine deaths’. Of the alleged 27 million famine deaths during 1959-61, as men- tionedmore than three-fifths were not excess of deaths but deficit of births, owing to fall in the birth rate. There is a world of difference between people not being born at all, and actually living people dying of starvation or disease during a famine. The second may be legitimately referred to as ‘famine deaths’ but not the first, for one has to be alive in the first place to suffer death. While it is true that there is a causal link between nutrition and fertility in that periods of low nutrition are also those of low fertility, there are many factors other than nutrition which can affect the birth rate, and the rapid institutional changes taking place during the Great Leap for capital formation purposes (involving rural labour mobilisation, communal kitchens, etc.), which disrupted family life, were precisely such that the birth rate might be expected to fall to some extent, quite independently of what was happening to nutrition levels. This is also supported by the fact that the birth rate in China apparently rebounded and overshot at a far faster rate than output did as some institutional changes were reversed; between 1962 and 1965 excess births (continuing to use the 1958 birth rate as the benchmark) totalled 27.7 million, more than making up for the earlier population deficit during 1959-61 while reduction in the death rate relative to 1958, also contributed additional population (see Table 2.6).
Finally, postulated trends in any set of data or results, should be checked for consistency with values of other known important macroeconomic variables and their trends. The share of the rural poor in the rural population in India, for instance has been shown in various econometric exercises to vary inversely with per capita food production and directly with food price, with variations in rural public expenditure pla)dng also an important explanatory role through employ- ment generation (see Ravallion and Datt, 1995; Sen, 1996; Sen and Patnaik, 1997). In China food prices were strictly controlled administered prices, while employment generation was always privileged and days employed per worker rose steadily before 1980, 4 so the main independent variable affecting variations in <poverty becomes the state of the harvest.
On famine and Measuring ‘famine Deaths’ * 55
Looking at the per capita output of foodgrains in China in the worst years of output decline, compared to the years of lesser crisis in India, we find that Chi- nese per head availability from domestic output alone still remained noticeably higher than availability including imports in India, which did not suffer any generalised famine as China is alleged to have done. Given the more egalitarian nature of Chinese society compared to India’s by that date, in the light of the fairly adequate levels of per head food output in the worst crisis years in China (higher than India’s in the 1960s, and higher when compared to 1987-88, India’s last drought year), even the official order of rise in the death rate in 1960 to 25.4 per thousand, seems quite surprising.
The much higher death rate figures, which are figures which moreover diverge from each other as well, which have been constructed inferentially on various more or less arbitrary assumptions by Coale and by Banister, in particular the death rate for 1960, seem to be quite improbable. While Coale (1984) has no reference at all to food output trends in the brief report of the committee he chaired, Banister ( 1987) in her much more detailed book, shows an uneasy aware- ness of the inconsistency of her derived figures of death rate and birth rate, with the firm food output data. But instead of tackling the problem head on, or modi- fying her apocalyptically high constructed 1960 death rate, she relegates comment on the inconsistency to a long unnumbered footnote on page 233. But this incon- sistency is the very crux of the matter, which seriously undermines the credibility of both Coale and Banister’s rewriting of China’s demographic history.
In the section below let us take up these questions one by one, starting with an estimate of the excess mortality and deficit of births implicit in the Chinese official statistics quoted in Liu Zheng, Song Jian and others ( 198 1 ), and in Riskin ( 1981 ).
Jin Estimate of the Order of Excess (Deaths and (Deficit (Births
In Table 2.4 we reproduce the vital rates for China as given in C. Riskin (1981). China’s Political Economy — the Search for Development since 1949, after cross-check- ing these rates against the China Studies Series publication from Beijing, China’s Population: Problems and Prospects by Liu Zheng, Song Jian and Others (1981). We find that while the vital rates and rate of natural increase for different years agree in the two sources, there is some inconsistency of the absolute population figures given by the Chinese scholars, with their own rates of natural increase. Accord- ingly we have worked out the population series implied by their rates of natural increase, taking as the fixed point the 1964 population estimate of 694.6 million (see Table 2.4).
The estimated death rates for China at the time of Liberation vary quite widely, ranging from 28 to much higher levels of 35 and more. The initial rate of 28 for China is virtually identical with the 1941-51 inter-censal death rate of 26 to 30 in
56 • l>tsa<Patnatii
India. After the redistributive land reforms of the early 1950s which substantially did away with class inequalities, there were systematic mass drives for cleaning up the environment and eradicating disease-carrying pests, while with the formation of rural cooperatives from 1954, there was the inception of a rudimentary rural health care system. The emphasis was on public hygiene and preventive rather than curative medicine; in a poor, largely rural society this is likely to give good results. By 1958, the death rate in China is officially estimated to have declined to below 13 per thousand. (Coale, 1984 and Banister, 1987 place their constructed death rates using the 1982 Census . data and assuming under-reporting, much higher with the range between 18 and 21 for 1957-58, which if true would Still represent a considerable achievement for China since India did not reach this level until nearly two decades later.)
