AUTONOMY IN CHINESE STATE ENTERPRISES 189 Chinese state enterprises have permanent jobs, but an increasing number have been hired on fixed-term contracts. It is easier for a manager to refuse to renew a worker' s contract at the end of his term than it is to fire a permanent worker. According to aggregate data, a contract worker in 1989 was six times as likely to have a contract terminated as a permanent worker was to quit or to be fired [State Statistical Bureau 1990a, pp 204, 218 The introduction of an incentive-payment mechanism does not by itself guarantee that a factory' s productivity will rise. It might be that in practice bonuses are paid out equally, regardless of individual productivity, so that they have no incentive effect.Often it is difficult to define adequate output measures, and basing ayment on the wrong measures of performance can be counterpro ductive. Workers might collude against management, subverting attempts to reward good performers by imposing social sanctions on anyone who works too hard. Similarly, although workers on contracts in principle can be laid off at the end of their term so that they have some incentive to exert effort, in practice it may be that their contracts are always renewed, and thus they are effectively the same as permanent workers. Thus, it is necessary to look at the data to see whether the strengthening of worker incentives was real or just apparent and whether the new incentives actually succeeded in improving productivity II. TRENDS IN AUTONOMY, INCENTIVES, AND PRODUCTIVITY The data we use come from surveys conducted by the Institute of Economics, Chinese Academy of Social Science(CASS),in consultation with the authors of this from the University of Michigan and Oxford University. Annual data for 1980-1989 for 769 enterprises in four provinces(Sichuan Jiangsu, Jilin, and Shanxi)give details of the firms'internal incentives, the firms cost and revenue accounts, and the nature of lationship between the firms and the state. The questi naires were sent out by the provincial System Reform Commis- sions (which are responsible for assessing and implementing reform measures)to 800 enterprises, and 769 valid questionnaires were returned. The System Reform Commission does not directly supervise enterprise activity, but it is an official government body with which the enterprise has regular interactions, which may account for the esponse rate. The questionnaire had two parts. The first part, directed specifically to the factory manager