Computational analysis of macroeconomic policies on income inequality, growth and intellectual property rights.

摘要

The goal of this dissertation is to develop and to use several computational methods to analyze macroeconomic policies on income inequality, growth and intellectual property rights. First, the inverted-U hypothesis which has failed to capture the development pattern of a newly industrialized country such as Taiwan,--better known as the "Kuznets puzzle"--is analyzed so that we can understand the economic performance of Taiwan through key government policies after the Second World War. Therefore, the "Kuznets puzzle" can be solved by using a more rigorous methodology. Control theory is then used to understand the dynamics of policies in the real world.; In the second essay, a numerical approach is used to evaluate how a country discriminates against foreign patent applicants. The results indicate that the U.S., Japan and Germany discriminate against foreign applicants through different means. The U.S. tends to treat patent applications of nonresidents as "complicated technology" and to prolong the length of pendency of "simple technology" and "complicated technology" for nonresidents. Japan seems to be more fair in classifying "simple technology" and "complicated technology" for residents and nonresidents. However, Japan prolongs the length of pendency of both technologies for nonresidents. Additionally, it seems that getting a patent is more uncertain in Japan. Contrary to the U.S. and Japan, Germany is in favor of nonresident applicants of "simple technology". However, Germany also tends to categorize the applications of nonresidents as "complicated technology" rather than "simple technology". But the average length of pendency is almost the same for residents and nonresidents.; Finally, a time-series index of tightening Intellectual Property Rights (IPRs) is constructed for the U.S. to investigate the effect of IPRs on the efficiency of R&D, the real wage in manufacturing and per-capita patents. Then, the model is solved using stochastic control methods in order to compare Certainty Equivalent (CE) and Open Loop Feedback (OLF). It is concluded that OLF is better than CE especially when more uncertainty is added to the model.