Looking at the years 1959 to 1961 in Table 2.4, we find that the death rate rose in 1959 by 2.6 per thousand compared to 1958, and rose further sharply in 1960, by as much as 10.8 per thousand over that of 1959, reaching over 25 per thou- sand. While it declined as sharply in the next year it was still 2.2 per thousand higher than in 1958. There is no doubt that the 1960 rise was highly abnormal, and the reason, as Riskin (1981) plausibly argues, lay not only in the nearly 30 percent output decline owing to the second successive bad agricultural year; the shortfall’s effects were aggravated by the excessive procurement burden placed on the peasantry, which reduced retained output in their hands more severely than gross output was reduced.
Taking 1958 as the benchmark year, there were excess deaths during 1959-61 totalling 11.6 million, the largest impact being in 1960 when excess mortality was over 8 million. Because China in the single preceding decade before 1959 had reduced its death rate at a much faster rate (from 29 to 12 comparing 1949 and 1958) than India had, this sharp rise to 25.4 in 1960 in China still meant that this abnormal, famine death rate was virtually the same as the prevalent death rate in India which was 24.6 per thousand in 1960, only 0.8 lower. This latter rate was considered quite ‘normal’ for India; further, in both the preceding and the suc- ceeding year India’s crude death rate was 8 to 10 per thousand higher than in China. Of course, each economy has to be judged in relation to its own internal performance; and no doubt the rise in the death rate during the worst years of output shortfall is a bad blot for China on its otherwise very impressive record of rapid decline and good food security. But it is not correct to say that ‘famine deaths’ totalled as much as 30 million; and it is not correct to imply that absence of press freedom meant that China’s then leaders, despite knowing about such massive deaths, were so cynical and depraved that they could mislead the world
\ successfully.
If China had continued to have the same birth and death rates as in 1958, the normal benchmark year, then its rate of natural increase would have remained at 174 per thousand. Applying this rate, the projected population for China in 1961 would have been about 27.3 million higher than it actually was. Thus the popula- tion deficit during the three abnormal years turns out to be 27.3 million (Table 2.5). By a slightly different measure, taking the 1958 benchmark, the excess of
On Taurine atuf Measuring ‘famine Deaths’ • 57
deaths as mentioned totals 1 1 .6 million and the deficit of births totals 15.3 million (Table 2.6). We have also estimated the population series implied in Coale’s ‘roughly adjusted’ vital rates and we find that the projected population in 1961 using his data compared to the actual estimated population, turns out to be 37.98 million less (see Table 2.8). The steeper rise in death rate and much steeper fall in birth rates assumed in Coale’s adjusted series leads to a difference with respect to esti- mated excess of deaths and deficit of births. As Table 2.9 using Coale’s adjusted rates shows, the excess of deaths is 13.7million and the deficit of births is as much as 23.7 million during 1959 to 1961. In other words, if we apply the conceptually transparent and correct measure of excess deaths to Coale’s own series, then we obtain an estimate of 13.7 million excess deaths. This is likely to be a substantial overestimate because the constructed death rate itself is an overestimate and, as we shall see, is also inconsistent with food output levels.
Table 2.5
Projected Population in 1961 using the 1958 Rate of Natural Increase
Rate of Natural Increase in 1958: 17.24 per thousand and Population 631.01 million Projected population by 1961 @ 17.24/1000 RNI 664.212 million Actual population in 1961 636.925 million Deficit of population 27.287 million
Source: Calculated from the data in Tabic 2.4.
Table 2.6
Excess/Deficit of Births and Excess/Deficit of Deaths, 1959-61, and 1962-65, Assuming 1958 as Base Rates
1959-61 Base: 1958, with birth rate (29.22) and death rate (11.98)
Tear
Deficit of
Excess of
Sum
Births
Deaths
1959
2.8306
1.6637
4.4943
1960
5.3047
8.5344
13.8391
1961
7.1336
1.4395
8.5731
Total
15.2689
11.6376
26.9065
1962-65 Base: 1958
Tear
Excess of
Deficit of
Sum
Births
Deaths
1962
5.0956
1.3113
6.4069
1963
9.5643
1..2242
10.7885
1964
6.8904
0.3334
7.2238
1965
6.1859
1.7715
7.9574
Total
27.7362
4.6404
32.3766
Source: Calculated from Table 2.4 by applying the difference between the death rate or birth rate in a specified year and the corresponding rate for 1958 treated as base year, to the specified year’s population. Note: The sum of annual deficit of births and excess of deaths will not add up to the population deficit estimated in Table 2.5, because the population base in this table is altering in successive years compared to Table 2.5.
S8m Utsa<Patnai^
Table 2.7
A.J. Code’s ‘Roughly CorMCtetf Vital Rates
Tear
Death Rate
Birth Rate
RNI
Estimated Population’
1954
29.1
44.4
15.3
589.28
1955
22.4
41.3
18.9
600.42
1956
20.8
40.2
19.4
612.06
1957
19.0
41.1
22.1
625.59
1958
20.4
37.7
17.3
636.41
1959
23.3
28.3
5.0
639.60
1960
38.8
25.2
-13.6
630.90
1961
20.5
22.3
1.8
632.03
1962
13.7
40.9
27.2
649.22
1963
13.0
47.3
34.3
671.49
1964
13.5
40.7
27.2
689.75
1965
11.1
39.7
28.6
709.48
1966
10.4
38.3
27.9
729.28
Source: Coale, 1984: Table 5, p. 47 and Tabic 10, p. 69.
