The sent-down youth of China: The role of family origin in the risk of departure to and return from the countryside

Abstract

Between 1967 and 1978, over 17million urban youths in China systematically migrated to the rural countryside in a massive relocation movement known as the Sent-Down Movement. The youths who relocated—some by choice, though many forcibly—were part of a grand scheme envisioned by the then ruling Communist Party leader, Mao Zedong, who sought to reeducate urban youth by having them live and labor amongst their rural compatriots. Known as the “sent down youth”, these youths' experiences and struggles of moving to and returning from the countryside offer considerable insight into the ideological importance of family origins. Most notably, the revolutionary movement which swept over China at the height of the sent-down movement in the late 1960s reversed the hierarchical order of society; individuals with higher family origins were now criminalized making them prime targets for relocation to the countryside. In this quantitative analysis, I examine the relationship between family origins and the risk of departure to and return from the countryside for urban youth, using a unique longitudinal retrospective dataset, Life Histories and Social Change in Contemporary China. I analyze how this relationship plays out not only during the height of the movement, but throughout a more expansive time frame under which youths were reportedly sent-down (1957 to 1980). Through discrete-time survival analysis, I estimate that urban youths from higher family origins (rich peasant and landowner classes) experienced a greater risk for being sent-down versus their counterparts from lower family origins. Most interestingly, youths of higher family origins also experienced a lower risk for returning from the countryside; one interpretation of this finding is that even after spending time in the countryside in the pursuit of absolving themselves of their higher family origins, the stigma attached to their higher family origins persisted.

Keywords:

• China
• Sent-down youth
• Family origin
• Cultural Revolution
• Discrete time survival analysis

Acknowledgements

The author would like to thank Professor John B. Willett, colleagues at the Harvard Graduate School of Education and an anonymous reviewer for their insightful comments on previous drafts of this work. All errors are my own.

Notes

1 I use the term “family origin” which is often used interchangeably with the term “social origin.” It is through one's family origin, which is derived from the patriarchal line one generation above, that one is hierarchically ranked into social classes. (Kuhn, 1984)

2 During the revolutionary movement (the mid- to late-1960s), individuals were labeled “bad” if their backgrounds were associated with either landlord, rich peasant or counter-revolutionary classes (Unger, 1984). Landlords were the considered the lowest among the “bad” origins hierarchy—to the extent that they were, “treated like lepers.” (Unger, 1984: 124. For a more detailed discussion of the “bad” class distinction, see Unger (1984: 124–141).

3 According to Berstein (1977a) and Zhou (2004), the sent down policy was also practiced during the pre-Great Leap Forward era (1958–1960) continuing until the early 1960s, suspended briefly between 1966 and 1968 and resumed in full force between 1968 and 1977. Pan (2003) documents sent-down youth as early as 1953 and as late as 1980.

4 Zhou (2004) uses the term “exploiting class” to classify individuals who were, “large business owners, landlords, “rightists,” “bad elements,” etc.” (Zhou, 2004: 60).

5 Data is publicly available at: http://www.sscnet.ucla.edu/issr/da/lhsccs/chinaweb.html.

6 For a more detailed information about sampling procedures, see Treiman (1998). The sampling procedures as outlined here are derived from Treiman (1998).

7 I also fit several other models not shown in Table 4 including one with interaction variables between the family origin predictors and time. Based upon the goodness of fit statistics, I concluded that these interactions did not significantly improve the fit over Model 1B.

8 Goodness of fit statistics show that though the fourth-order polynomial specification for time is a better fit over the cubic specification for time (Δ−2LL=10.18; χ ² (d.f.=1, α=.05)=3.85), I decided to use the most parsimonious specification for TIME that best smoothed out the fluctuations in the fitted hazard function.

9 Note that of the demographic controls, the parameter only on gender and POST_1969 remained statistically significant. (p<.001, α=.05). I also fit several other models not shown in Table 4 including one which includes interactions between family origin predictors and time. Based upon the goodness of fit statistics, I concluded that these interactions did not significantly improve the fit over Model 2B.

10 This model is not shown in Table 4.