Industrial policy, technology advances and industry development: Evidence from the Chinese software industry.

摘要

The idea that government should steer economic development by strategically picking and managing certain industries is controversial but appealing to many developing countries that are eager to upgrade their industries. I investigate the impact of activist industrial policy on the development of the industry it targets by studying the case of the Chinese software industry.;To keep the problem tractable, I focus on the government sponsored science and technology (S&T) programs that recruit all organizations in the software industry, regardless of the ownership type and profit orientation, to develop cutting edge technology. I evaluate the effectiveness of these programs in terms of the performance of the affiliated software development efforts, the building of collaboration networks among the software producers, and finally the strategic responses of these organizations to program participation.;From the data archive of China's Ministry of Science and Technology, I gather a unique dataset of 3437 software development projects which were completed between 2000 and 2004. I supplement the dataset with substantial follow-ups, including semantic analysis of the project description texts and an expert panel survey. I employ a propensity score matching method to benchmark the program-affiliated projects with those started by software producers themselves. As a result, this study offers the first large-scale systematic analysis of the Chinese software industry.;I find that the Chinese government's activist industrial policy to promote the country's software industry---in particular, the various government sponsored S&T programs---has a modest negative effect on the performance of the software development efforts. Government initiated projects are 4% to 14% less likely to get adopted than projects initiated by software producers. When the program affiliated projects are carried out by the State Owned Enterprises (SOEs), it takes six to seven months longer for the projects to be completed.;On the other hand, the government sponsored S&T programs are shown to be instrumental in generating momentum for endogenous growth in terms of building collaboration networks among software producers. These programs are 4.7% more likely to induce inter-organizational collaboration events than market forces.;Finally, the general institutional environment matters. The more liberal and market economy oriented, the more likely that software producers will start their own projects instead of getting involved in the central government sponsored S&T programs. Relaxing financial constraints for entrepreneurs in the non-state sector improves the likelihood of community integration. In particular, universities and research institutes are more likely to venture with partners from the business world and pursue their own research agendas.