Note: ‘This series of estimated population is not in Coalc. It has been calculated by us using his adjusted vital rates and taking the 1953 Population Census figure of 580.4 for mainland China as base.
Table 2.8 Population Deficit, 1958-61
Projected Population @ 17.3/1000 (1958 RNI) by 1961 Actual Estimated Population in 1961 Population Deficit
Source: Calculated from Table 2.7 data.
Table 2.9
Excess/Deficit of Births and Deficit/Excess of Deaths, 1959-64
Tear
Excess of Deaths
Deficit of Births
Tear
Deficit of Deaths
Excess of Births
1959
1.85
5.98
1962
4.48
2.11
1960
11.77
7.88
1963
5.05
6.55
1961
0.06
9.S8
1964
4.83
2.10
All
13.68
23.74
14.36
10.76
Source: Tabic 2.7 data with 1958 as base. Note to Tabic 2.6, applies here as well.
The first route of estimating famine deaths in China by taking it as identical with the deficit of population by 1961, is clearly wrong. The ‘deficit of population’ is a valid enough measure provided it is termed as such; what makes for tenden- tiousness is to say that 27 million people ‘died in the famine” which clearly, and wrongly suggests to the unwary reader and non-expert that they all actually existed to begin with.
Periods of food shortage do lead to decline in fertility; diis is a fairly well established proposition. Periods of mass mobilisation of males, for military ser- vice for example, also get reflected in a decline in the birth rate. There was no
670.01 632.03 37.98
On famine and Measuring ‘famine (Deaths’* 59
military conscription at this date in peacetime China, but there was something equivalent affecting both males and females in the villages. There was a massive mobilisation of both male and female workers for a stupendous construction effort during this period of early commune formation. Able-bodied women were drawn from the domestic sphere into construction work. The established peasant family living and work patterns were radically re-organised with the formation of the communes. Large bands of men and women set out in teams and brigades for constructing water management systems, cleaning up the environment and eradi- cating disease-carrying organisms, afforesting hills, terracing and bunding and so on. The private plot was abolished temporarily to facilitate mobilisation on a virtual war footing. Workers including women spent weeks on the work-sites, and there were communal kitchens and creches to look after children. (A detailed account at the village level may be found in W Hinton’s Shmfcm [1984] ). It is not surprising if this comprehensive disruption of normal peasant family life in the interests of construction, also contributed greatly to the observed decline in the birth rate, as birth decisions were postponed and in any case physically rendered more difficult. Abolition of private plots and insistence on communal kitchens were soon attacked as ‘ultra-left’ measures as output declined in 1959. With the stabilisation of the new system, there was a dismantling of communal kitchens and reversal to family life. The birth rate again surged to unprecedented heights, from 1962 onwards peaking at 37.9 in 1964, as though there was a pent-up demand for children which was now being given full play.
It is inconceivable that such a large number of ‘famine deaths’ as even the realistic 11.6 million should have been wilfully suppressed by a state in China which had demonstrated its commitment to peoples’ welfare by undertaking mea- sures to reach basic food security and health services to the poor, and which had achieved a much faster reduction in infant mortality and the death rate in the very first decade of independence than had India. We may try to provide a hypothesis regarding why no-one including the Westerners in China, even noticed that mor- tality was higher during these years.
As regards the genuine excess mortality during China’s difficult famine years, while shortages and deficiency diseases were very real and visible, massive famine appears to have been near invisible to all including the foreign diplomats and journalists present at that time in China. Even though there was famine, we may hypothesise that its precise dimensions may have been, paradoxically, rendered opaque by the substantially egalitarian structure of Chinese society which was no longer characterised by sharp class differences. The food shortage was not con- centrated in a sharp drop in consumption by the members of a particular deprived classes like poorer peasants and labourers who then migrated to cities and died in the sight of all, while others had more than enough to eat, as typically happens with famine in class societies, and as happened for example in the great Bengal \ famine of 1 943-44. Food shortage was spread out over the rural consuming population much more evenly and therefore must have led to higher rates, but not immediately or obviously visible higher rates of mortality for particular groups. We would also expect the shortages to have led to higher mortality in the particularly )
60 • VtsatPatnatii
vulnerable segments in an otherwise equal society — parturient mothers, infants and the very old It is a mistake to think that all real trends in vital indicators are necessarily visible to the individuals at the time.
Thus even though we ourselves in India have lived through the period when infant mortality has fallen greatly, it is a matter we are convinced of not from our direct experience of it, but after the numbers have been counted and presented to us. China’s leaders were not guilty of wilful suppression of knowledge of the
/ higher mortality; the knowledge itself was built up much later than the evenis, and an approximation to the correct estimate as we have seen is just over one-third of the sensationalised estimates which are still being circulated. 5
Let us now turn to the second route through which massive famine deaths are estimated, namely by re-writing the death rates and birth rates in China for the two decades after Liberation by substantially raising both, and then fitting linear trends. This is the exercise that a number of US demographers have been carrying
‘outj starting with the committee with Ansley J. Coale in the chair, whose report on Rapid Population Change in China 1952-1982 became available in 1984 while Judith Banister published her book in 1987, based on following much the same method as Coale (1984) but arriving at even more apocalyptic findings. It is important to study how exactly they have arrived at their results regarding 27 to 30 million ‘excess deaths’ since none of those who freely throw the 30 million figure around seem to have looked at the original procedure of arriving at it.
