| dc.description.abstract | In the early empirical studies, many economists focus on the determinant of firm growth but few papers to investigate the time series phenomenon of firm growth. We investigate the Gibrat’s law by using panel unit root test. Panel unit root can increase power in contrast to conventional individual ADF test. At first we use panel unit root test to testify Gibrat’s law under independent and identical distribution. The test results reject the null hypothesis of Gibrat’s law. Independent and identical distribution is not reasonable in real situation. Any firm in a given industry may have some correlation with other firms. The limiting distribution of Im, Pesaran, and Shin(IPS) statistic is invalid and it will produce large distortion. We apply Taylor and Sarno(1998) MADF test to deal with cross-sectional correlation problem and study the issue. We find that the conclusion is not the same.
Next to the finding of chapter 2, we provide further evidence on Gibrat’s law from the panel unit root test of the Taiwan electronic industry. The test results of Gibrat’s law are not the same as previous empirical literature. We use the panel estimation of seemingly unrelated regressions for Augmented Dickey-Fuller tests (SURADF) to consider cross-sectional dependence and heteroscedasticity. The results show that four companies reject Gibrat’s law.
Firm growth will make the change of firm rank. The movement of firm rank is an dynamic indicator for the rigidity of market structure. In Chapter 4, we adopt the top 500 firms as study object, and divides them into 33 industries. We use the probability of the turnover of firm’s ranking in each industry as the basis of measuring industry mobility. We consider Markov chain, Gibrat’s law and three states Markov chain proposed by this study as the comparison basis of the turnover of firms’ ranking. Analytical results indicate that the three states Markov chain has the transition probability that is most close to the ranking true probability. It can be shown that the mobility which is estimated by Gibrat’s law is highest from mobility indicator. In addition, Geroski and Toker (1996) is not suitable for studying Taiwan’s industry mobility by considering top 5 firms in a given industry, it will overestimate industry rigidity.
Chapter 5 uses the Industry, Commerce and Services (ICS) sampling data to surdy the relationship between firm growth and R&D spillover effects considering firm exit behavior. There are a few papers to investigate this issue. We divide the R&D spillover effects into inter-industry and intra-industry spillover effects and apply the Heckman two-stage model to deal with empirical research. We find that firm age, total R&D expenditure, whole industry sales, advertising, and employee’s salary have positive contribution to firm survival. Firm size, inter-industry spillovers, intra-industry spillovers, export, industrial growth, and capital/labor ratio also have positive contribution to firm growth.
The final Chapter is based on an endogenous growth model and we divide an innovation sourcing strategy into an internal source and an external source. With respect to the determinants of the decision by the innovative firm to produce technology itself or to source technology externally, the representative firm maximizes its profit function and the first-order condition can be derived. Based on our model, the optimal ratio of an internal technology-sourcing strategy and external-sourcing strategy can be found. The technology stock of the representative firm to total industry, the technology absorption capacity, and the internal and external R&D efficiencies will all influence the optimal ratio. | en_US |