This paper looks at how China’s state-owned enterprise (SOE) privatizations changed the kind of productivity firms use. The author measures three separate kinds of productivity in Chinese manufacturing from 1998 to 2008: the kind that makes capital (machines and equipment) more productive, the kind that makes labor (workers) more productive, and the kind that makes materials (intermediate inputs) more productive. The main finding is that technological change over the decade was strongly biased toward capital and labor. Privatization produces a clear short-term rise in labor-focused productivity at the time of the transition, but no lasting increase in capital-focused productivity and a small decline in material-focused productivity.

To get at these differences the researcher develops a new method to recover the three factor-specific productivities from firm data. The approach starts from a common production model called CES (constant elasticity of substitution), which lets inputs be substituted for one another in a controlled way. Using firms’ cost shares and basic cost-minimizing conditions, the method rewrites the model as three simpler equations and estimates them together with a dynamic panel technique. Productivity is treated as a process that depends on its past values and on ownership status, and the paper uses timing rules in the reform to help separate cause and effect.

On levels, non-state firms (private, collective, and foreign) operate above SOEs on capital- and labor-focused productivity. SOEs, however, appear more productive on materials. The author interprets that pattern as partly reflecting SOEs’ preferential access to intermediate inputs under administrative allocation. Over the decade the capital cost share in manufacturing fell substantially — a decline concentrated among SOEs and firms that privatized rather than among always-private firms.