CES - Discussion paper series, DPS14.15 pages:1-35
Over the reform period, industry has been the source of forty percent of GDP, and has contributed 90% of China’s exports. Annual firm level surveys that begin in 1992, along with industry-wide census in 1995, 2004 and 2008 are rich sources of data on firms’ actions in this important sector. It is well-known that working with Chinese data requires overcoming challenging measurement issues. Macroeconomic series are often suspected to suffer from political interference or reporting biases that stem from political incentives. Working with the firm-level data has its own challenges. Making sure that comparisons over time are consistent is perhaps the most difficult and pervasive issue. This is because of sampling as well as measurement issues for key variables, such as ownership type, real output, value-added, wages, and the capital stock. These problems are apparent, for example, in discrepancies between the evolution of aggregates from the firm-level data and aggregate statistics in the national income accounts. In this paper, we provide an introduction to these data sets. We discuss and illustrate several of the issues that make comparability over time difficult and we suggest solutions for many of them. The importance of a particular measurement issue often depends on the exact application. We illustrate this point by tracing the evolution of the relative productivity level of entrants and incumbents over time, trying to distinguish between changes in actual performance and changes driven by measurement problems. We conclude by identifying a few promising areas of future research and margins on which collaboration among users to improve these data might be beneficial.