Estimation Of Total Factor Productivity Of Different State Firms

China’s constant economic development requires urgently the reform of enterprise ownership, especially the state-owned enterprises (SOEs). Ownership reform has been one of the core problems facing China’s economic transformation, and the key point of this problem is the enterprise production efficiency. It is generally considered that the production efficiency of SOEs is lower than non-state-owned ones. The production efficiency has been used, in recent years, to measure whether the ownership reform is effective.Total factor productivity (TFP for short) is generally defined as the "surplus" of total output that cannot be explained by input elements. TFP includes not only science and technology progress, but also management skills, organizational innovation, knowledge level of material production and system environment. It is a comprehensive measurement of enterprise production efficiency.Micro data being increasingly accessible, TFP calculation at the micro level gets growing application. However, it is seldom used in domestic to measure enterprise reform effectiveness. This paper will calculate from the micro level the TFP of different ownerships, and discuss enterprise production efficiency and ownership reform effectiveness. It is expected to be meaningful for China’s economy development.This paper first reviews the micro calculation method of TFP:the OLS estimation method, the fixed-effect estimation method, the OP estimation method(Olley and Pakes,1996)and LP estimation method(Levinsohn and Petrin,2003). Advantages and disadvantages of these micro methods in solving the simultaneity bias and sample selection bias are analyzed and compared. Next, these methods are used to calculate TFP of China’s manufacturing firms during the period from1999to2007, and the results are comparatively analyzed. Then, this paper divides these enterprises into SOEs and non-state enterprises, and non-state enterprises into collectively owned enterprises (COEs), privately owned enterprises (POEs) and foreign-invested enterprises (FOEs). TFP of each enterprise type is calculated and the results are compared both horizontally and vertically. The results show that the OP method and LP method can effectively solve the simultaneity bias and sample selection bias that exist in OLS estimation, but the LP estimation method, whose proxy variables is the intermediate inputs, is not significantly better than the OP estimation method, whose proxy variables is the intermediate investment. Horizontally speaking, the average TFP absolute level of non-state owned firms is higher than that of SOEs, but the average growth rate of SOEs’ TFP is higher than that of non-state owned firms. Vertically speaking, the TFP of both SOEs and non-state owned enterprises is growing annually except for certain individual years, but the growth rate is low.