JOURNAL OF POLITICAL ECO years, respectively. This compares to the average tenure of currently serving Chinese managers in our sample of 5.5 years. Thus Chinese managerial turnover in the 1980s appears to be slightly more fre. quent than that of managers in the United States and Japan. Since one of the key features of an active, mature market is a high number of transactions, this evidence on turnover of managers is supportive of the idea that there is developing in China a labor market for mana genial resources Active managerial turnover contrasts sharply with limited turnover of production workers in China. Most ordinary workers enjoy lifetime employment and almost never change jobs. For example, in our sam le,in 1984, only 3.7 percent of the workers quit, were fired, or were transferred(1.3 percent retired); in 1989, only 2.8 percent quit, were fired, or were transferred (1.3 percent retired). (An average tenure or managers of 5.5 years corresponds roughly to an 18 percent arly turnover. )It is worth stressing, then, that while ordinary Chi nese workers rarely change jobs, Chinese managers in the 1980s fre- quently did. This activity contrasts sharply with the conventional view of a relatively rigid management structure B. Manager Selection and Contract Terms Both before and after the reforms, the managers of Chinese state- owned firms were selected by the industrial bureau charged with regulating the firm. There is a classic princip al-agent relationsh between the industrial bureau and the manager. The agent-the pro- spective or current manager--has some relevant information that ot known by the principal-the industrial bureau. This information may concern the manager's own abilities or the firm's potential pro- ductivity. Either kind of information can generate adverse selection In addition, the manager may take actions that affect the firms pro- ductivity and cannot be directly observed by the industrial bureau moral hazard is present. The bureau cannot know whether poor per formance by the firm is attributable to(a) inherently low productivity firm,(b)managerial incompetence, (c) managerial decisions that pursue goals other than productivity, or(d)bad luck The selection of a new manager provides an opportunity to observe the principal's actions designed to motivate the agent. For example, the treatment of the former manager should be related to firm per Thus a model of the firm similar to that of Laffont and Tirole(1986)applies here xpected to be at least as ill informed as the industrial bureau, with one exception, namely, the benefit of hindsight: a firm's ex post perfor