Hypothesis 1.
The probability of promotion (termination) for provincial leaders increases (decreases) with the provincial economic performance.
Besides the GDP growth rate, the characteristics of provincial leaders may affect their likelihood of promotion and termination. Provincial leaders' personal attributes such as education, age and tenure of office are included in our estimations to control their effects on turnover.11 Education, an indicator variable that equals one for leaders with college education and zero otherwise, measures a leader's human capital, and thus we expect it to have a positive (negative) effect on promotion (termination). Age has become a critical variable determining turnover, especially terminations after the age-based retirement rule was implemented in 1982. To capture the potential nonlinear effect of age on the probability of turnover, and in particular the effect of the 65-years-of-age retirement rule, we add a dummy variable “age65”, which equals one if the leader is 65 or older and 0 otherwise. We also control tenure in the regressions, which measures how many years a leader has been in the post.
Provincial leaders' connections with the central government could also affect the likelihood of turnover. A provincial leader's experience in the central government may allow her/him to maintain stronger connections with the center and better knowledge of the workings of central appointment procedures, which will result in better turnover prospects. The central experience can also help the leader cultivate informal connections with central leaders who can influence personnel evaluations.12 Although connection with the center is so important, to the best of our knowledge, no empirical study has been undertaken that directly links this factor to the career mobility of provincial leaders. We will include central connection in our empirical analysis to examine the effect of political connections on career mobility. The central connection variable is an indictor that equals one for leaders having previous experience or holding joint-appointment in the central government, and equals zero otherwise.
Our regressions also take into consideration the effect of both provincial characteristics and cyclical policy shocks common to all provinces.13 The level of development of a province could affect the career prospects of its leaders. We use the lagged provincial per capita GDP to control for this potential effect. The provincial location could also matter in a leader's likelihood of promotion. For example, starting from 1979, the central government introduced a variety of preferential economic policies that favor coastal provinces, such as Guangdong and Fujian. One may wonder whether the central government has a special preference towards the leaders in those coastal areas. In order to control for the potential bias of turnover decisions in favor of certain areas, we include a set of provincial indicators in x.
Evidence shows that central policies regarding personnel management change over time. For example, our data show that there were two large waves of retirement in 1983 and 1985 after the central government implemented the mandatory retirement rules (see Section 4). There is also a cyclical pattern in that many personnel changes occur around the time of the party congress and people's congress held every 5 years. To control for the effect of policy changes over time and political cycles, we include a set of year indicators in estimating the ordered probit model.14
To summarize, we have the following control variables in the ordered probit model that estimates the effect of economic performance on turnover: the leader's age, age65, education, tenure, a central connection indicator, the lagged provincial per capita GDP, and a whole set of provincial and year indicators.