Measuring market concentration in China: the problem with using censored data and its rectification

- We use firm level data to measure China's concentration trend between 1998 and 2009.
- We confirm a systematic bias of concentration estimation with censored survey data.
- We verify applicability of the Zipf distribution for China's industrial firms.
- We develop an alternative measure of concentration.
- China actually experiences much less competition improvement in many industries.

This paper utilizes the non-linear estimation method to simulate the Zipf distribution, and constructs an alternative measure of Hirschman-Herfindahl index (HHI), in order to reveal the real changes in monopoly of China's industrial markets. Based on the annual waves of the Chinese Industrial Enterprises Database between 1998 and 2009, it finds that: 1) systematic bias of deceptive declining concentration would be very easy to appear when directly using censored survey data with some invariant threshold; 2) with method in this article, an alternative measure of China's market concentration (namely, the estimated Zipfian parameter) can be produced to better depict monopoly trend, even though small firms are censored out in the market surveys and commonly used HHI cannot avoid such systematic bias; and 3) China actually experiences much less competition improvement or monopoly reduction in many industries during this period.