Skip to main content Accesibility Help

Agency and Famine in China's Sichuan Province, 1958–1962*

  • Chris Bramall (a1)

A revisionist literature on the Great Chinese Famine has emerged in recent years. These revisionists focus primarily on the question of agency. They claim that that neither poor weather nor the excesses of local cadres can explain the extent of mortality; rather, responsibility lies squarely with Mao and the CCP leadership. Using county-level data on mortality, output, rainfall and temperature for Sichuan province, I argue that this revisionist view is unconvincing. Weather admittedly played only a minor role, and the zealotry of the Party centre contributed significantly to the death toll. However, variations in mortality between Sichuan's counties appear to have been essentially random – suggesting that differences in local cadre responses to central government policy were decisive in determining the scale of famine.

Hide All

1 “Excess” meaning actual deaths minus the number of deaths that would have occurred if conditions had been normal. For conventional estimates of mortality, see Ashton, Basil, Hill, Kenneth, Piazza, Alan and Zeitz, Robin, “Famine in China, 1958–1961,” Population and Development Review, Vol. 10, No. 4 (1984), and Banister, Judith, China's Changing Population (Stanford, CA: Stanford University Press, 1987).

2 See especially Thaxton, Ralph, Catastrophe and Contention in Rural China (Cambridge: Cambridge University Press, 2008), Bernstein, Thomas P., “Mao Zedong and the famine of 1959–1960,” The China Quarterly, No. 186 (2006) pp. 421–45, and Manning, Kimberly Ems and Wemheuer, Felix (eds.), Eating Bitterness: New Perspectives on China's Great Leap Forward and Famine (Vancouver: University of British Columbia Press, 2011).

3 Tongjiju, Guojia (State Statistical Bureau) (SSB), Xin Zhongguo 50 nian nongye tongji ziliao (Materials on 50 Years of Agriculture in New China) (Beijing: Zhongguo tongji chubanshe, 2000), p. 37; SSB, Zhongguo tongji nianjian 1983 (Chinese Statistical Yearbook) (Beijing: Guojia tongji chubanshe, 1984), pp. 422, 438. Chinese data for this period are unreliable, because of pressure to exaggerate output and the collapse of the statistical system; see Wei, Yue, Dangdai Zhongguo de tongji shiye (China Today: Statistics) (Beijing, Zhongguo shehui kexue chubanshe, 1990). However, the sharp decline in output and rise in mortality is indisputable; we cannot be precise in analysing the famine, but we can still say much about its scale and its causes.

4 Thus supporting Sen's famous rejection of FAD as a general theory of famine. See Sen, Amartya, “Food and freedom,World Development, Vol. 1, No. 6 (1989), pp. 769–81; Lin, Justin Yifu and Yang, Dennis T., “Food availability, entitlements and the Chinese famine of 1959–61,Economic Journal, Vol. 110 (2000) pp. 136–58.

5 For procurements, see Bernstein, Thomas P., “Stalinism, famine and Chinese peasants,” in Theory and Society, Vol. 13, No. 3 (1984), pp. 339–78; Ash, Robert F. (ed.), Agricultural Development in China, 1949–1989 (Oxford: Oxford University Press, 1998); Walker, Kenneth R., Food Grain Procurement and Consumption in China (Cambridge: Cambridge University Press, 1984).

6 For communal dining, see Yang, Dali, Calamity and Reform in China (Stanford, CA: Stanford University Press, 1996), and Chang, Gene H. and Wen, G. James, “Communal dining and the Chinese famine of 1958–61,” Economic Development and Cultural Change, Vol. 46, No. 1 (1997) pp. 134. For increased food demand, see Bramall, Chris, In Praise of Maoist Economic Planning (Oxford: Oxford University Press, 1993); Han Dongping, “Farmers, Mao, and discontent in China,” at The over-consumption argument is controversial because rationing replaced “free supply” in communal canteens before starvation appeared (Thaxton, Catastrophe and Contention in Rural China, pp. 124–25). There is also little econometric support: see Kung, James Kai-sing and Lin, Justin Yifu, “The causes of China's Great Leap famine,” Economic Development and Cultural Change, Vol. 52, No. 1 (2003), pp. 5173.