First, high fertility figures were obtained from a large sub-sample (0.1 per cent of the population) in the 1982 Census, which were then applied to work back- wards and obtain a completely new constructed series of births starting from thirty years or more earlier (from 1953 to 1981 in Coale and from 1949 to 1984
‘in Banister), which were found to exceed the estimated births in the official se- ries. According toNai Ruenn Chen’s useful book (Nai-Ruenn Chen, 1966: 6), a very large survey of 30.18 million persons had been carried out in addition to the
/ 1953 Census, specifically to ascertain vital rates. This was a 5 per cent sample of the population and yielded a birth rate of 37 and death rate of 20; but these findings are brushed aside by the US demographers who reconstruct much higher birth rates — in the 40s — and thereby, much higher death rates — in the high 20s — for the first half of the 1950s). Total reliance was thus placed by them on the Census 1982 responses to questions about births, even though this showed im- probably high fertility rates for the late 1950s and 1960s compared to similar societies like India; 6 and even though biases might have been introduced by the one-child policy already in force for a decade (by inducing the possible backward shifting of reported births to years before 1970).
If extra people were born, some of those extra people must have died. The difference between the constructed total of births over the inter-censal period 1953-64, and the official inter-censal increase in population, is treated as the actual total of deaths, which exceeds the official total by a heroic 60 per cent. However, it is not known when these assumed extra deaths took place. These assumed extra deaths are then allocated arbitrarily over the inter-censal years, in such a manner that the official death rates are adjusted upwards by a significantly
On famine and Measuring ‘ famine (Deaths’ • 61
higher proportion in the early 1950s and during 1958–60, than in other years (Coale, pp. 30, 66-68; Banister, pp. 114-15). All these statistical manipulations require total faith in the accuracy of the official Census population totals, even while every official vital rate is assumed to be underestimated.
The element of arbitrariness thus enters at two points, first in the rejection of all other official data even from very large samples, but acceptance of 1982 fertility data; to rework birth rates for even thirty years earlier; and the necessarily arbi- trary allocation of the resulting assumed under-reported deaths to the various inter-censal years. Both A.J. Coale (1984) and J. Banister (1987) follow this pro- cedure, and since neither gives any results of making alternative assumptions regarding assumed under-reporting from the ones they have actually made, in order to allocate the extra deaths to particular years, the sensitivity of their results to their specific assumptions remains unknown to the reader. (In fact, a quick calculation shows that the constructed death rate in a particular year is highly sensitive to the assumed degree of under-reporting and that a slightly different assumption would make a large difference.)
Because of their manner of calculating and allocating the derived extra deaths, both therefore obtain (a) much higher values of death rates and birth rates during 1950 to 1961 than in the official data, higher by 8 to 22 per 1000; (b) a steeper rate of fall — in Coale but not in Banister — in the death rate up to 1958 compared to official data; and (c ) a steeper rate of rise in the death rate from 1958 to 1960. The arbitrariness of the procedure is illustrated by the variation between the two estimates themselves: Coale has the death rate declining by 30 per cent during 1954 to 1957 while Banister has it declining by 15 per cent Coale has the death rate rising by 90 per cent between 1958 and I960, from 20 to nearly 39 (see Table 2 7) while Banister has it rising by 118.4 per cent over the same period from 20 to 44.6 (see Table 2.10). Coale has a 1960 death rate which is nearly 39 compared to the official 25.4 while Banister goes one better, making assumptions which hike the 1960 death rate to 44.6 (Each one-point rise in die imputed death rate for 1960, automatically adds 0.63 million to the estimated excess deaths in that year alone, even following the benchmark method).
The reader no longer knows what the death rate in 1960 in China was. A variation from 25.4 (official value) to 44.6 (Banister) is no mean variation. The 1960 death rate becomes like a hot-air balloon which drifts higher and higher with the degree of stoking and the heat generated below it With appropriately different assumptions about under-reporting in that year the rate could be hiked up further: why not to 50, or even more? With alternative, equally plausible or more plausible assumptions than those made by Coale and Banister, one could also reconstruct the death rate in 1960 to a level lower than the official one of 25.4. There is absolutely no reason to accept the Coale or Banister reconstruction rather than any other; indeed their reconstruction comes into patent conflict with the output data at several points. Coale’s reconstructed death rate of 29 for 1954, which is his peak death rate for the entire 1950s decade, becomes meaningless when we see that 1954 was a normal year when grain output was 2.2 per cent higher than in 1953 and there was absolutely no reason for the death rate for the 1950s to have peaked in that year.
62 • VtsaVattiaili
Table 2.10
Banister’s Constructed Vital Rates, 1953-65
Tear
Death
Birth
Rate of
Rate
Rate
Natural Increase
1953
25.8
42.2
16.4
1954
24.2
43.4
19.2
1955
22.3
43.0
20.7
1956
20.1
39.9
19.8
1957
18.1
43.3
25.2
1958
20.7
37.8
17.1
1959
22.1
28.5
6.4
1960
446
26.8
-17.8
1961
23.0
22.4
-0.6
1962
14.0
41.0
27.0
1963
13.8
49.8
36.0
1964
12.5
40.3
27.8
1965
11.6
39.0
27.4
Source: Birth rate using 1982 fertility survey and reconstructed death rate (Banister, 1987: 352, Table 10.1).
