Why do people want to change jobs? A survival analysis on job turnover in China

China introduced a household registration system (known as hukou) in 1950s, which initially used to collect and manage residential information. By having different types of registration, the population is divided into urban and rural parts. However, the labour market is segregated by this system as well, preventing an efficient flow of labour and achievement of full benefits from mobility. In the pre-reform Chinese urban labour market, jobs were allocated to urban residents either from local communities or educational institutions. Employers were mainly in the public sector, which includes state-owned or collective-owned enterprises and government departments. These employers were not allowed to recruit and dismiss employees freely as they wanted. They should follow an employment plan made by the central government. By contrast, from employees? side, they should accept the allocated job offers and cannot make decisions on job turnover. Though jobs were secure in the pre-reform labour market, disincentives and skill mismatch might occur. However, life-long employment was gradually broken after introducing the policy of reform and opening up in 1978. For example, the state-owned enterprise reform in 1990s led to huge downsizing and reorganisation. In addition, universities abolished the policy to guarantee life-long employment for graduates and expanded the recruitment scale in late 1990s, which added competition and pressure to the Chinese labour market. By contrast, in rural, surplus labour force was created after adopting a household responsible system for agricultural production in 1978. Therefore, many township and private-owned enterprises were established to absorb surplus rural labour. Some rural people also migrate into urban for jobs after the restriction on migration was eased. The labour mobility is associated with job turnover, which motivates this study on job turnover in China. Employment histories of respondents were asked in the China General Social Survey 2008, in which, there are 1750 pieces of uncensored employment records (58.59% of the total) with an average duration at 6.18 years. Using this micro-level data, a discrete-time duration model is built to find determinants on job turnover. As urban and rural people may have different concerns about job turnover, it is necessary to study them separately. However, similar results are found after controlling for the household registration type in the estimation. Firstly, non-parametric analysis supports the fact that most of new jobs end early and the probability of job turnover decreases with job tenure. Secondly, the probability of job turnover is high among young people. Thirdly, social insurances are the most significant determinants on job turnover for both urban and rural people. Based on these findings, this study concludes that urban and rural people change jobs in return for more secure jobs. Although this research is designed from the labour supply side, the implication is that employers can design better jobs so as to reduce employment costs from job turnover.