7 See for example Peter Nolan, “The causation and prevention of famines: a critique of Sen, A.K.,” Journal of Peasant Studies, Vol. 21, No. 1 (1993), pp. 128, and Sen's “Reply” in the same issue.

8 Shuji, Cao, Da jihuang (The Great Famine) (Hong Kong: Time International, 2005), p. 282; Jisheng, Yang, Mubei (Tombstone) (Hong Kong: Cosmos Books, 2008), p. 904; Chang, Jung and Halliday, Jon, Mao: The Unknown Story (London: Jonathan Cape, 2005); Dikötter, Frank, Mao's Great Famine (London: Bloomsbury, 2010).

9 Dikötter's estimate is not new, as he admits (Dikötter, Great Famine, p. 333). Rather, it is simply a re-statement of the claim made by Chen Yizi and reported in Becker, Jasper, Hungry Ghosts – China's Secret Famine (London: Murray 1996), pp. 271–72. Chen's claim is suspect; he collected the data whilst a member of an investigative team appointed by Zhao Ziyang in the early 1980s which had the remit of discrediting the Maoist regime (and hence consolidating the political position of the reformers). And some of Dikötter's “new” provincial estimates have been in the public domain for many years: for Sichuan, for example, see Bramall, , In Praise of Maoist Economic Planning, pp. 296–97. More generally, Dikötter's claim that the Chinese archives provide more reliable data on mortality than published data is problematic. He asserts that archival reports can be trusted because “There was no political advantage to be had from declaring extra deaths” (Dikötter, Great Famine, p. 332) but this ignores how officials exaggerated deaths to evade or diminish state-imposed grain procurements.

10 For under-reporting, see Banister, Changing, pp. 233–35.

11 Bernstein (“Stalinism, famine and Chinese peasants”) has long adhered to this view.

12 The verdict was reversed after Mao's death; see Han Dongping, “Farmers, Mao, and discontent in China.”

13 The revisionist approach to the Chinese famine has been influenced by Scott's description of how authoritarian states cause famine by pursuing a “high modernist” agenda designed to imposed order upon an “anarchic” countryside; see Scott, James C., Seeing Like a State (New Haven: Yale University Press, 1998). In the Chinese context, it can be contrasted with Shue's view that the “reach” of the central Maoist state was rather limited; see Shue, Vivienne, The Reach of the State (Stanford, CA: Stanford University Press, 1988).

14 Joseph, William A., “A tragedy of good intentions: post-Mao views of the Great Leap Forward,Modern China, Vol. 12, No. 4 (1986), pp. 419–58; Thaxton, Catastrophe and Contention in Rural China, p. 325.

15 For the Henan famine, see Felix Wemheuer, “Dealing with responsibility for the Great Leap famine in the People's Republic of China,” The China Quarterly, No. 201 (2010), pp. 176–94, and Becker, Hungry Ghosts. According to Chen, “Anhui's Zeng was a blatant political radical who almost single-handedly damaged Anhui” (Chen Yixin, “Under the same Maoist sky: accounting for death rate discrepancies in Anhui and Jiangxi,” in Manning and Wemheuer, Eating Bitterness, p. 216). This was because Zeng played a key role in promoting massive irrigation projects in Anhui and, because they absorbed a large proportion of the farm workforce, these projects contributed significantly to the output decline.

16 Thaxton, Catastrophe and Contention in Rural China, p. 325.

17 For examples cited by revisionists, see ibid. and Manning and Wemheuer, Eating Bitterness.

18 Provincial-level revisionist studies make little systematic use of the data and this limits their usefulness. See for instance Chen, “Same sky,” in Manning and Wemheuer, Eating Bitterness.

19 Wemheuer (“Dealing with responsibility for the Great Leap famine,” p. 178) is aware of this problem and claims that he is not assigning responsibility, but simply describing how memories of the famine vary across different social groups. Nevertheless, it is clear that he assigns most of the blame to the Party centre.

20 Kueh, Y.Y., Agricultural Instability in China (Oxford: Clarendon Press, 1995), pp. 207, 216, 224.

21 Daniel Houser, Barbara Sands and Erte Xiao, “Three parts natural, seven parts man-made,” 2005, at, p. 23.

22 I focus on output and mortality data for the counties because that is the lowest administrative level at which systematic analysis is possible.