Banister recognises reluctantly in a footnote (Banister, 1987: 233) that ‘there is some discrepancy between China’s grain production and the apparent timing of the famine.’ Banister’s assumption that the famine started early in 1958 because die birth rate starts dipping in that year, even though output was normal, is quite unwarranted and is the result of ignoring all factors other than nutrition which might have affected fertility (we have already indicated such factors as drastic .change in living and working habits during collective labour mobilisation at the inception of the communes, so well described by William Hinton at the village level in Sbenfan.
There is a large discrepancy with regard to output trends and Banister’s story on fertility. As Banister herself recognises (p. 233), fertility is expected to decline nine months after the worst nutritional stress, whereas according to her own reconstruction, it declined much more in 1959 over the previous year (by as much as over nine per 1000) though 1958 was a good harvest year, than it did in 1960 over 1959 (less than two), or in 1961 over 1960 (just over four: see Table 2.10). She is unable to explain this. In fact, her rigidly held assumption that the initial large decline in the birth rate was necessarily only because of lower fecundity from lower food intake, itself is a wrong assumption and arises from the evident lack of any knowledge of or insights into the processes of profound changes in conditions of peasant collective work taking place at that time, which would necessarily have impacted on the birth rate.
To convert underemployed labour into capital assets, especially irrigation as- sets, was at the core of the massive labour mobilisation strategy of the Great Leap involving commune formation. This in itself was both conceptually bold and cor- rect in terms of economic logic as a number of scholars have recognised— some 16.5 million mm was added to irrigated areas during 1958-61 apart from other asset creation. 7 In actual implementation there were problems because, first, there
On Tamine and Measuring ‘ ‘Famine OeatAs’* 63
appears to have been excessive initial centralisation and hence arbitrariness, with the entire huge commune as the unit of decision making and accounting, rather than the brigade (village) as was the case later; second, there was drastic overesti- mation of immediate output gains, without taking account of the fact that all capital formation projects involve lags before yielding returns, and therefore there was excessive procurement arising from inflated output estimates; 8 and finally the entire enormous exercise in labour mobilisation through institutional change, hap- pened to coincide with a run of bad harvest years. Once these problems were recognised and rectified and the commune system had settled down however, from 1962 onwards it seems to have generated a very impressive order of tangible capital formation and employment both on farm, and in off -farm rural enterprises (see Bramall, 1993; Rawski, 1979, 1982; Wheelwright and McFarlane, 1970), leading to a substantial rise in the real standard of living up to 1978, which was contributed to by greatly improved rural health and primary education systems set up out of collective funds.’
Returning to the estimates of famine deaths, the next and equally dubious part of the reconstruction exercise in Coale (1984: 7, 69-70), is to fit a linear trend to the constructed deaths for 1953 and 1964, obtained by applying the initial imputed much higher death rate and later imputed steeper rise, and then to say that the entire difference between the linear trend and constructed deaths for particular years represents excess deaths (27 million in Coale). Banister does not specify what procedure she has followed but merely informs the reader that ‘the computer reconstruction of China’s population trends utilised in this book, which .assumes under-reporting of deaths in 1957 as well as in all the famine years, results in an estimated 30 million excess deaths during 1958-61’ (Banister, 1987: 85; emphasis added). The computer will work out the procedure fed into it, and from the similarity to Coale’s result we may perhaps infer linear trend fitting in this case as well
Why anyone should at all go about fitting linear trends to deaths derived from a variable, the death rate, which is known to behave necessarily in a non-linear manner (declining and then bottoming out), is not at all clear. In fact, it is a wrong procedure, involving the implicit assumption that the death rate should continue to decline, at a rate influenced by the extremely steep initial decline implicit in the constructed series for 1953-58. But this is impossible; we know that the death rate in a population can never go on declining to hit a value of zero; not even the Chinese have found a way of ensuring immortality.
Let us consider Coale’s linear trend formed, as he tells us on page 70, by using hypothetical deaths for 1953 to 1964 (10.4 million and 9.4 million). On project- ing this linear trend the reader finds it should have hit the x-axis around 1990 by which time the Chinese population should already have zero deaths and have attained immortality (see Coale, 1984: p. 7, Graph 2.1 on which these values may be plotted by redefining the y-axis as total deaths) . In Banister’s similarly reconstructed graph (Banister, 1987: p. 353, Egure 10.1) the linear trend of deaths (15 million and 6.7 million) would when projected reach the zero deaths level even earlier, by 1977.
If this procedure of projecting a linear trend is patently absurd in the case of a variable like deaths, so is the actually undertaken procedure by Coale and Banister
64. VtsaPatnaili
quite meaningless, of fitting a linear trend to hypothetical deaths for positive values of death rate, and calling the difference of the hypothetical values for par- ticular years from this trend, as ‘excess deaths’.
