Abstract
Elitism argues that the values and experiences of the political elite shape policy, while institutionalism contends that an individual's behaviour is constrained by institutional settings. This article shows that both perspectives work well overall, while offering persuasive arguments, but the former is dominated by the latter. This explains the reluctance of leaders to equalize education if doing so means sacrificing their future promotion. Although provincial leaders have substantial influence on education equalization, such influence is the consequence of the Chinese Communist Party's (CCP) personnel regulations. Two distinct patterns are revealed: whereas governors, being more promotable, are more likely to be significantly constrained by the “promotion rule”, party secretaries, as more terminal officials, are more likely to be restricted by the “rule of retirement”. A heteroskedastic linear model that takes into account the influence of the speed of a leader's promotion on the variance of their behaviour shows that the “faster runners” – those who are more quickly promoted – in a hierarchy are also the faster learners in their institutions and their behaviour regarding education equalization is more convergent.
Acknowledgements
This research was supported by the Asia Research Centre at Fudan University. For their comments on an earlier draft of this article, the author would like to express his gratitude to John P. Burns at the University of Hong Kong, John Armstrong at Fudan University, and the three anonymous reviewers.
Notes
1. The research for this project was mainly conducted in 2007–2008 when the author was a Fulbright Visiting Fellow at the Fairbank Centre, Harvard University.
2. The greater the Coefficient of Variance (CV), the more unequal the education will be. The measure will be explained in detail later.
3. As will be discussed later, other factors play a role, such as the key school policy. This offers an institutional incentive for officials to distribute educational recourses unequally. As the children in their families will enrol in these so-called key schools, there is also an element of self-interest.
4. Moreover, the children of families of the top county leaders usually study at the county's key schools (see Wang, Citation2003b). Most provinces have a couple of these provincial-level key schools in which the children from top families are enrolled.
5. In fact, the CCP's centre (the Organization Department) manages all vice and full provincial officials, even non-CCP officials, as long as they attain such ranks. The system works depending on an official's political rank rather than political identity.
6. However, the study does not suggest that provincial leaders would have had a natural tendency to equalize education had it not been for their career concerns. The given argument just implies that without concerns for the potential negative effect of education on their career, the leaders would be more likely to follow Beijing's formal policies on education equality and their personal values would be more likely to be expressed.
7. For the case of Zhejiang Province, see http://www.zj.gov.cn/gb/zjnew/node3/node22/node165/node1749/index.html.
8. In statistics, CV is a measure of dispersion in a probability distribution. It is defined as the ratio of the standard deviation to the mean. It is often reported as a percentage by multiplying the above calculation by 100. It is a useful statistic for comparing the degree of variation between one data series and another, even if the means are drastically different. The higher the CV, the higher the variability of the data. Thus, in this study, the higher the CV, the more unequal the case is.
9. The data includes the county-level city but omits county-level urban districts since the data for districts is unavailable. Statistically, this might underestimate the intra-provincial inequality in education, since the urban-rural gap in education spending also contributes to the total inequality within a province.
10. By this definition, the value of each leader's promotion speed is constant from 1994 to 2001, while the value of the rule of retirement for each leader will change annually. The correlation between these two variables is less than .35. There will be no significant collinearity if we put both of them in a regression model.
11. The baseline (omitted) is the year 2001.
12. It was estimated using a Maximum Likelihood Heteroskedastic Linear Model. A Maximum Likelihood Normal Regression Model assumes yi is normally distributed. However, in this homoskedastic case, a subscript is added to the variance, producing the model below. In principle, the variance now is unique to each observation, as are the expected values (xiβ). The statistical analysis was done using the R statistical program, and the program packages of MLLibrary and MLhetReg, written by Charles H. Franklin at the University of Wisconsin–Madison.
13. Teachers, as intellectuals, seriously suffered under many campaigns initiated by Mao in the 1950s and 1960s. Most officials in this study were born in the 1930s and 1940s.