23 By “Sichuan” I mean the province of the early 1960s, which comprised the cities of Chengdu, Chongqing and Zigong, and 15 special districts (zhuanqu, abbreviated as district throughout this article). For simplicity, “Chengdu district” includes Chengdu city and the counties of Wenjiang district, “Chongqing district” includes Chongqing city, the counties of Baxian, Qijiang and Changshou, and the counties of Jiangjin district, and “Yibin district” includes Zigong.

24 Note that the death rate fell back to its 1957 level in 1962; for this reason, I focus on 1958–61.

25 For the literature on Sichuan, see Cao Shuji, The Great Famine, pp. 193–215 and Yang Jisheng, Tombstone, pp. 161–228; Cao Shuji, “1958–1962 nian Sichuan sheng renkou siwang yanjiu” (“Research on mortality in Sichuan province 1958–1962”), Zhongguo renkou kexue (Population Science in China), No. 1 (2004) pp. 57–67); Bramall, In Praise of Maoist Economic Planning, pp. 281–334; Stephen Endicott, Red Earth (London: I.B. Tauris, 1988), pp. 51–67. Ding Shu, “1960 nian qianhou Sichuan sheng de fei zhengchang siwang” (“Abnormal deaths in Sichuan before and after 1960”), 2009, at; Wang Dongyu (Dong Fu), “Dayuejin shiqi Sichuan sheng dajihuang de teshu” (“Specific causes of the Great Famine in Sichuan”), 2009, at; Wang Dongyu, “Maimiao erqing cai huahuang – Chuanxi dayuejin jishi” (“Ripening wheat seedlings and golden cauliflower – a record of the Great Leap Forward in western Sichuan”), 2000, at For the political background, see Goodman, David S.G., Centre and Province in the People's Republic of China (Cambridge: Cambridge University Press, 1986).

26 SSB, Quanguo gesheng zizhiqu zhixiashi lishi tongji ziliao huibian (Collection of Historical Statistical Materials on China's Provinces, Autonomous Regions and Centrally-Administered Cities) (Beijing: Zhongguo tongji chubanshe, 1990), pp. 2, 690 (LSTJ 1990).

27 9.4 million compared with 6.3 million according to Cao Shuji, The Great Famine, p. 282.

28 The mortality crisis began in Sichuan in 1958, even though there was no apparent decline in per capita grain output (see Bramall, In Praise of Maoist Economic Planning, pp. 309–13). However, because of space constraints, I concentrate here on 1959–61, when most deaths occurred.

29 Urban grain availability was higher but mortality nevertheless still rose significantly.

30 This includes an allowance for animal feed and seed, and assumes the consumption of additional calories from meat, vegetables and fruit (see Ash, Agricultural Development, p. 141)

31 Of the 103 counties for which I have data for 1957 and for 1959–61, only eight (all located in mountainous western Sichuan) reported a rise in average per capita output whereas 95 reported a decline. For the sample, the average decline in per capita output was 28%, similar to the figure for the entire province (32%).

32 Villagers across Sichuan responded to famine by consuming brown rather than white rice, and by replacing wheat and rice with calorie-dense tubers. A kilogramme of unhusked grain thus translated into more calories in 1961 than in 1956–57.

33 I have used data on CDRs for 123 Sichuan counties to construct Figure 1, and the district median as a proxy where data are missing. CDRs provide an imprecise measure of the demographic impact of famine, but population growth rates between 1957 and 1964, and survival rates by age and county at the time of the 1982 Census, confirm the spatial patterns shown.

34 Sichuan tongjiju (Sichuan Statistical Bureau), Sichuan tongji nianjian 1990 (Sichuan Statistical Yearbook 1990) (Beijing: Zhongguo tongji chubanshe, 1990) (SCTJNJ 1990), p. 58; tongjiju, Chongqing (Chongqing Statistical Bureau) Chongqing jianshe sishinian (40 Years of Building Chongqing) (Beijing: Zhongguo tongji chubanshe, 1989); Chengdu tongjiju (Chengdu Statistical Bureau), Chengdu tongji nianjian (Chengdu Statistical Yearbook 1999), at

35 Sichuan sheng renkou pucha bangongshi (Sichuan Population Census Office), Sichuan Zangzu renkou (The Tibetan Population of Sichuan) (Beijing: Zhongguo tongji chubanshe, 1994), p. 172.