Inconsistency of Assumed High MortaLity
with ToocC fivaitabitity
The very high death rates ‘constructed’ by Coale and Banister for the early 1950s and for 1960 (as we have seen, for the latter year, nearly 39 in Coale and as much as 44.6 in Banister compared with 25.4 in official sources) also appear to be quite inconsistent with the known per capita food grains availability out of domestic output in China in 1960. Indeed even the much lower official death rate of 25.4 in 1960, needs to be explained in the light of the fact that the lowest possible calcu- lation of per capita food output that we can make for China, gives us figures which are not only substantially higher than the per head output, but are also much higher than availability (including imports) in India during its period of food shortage and stress in 1965-66 and 1966-67, a few years after China’s crisis, which produced no generalised famine. China’s worst year was 1960. The output per head figure for that year, no matter how hard we try to obtain a minimal estimate, also remains stubbornly above the corresponding figure for a much later minor drought year, 1987-88 in India after two decades of green revolution.
China had a per capita grain output which was about one-third higher than India’s even in 1951, and has steadily increased the margin in its favour, apart from 1959-61 when its output per head came closer to India’s but still remained higher. 10 Foodgrains output was substantially lower than normal for two succes- sive years in India, 1965-66 and 1966-67, which did not experience famine, despite its highly inegalitarian rural structure. In Table 2.11 we have adjusted
Table 2.11
Per Capita Food Output for China (1958-61)
Year
CDR>
CBS. 1
Total Population
Per Capita Gross
Per Capita Net*
Grain Output? (%)
Grain Output (kg)
1958
20.4
37.7
631.01
278.9
300.9
1959
23.3
28.3
637.44
234.7
236.2
1960
38.8
25.2
634.53
199.0
203.9
1961
20.5
22.3
636.93
203.8
216.4
Source: Vital rates and Population from Tabic 2.4. Notes: “CDR = Crude death rate. 2 CBR = Cmde birth rate.
‘Gross grain output is output of grain and tubers, roughly adjusted for unhusked paddy by taking 0.88 times grain output given in Table 2.4. Per capita value is obtained by dividing this by total population.
4 Net grain output is obtained by deducting procurement from output in Table 2.4, then taking 0.88 times the result. It is then divided through by rural population (assumed to be 83 pa – cent of the total on the basis of urbanisation data in Table 2.3) to give retained
On Famine and Measuring ‘Famine <Deatks’» 65
China’s grain output downwards by 12 per cent to take account of unhusked paddy in her food statistics compared to rice in India. Per capita output for China is on two bases: gross output, and net output, after deducting procurement. It is assumed that all procured output in China went to urban areas and no part was redistributed to food-deficit rural areas, so the output net of procurement is di- vided through by the rural population to obtain rural availability. Finally, potatoes have been included along with foodgrains in India following the Chinese practice, using the same conversion rate (5 kg potatoes equal to 1 kg grain) as in China (Table 2.12).
Table 2.12
India: Per Capita Food Output Availability, 1966-68, 1988
Tear
Papulation
Gross Grain
Per Capita Availability
Total 1
(Million)
Output
Domestic
Imports
Vital
(Inclusive of
(tnn. ton)
Output
<fy)
(kg)
Tubers)
<fe>
M
1966
493.2
71.1
144.2
20.9
165.1
166.4
1967
504.2
73.1
145.0
17.3
162.3
163.7
1988
799.2
132.9
166.3
3.5
169.8
173.3
Source: Calculated from Economic Survey 1995-96, 1998-99. Read 1966 as agricultural year
1965-66 and so on. All except the last column, excludes tubers. Note: ‘Grain output figures in India do not include tubers, so this has been added in the last
column only, at the conversion rate employed in China, for comparability with figures in
Table 2.11.
1 1 may be seen that China’s rural per head food output net of procurement, for 1960 was 57.5 kg higher than India’s overall per head output (including tubers) without deducting procurement in the worst year 1966-67, and was 40 kg higher than India’s availability (which included large net imports of 10 million ton or 17.3 kg per head). China’s overall gross per head grain output in 1960 was more than 50 kg higher than India’s corresponding 1966-67 output and 35.3 kg higher dian India’s availability in that year. 11 Given the egalitarian structure of rural soci- ety in China by this date, the fact of famine in China when no famine was experi- enced in India is certainly a major puzzle. After two decades of Green Revolution, India experienced a drought year in 1987-88, when per capita food output includ- ing tubers clipped to below 170 kg and availability to around 173 kg. There was no question of ‘famine’ in the sense of a general rise in the rural death rate even though at least two-fifths of the population was below the poverty line. China’s per head gross output in 1960 turns out to be nearly 30 kg higher and per head net output, nearly 35 kg higher than in India in 1987-88. Why China should have experienced a famine at all in 1960 when its food output per head was so much higher than in India decades after the latter’s Green Revolution, is a question which needs to be investigated and answered before it is possible to accept uncritically even the official Chinese demographic account for the years concerned
66. lksa<Patiiaili
9fO<T<ES
1. The classic exposition was by R.C. Dutt (1970a and b) The Economic History of India (2 vols.) for a summary of historians’ arguments, sec Patnaik (1999), ‘The Process of Commercializa- tion under Colonial conditions’ in The Long “transition— Essays on Political Economy.
