Building original series of physical capital stocks for China's economy methodological problems, proposals for solutions and a new database

Abstract

There are to date no official Chinese statistics relating to capital stocks. This lacking data hinders econometric studies of growth in this country. Series of such stocks are proposed in the literature, but most available empirical work on this topic suffers multiple deficiencies. The purpose of this article is to build the most reliable and longest possible statistical series of capital stocks for China. Our initial capital stocks are calculated on the basis of an output-capital ratio which is less approximate (and lower) than those generally provided. Our investment flows are consistent with the perimeters of the initial stocks. Our investment price indices are strictly tailored to the content of these stocks, and the unit root tests show that all the indices are non-stationary and cointegrated to the order of 2. This means they cannot be substitutes, as supposed in many other studies. Our depreciation rates are estimated by type of capital, under assumptions consistent with age-efficiency and retirement. Investment shares are used to approximate an overall capital structure and to calculate a total depreciation rate. Built from 1952 to 2014, our original series are available to econometricians seeking to conduct new empirical studies on China, over the long run.