36 In the counties of Pixian (175 per 1,000 in 1960), Shizhu (168 in 1960), Rongxian (165 in 1960), Fengdu (163 in 1960) and Yingjing (151 in 1959); see Jisheng, Yang, Tombstone, p. 533 for Rongxian and Shizhu. Other rates from Xianzhi (County Records) (XZ): Pi XZ (Chengdu: Sichuan renmin chubanshe, 1989), p. 131; Yingjing XZ (Chongqing: Xinan shifan daxue chubanshe, 1998), pp. 111–13; Fengdu XZ (Chengdu: Sichuan kexue jishu chubanshe, 1991), pp. 9495.

37 Yang Jisheng, Tombstone, p. 533.

38 Zhongguo kexueyuan Chengdu dili yanjiusuo (Chengdu Geographical Research Institute, Chinese Academy of Science), Sichuan nongye dili (An Agricultural Geography of Sichuan) (Chengdu: Sichuan renmin chubanshe, 1980).

39 Bramall, In Praise of Maoist Economic Planning, pp. 282–91. The infant mortality rate in 1937 was seemingly higher than in 1959 in Shifang county on the Chengdu plain; see Lavely, William R., “The rural Chinese fertility transition,” Population Studies, Vol. 38, No. 3 (1984) pp. 365–84.

40 Rainfall and temperature data come from the 205 and 160 station national sets at and respectively. Some of Sichuan's County Records (e.g. Shehong, Pixian and Anyue) also report rainfall and temperature data.

41 These five stations (Pengshui, Nanchong, Chengdu, Yibin and Bazhong) together cover the key grain-producing parts of the province. I ignore western Sichuan, where grain production and population density were low. For aridity indices, see Oury, Bernard, “Allowing for weather in crop production model building,” Journal of Farm Economics, Vol. 47, No. 2 (1965), pp. 270–83. Here I use the de Martonne index, but the conclusions are insensitive to the index used.

42 The late 1950s and early 1960s were extraordinarily wet across the western Sichuan basin (including Chengdu). However, the County Records make little mention of damage caused by flooding; drought was evidently the main problem.

43 The 1958 mortality crisis was not due to output decline, and both high procurements and increased food demand contributed significantly to mortality during 1959–61.

44 For example, I proxy the supply of fertilizer using the pig stock (chemical fertilizer use was negligible in the late 1950s). The pig stock was certainly a determinant of grain output (more pigs meant more fertilizer). However, causality also ran in reverse: increased grain output meant more pig feed.

45 Dianjiang XZ, p. 338; Fuling XZ, p. 379.

46 Li escaped much of the official blame for the Sichuan catastrophe but many scholars are less charitable; see Ding Shu, “Abnormal deaths in Sichuan,” and Wang Dongyu, “Specific causes of the Great Famine in Sichuan.”

47 Goodman, Centre and Province in the People's Republic of China, pp. 50–51.

48 Ash, Agricultural Development, p. 130.

49 Shifang XZ (Chengdu: Sichuan renmin chubanshe, 1988), pp. 1020, 18–55. Of this, 30 million kgs went to the USSR.

50 Bramall, In Praise of Maoist Economic Planning, pp. 322–24.

51 LSTJ 1990, pp. 3, 691

52 The scale and impact of migration during the famine has been little explored. However, it appears that the controls on migration imposed by the Party Centre helped to intensify the scale of mortality. By preventing able-bodied workers from returning to their home villages, the harvest was lower than it would otherwise have been, especially in 1958. Moreover, the rationing of food supplies via communal canteens made it hard for “outsiders” to migrate to communes with higher per capita food supplies: such outsiders had no entitlement to food and therefore migration offered little panacea.

53 Chao, Yang, Dangdai Zhongguo de Sichuan (Contemporary China: Sichuan) (Beijing: Zhongguo shehui kexue chubanshe, 1990), p. 93fn.