2. It must, however, also be noted that there are real handicaps that China experiences compared with India because it lacks democratic freedoms. This is particularly so when it comes to flexibility of economic policy and responsiveness of public action to social crisis and unforeseen disasters. The most prominent contrast lies perhaps in the fact that China has had what is almost certainly the largest recorded famine in history (when thirty million people died in the famine that followed the failure of the Great Leap Forward in 1956-61), whereas India has not
. had a famine since independence in 1947 (Sen, 1999: 34).
3. We do not sec these, or any other demographers or economists fitting linear trends by including the falling death rates in Russia between 1945 and 1990 when it was still a constitu- ent part of the Soviet Union, and then estimating the excess deaths resulting from market- oriented ‘shock therapy’ in the 1990s, by adding up all the actual deaths above these trends. Between 1990 and 1994 alone, the death rate for able-bodied males of working age in Russia has shown a rise from 48.8 to 84.1 per thousand and expectancy of life at birth has fallen significantly.
4. For steady rise in employed days in rural areas, sec Schran (1969) and Rawski (1979, 1982). A brief discussion of surplus labour mobilisation through the co-operatives and later com- munes is contained in Patnaik (1998).
5. On a visit to China in the eighties, at the time the inflated ‘famine deaths’ estimates were being talked about in the West, this author mentioned these estimates and asked some very senior Chinese economists involved in the planning process, about their own experience of this period. They were extremely surprised and said that while there were cases of many more deficiency diseases than usual they were not aware of widespread famine deaths.
6. The fertility rates from the 1982 Census in China gave TFR values well above 6.0 for many years in both the 1950s and the late 1960s and even 7 in one year. The TFR value has been below 6 in India from the second half of the 1950s and has steadily declined despite much poorer food and health security than in China.
7. Wheelwright and MacFarlane (1970), point out that ‘Nurksc-typc savings’ by conversion of deferred-wage labour to capital was the core of the strategy. Hinton (1984: Chapters 29-33) documents the enthusiasm with which villagers spoke of the Great Leap despite all the problems. See also my discussion (1998) of the relation between capital formation through surplus labour mobilisation, deferred wages and egalitarianism.
8. Hinton (1983, Part 111, gives a graphic account of how Long Bow village got some of the grain already procured, given back to it thanks to the insistence of a sensible local cadre that output had been overestimated, and thus avoided shortages faced by adjacent villages.
9. The leap in employment per worker and addition to farm productive assets up to 1975 documented by Rawski (1979, 1982) suggests that it was a successful strategy. During visits to a number of communes in 1983, this author was informed in each ca.se that ‘we built our water conservation system during the Great Leap’ and that yields had stabilised and steadily improved with assured irrigation. Sec Patnaik 1988a. Sec also J.E. Nickum’s 1974 thesis on A Collective Approach to Water Resources Development — the Chinese Commune System 1962- 1972′.
10. For an earlier comparison of Indian and Chinese grain output per head see Patnaik 1988b.
11. China’s gross per capita output (total output divided by total population) at 199 kg is seen to be a little lower than net rural per capita output at 204 kg (gross output minus procurement divided by rural population) because even in the worst year, 1960 the share of grain output which was procured, was less than the share of urban population in total population. Normally per head grain consumption is higher in rural than in urban areas and remained so in 1960 as well though the margin was reduced.
1
On famine and Measuring ‘Tanrine (Deaths’ • 6?^ <R<E’F<E<%E!HC’Ed
Banister, J. 1987. China* Changing Population. Stanford University Press. Barker, R., R. Sinha and B. Rose (cds). 1982. The Chinese Agricultural Economy. Colorado: Wcstvicw Press.
Blyn, G. 1966. Agricultural Trends in India 1891-1949: Output, Area and Productivity. Philadelphia .
PA: University of Philadelphia Press. Bramall, C. 1993. In Praise of Maoist Planning: Living Standards and Economic Development in
Sichuan since 1931. Oxford: Clarendon Press. Byres, T.J. and P. Nolan. 1976. Inequality: India and China Compared 1950-1970. Milton : i
Keynes: The Open University Press. Coale, A. J. 1984. Sapid Population Change in China, 1952-1982. Washington DC: Committee
on Population and Demography, National Academy Press. Dutt, R.C. 1970a. ‘Under Early British Rule’, in Economic History of India, Vol. 1. Ministry of
Information and Broadcasting, Government of India: Publications Division. 1970b. ‘In the Victorian Age’, in Economic History of India, Vol. 2. Ministry of Informa- i
tion and Broadcasting, Government of India: Publications Division. Hinton, W. 1984. Shenfan. London: Pan Books in association with Martin, Seeker and Warburg. Liu Zheng. 1984. ‘Population’, in Yu Guangyuan (cd.), China’s Socialist Modernization. Beijing:
Foreign Languages Press. Liu Zheng, Song Jian and others. 1981. China’s Population: Problems and Prospects. Beijing: New j
World Press. j Mitra, Ashok (cd.). 1988. China: Issues in Development. Delhi: Tulika. j Nai-Ruenn Chen. 1966. Chinese Economic Statistics — A Handbook for Mainland China. Edinburgh: ,
Edinburgh University Press. Nickum, J.E. 1974. A Collective Approach to Water Resources Development: the Chinese
Commune System 1962-72. Ph.D Thesis, University of California at Berkeley. Patnaik, U. 1988a. ‘Three Communes and a Production Brigade — The Contract Responsibility
System in China’, in Ashok Mitra (cd.), China: Issues in Development. Delhi: Tulika. 1988. ‘Some Aspects of Development in the Agrarian Sector in India’, Social Scientist ,
16(2). ‘ ” ; 1 991. Food Availability and Famine — A Longer View’, Journal of Peasant Studies, j
19(1 ); also included in The Long Transition — Essays on Political Economy. Delhi: Tulika. 1998. Alternative Strategies of Agrarian Change in relation to Resources for Develop- ment in China and India’, in D. Nayyar (ed.), Economics as Ideology and Experience: Essays j
presented to Ashok Mitra. London: Cass. j
1999. The Long Transition— Essays on Political Economy. Delhi: Tulika. j
2000. On the Inverse Relation between Primary Exports and Domestic Food Absorb- j
tion under Liberalized Trade Regimes. Paper presented at an international conference on j
Globalization, Structural Change and Income Distribution, 14-17 December, Chennai. ] Ravallion, M. and G. Dart. 1995. Growth and Poverty in Rural India, Background paper to the . {
1995 World Development Report WPS 1405 World Bank Washington DC. j Rawski, T.G. 1979. Economic Growth and Employment in China. New York: Oxford University ■ j
Press.
1982. Agricultural Employment and Technology’, in Barker, Sinha and Rose (eds), The
Chinese Agricultural Economy. Colorado: Westview Press. j Riskin, C. 1981. China’s Political Economy— the Search for Development since 1949. New York: ;|
Oxford University Press. j Schran, P. 1969. The Development of Chinese Agriculture 1950-1959. Urbana: University of Illinois j
Press. \ ;
Sen, A. 1996. ‘Economic Reforms, Employment and Poverty: Trends and Options’, Economic and j
Political TMekly 31(35-37). Sen, A. and U. Patnaik. 1997. Poverty in India. December; Riper under DSA Programme, Centre ,
for Economic Studies and Planning, Jawaharlal Nehru University. ‘;
68* Visa <Patnai£,
Sen, A.K. 1999. Development as Freedom. New York: A. Knopf.
Sivasubramonian, S. 1965. National Income of India 1900-01 to 1946-47 (mimeo) Delhi School of Economics.
Visaria, P. 1988; ‘Demographic Trends and Population Policy in China’, in Ashok Mitra (ed.),
China: Issues in Development. Delhi: Tulika. Wheelwright, E.L. and B. McFarlane. 1970. The Chinese Road to Socialism. New York, Monthly
Review Press.
Yu Guangyuan. 1984. China’s Socialist Modernization. Beijing: Foreign Languages Press.
Vses of History: India’s Contribution to the Study of Climate and <Disease
‘Elizabeth Wfiitcoml
The task of understanding the complex relations between climate and disease now recognised worldwide as a matter of urgency. India is uniquely placed to make a seminal contribution. The meteorology of that most fascinating of weatfo systems, the monsoon, has been systematically observed with unrivalled range and precision over the south Asian subcontinent from the establishment of the India Meteorological Service in 1875. These exceptional data are matched, in quantity and quality, by observations of central and provincial medical officers of the departments of the sanitary commissioners (later public health) with vital statistic consistently recorded monthly from the establishment of registration in 186< Annual meteorological and medical reports contain, in addition, detailed descriptive detail in text which accompanies, and fills out, the statistical tables of month) observations. The unique British Indian archive is richest perhaps in materi; from which we may reconstruct the meteorological and medical history of the greatest killer diseases of the subcontinent: cholera and malaria. Our reconstruction can draw on a comprehensive series of monographs on climatic — oceanic graphic and atmospheric, biological — marine and terrestrial, to establish patterns of events in the modern history of vector-borne diseases, the precise role of climate and the limits of climatic determinism.
The mound of available data is immense. A chance finding, that in certain famine years there is a highly significant association, in the extreme conditions c these years, between climatic events and in the pattern of principal causes of disease, sets a hypothesis with which we can begin to explore it.
The subcontinent of India in the late nineteenth century was visited repeatedly; by severe and widespread famine, with death rates reaching extraordinary levels of 25 even 35 per mille. On closer inspection of the official records for years marka by the most catastrophic of famines, records distinguished by meticulous medica and meteorological observation, a peculiar pattern of climatic conditions emerges not merely drought, prolonged and persistent, but drought interrupted, at irreau
Thinking Social Science
in India
s in Honour of JLRce ^hornet
‘Editors
Sujata (PateC Jasodhara (Bagcki %fishna <Rflj
Sage(Pu6(ications ‘New (DeOUAIhousand Oaks/London
6 thoughts on “On Famine and Measuring Famine Deaths”
Pingback: Mao's Famine Mao's Famine - Here Comes China!
Pingback: TOMBSTONE by Yang Jisheng. A Critical Review - Here Comes China!
Pingback: Why China Leads the World | Here Comes China..
Pingback: China's Great Famine - Here Comes China!
Pingback: Mao's Great Famine? - Here Comes China! - The Great Leap Forward
Pingback: IQ China and the Wealth of Nations - Here Comes China!