54 Ibid. p. 91.

55 Ibid. p. 91.

56 Goodman, Centre and Province in the People's Republic of China, pp. 150, 151, 236.

57 Ibid., pp. 138, 144, 154, 158.

58 Dongshuitian were prevalent in Sichuan; e.g. they comprised 70% of paddy fields across Fuling district in the mid-1950s; see Afanas'yeskiy, Ye.A., Szechwan (New York: JPRS Translation No. 15308, 1962), p. 301.

59 The abolition of dongshuitian contributed greatly to famine because, until the irrigation schemes begun in the 1950s were completed, farmers had no recourse to alternative water supplies in the event of severe droughts.

60 SSB, Materials on 50 Years of Agriculture, p. 34; SCTJNJ 1990, p. 130.

61 Although we lack proper evidence, the response of local cadres towards the policies of the Party centre in respect of migration was also crucial. In counties where local cadres limited the exodus of able-bodied workers to work on industrial projects outside the county boundaries, the 1958 harvest would have been higher. Conversely, cadres willing to accept large numbers of in-migrants ran the risk of pushing up mortality by exhausting local food supplies. In other words, the Party centre was responsible for imposing the institutions of collective farming and communal dining, but the interpretation and implementation of these policies by local cadres was critical in determining local mortality rates.

62 Pi XZ, pp. 19, 131, 253–54, 490, 492.

63 Guan XZ (Chengdu: Sichuan renmin chubanshe, 1991), pp. 121–22, 130, 460.

64 Fuling XZ (Chengdu: Sichuan renmin chubanshe, 1995), p. 233; Dianjiang XZ (Chengdu: Sichuan renmin chubanshe, 1993), p. 131; Ba XZ (Chongqing: Chongqing chubanshe, 1994), p. 656; Changshou XZ (Chengdu: Sichuan renmin chubanshe, 1997), p. 178.

65 Yingjing XZ, p. 112.

67 Yingjing XZ, pp. 281, 453. Rao Qing, Party secretary in 1960, was later sentenced to eight years in a labour camp for his failures.

68 Ó Gráda is right that the famine generally hit hardest in provinces which were poor, such as Sichuan, Anhui and Henan; see Gráda, Cormac ó, Famine (Princeton, NJ: Princeton University Press, 2009), pp. 252–54. However, my results show little correlation between output per head and mortality at a sub-provincial level in Sichuan.

69 In this set-up, I use average output for 1956–57 to smooth output anomalies in 1957. I employ adjusted grain output data because reported output provides a misleading impression of food supplies by taking no account of differences in output composition or husking rates; raw grain data thus exaggerate the level of per capita output in (say) the rice-producing counties of the Chengdu plain relative to the corn and potato-producing counties of the central Sichuan basin. Peripheral counties are those located outside the central riverine zone and away from the provincial railway network.

70 Land reform was not attempted here until the late 1950s because these counties were so poor, and because of separatist sentiment amongst China's Tibetan population in the early and mid-1950s.

71 A point emphasized in Chen, “Same sky,” pp. 197–225.

72 The evidence for Sichuan during the famine thus supports Shue's (The Reach of the State, p. 131) view that “far from serving as robotic handmaidens of central domination, these stubborn, savy, and often cynical local officials came to constitute a formidable obstacle to real and effective central penetration and control on the ground.”

* I am grateful to Kerstin Lehr and Julia Strauss for their comments on an earlier version of this article, to Tim Wright for alerting me to a number of sources, and to Kimberley Manning for allowing me an early look at the manuscript version of Manning, Kimberley E. and Wemheuer, Felix, Eating Bitterness: New Perspectives on China's Great Leap Forward and Famine (Vancouver: UBC Press, 2010).

Recommend this journal

Email your librarian or administrator to recommend adding this journal to your organisation's collection.

The China Quarterly
  • ISSN: 0305-7410
  • EISSN: 1468-2648
  • URL: /core/journals/china-quarterly
Please enter your name
Please enter a valid email address
Who would you like to send this to? *



Altmetric attention score

Full text views

Total number of HTML views: 0
Total number of PDF views: 0 *
Loading metrics...

Abstract views

Total abstract views: 0 *
Loading metrics...

* Views captured on Cambridge Core between <date>. This data will be updated every 24 hours.

Usage data cannot currently be displayed