| Summary: | 碩士 === 中原大學 === 國際經營與貿易研究所 === 106 === This thesis rewrites the three-factor model, developed by Fama-French (1993), as a panel smooth transition regression (PSTR) model to evaluate stock returns. First, we add the term of spatial dependence into the three-factor model to form a linear four-factor model and then estimate stock returns. Second, we rewrite the linear four-factor model as a four-factor PSTR framework and use the shareholding of QFII as the transition variable to investigate the threshold effects of the QFII shareholdings on stock returns and three risk premiums. In empirical, we select 27 semiconductor companies listed on the Taiwan Security Exchange Corporation as the sample objects. The sample period spans from 2008:Q1 to 2017:Q2. Thus, there are 1026 observations.
The empirical results can be summarized as follows: First, the traditional Fama-French three-factor model is estimated employing linear approaches. The empirical results show that the semiconductor stocks and semiconductor index are significant positive. Holding growth stocks can lead to higher returns than value stocks, and holding small size stocks can lead to excess returns than big size stocks. In addition, there is a significantly positive spatial dependence between stock returns. Second, the estimation results of the four-factor PSTR model reveal that three risk premiums vary with time and across stocks, depending on the shareholdings of QFII in different regimes, not constant obtained from the traditional three-factor model and four-factor model. Third, there has a positive spatial dependence as the QFII shareholding is lower than the threshold (21.301%). However, the dependence turns out to be negative as the QFII shareholding is larger than the threshold. Fourth, with the increase in the QFII shareholdings, the positive size effect weakens and even turns to a negative effect. That is, the function of external supervision begins fermenting, which is helpful for holding big size stocks (also the blue chip stocks) to earn excess return. Fifth, holding growth stocks has an excess return than value stocks; however, the effect is not significant and decreases with the increase in the QFII shareholdings. Sixth, the market premium of stock returns is significantly positive and increases with the increase in the QFII shareholdings. However, the traditional linear models underestimate the positive contribution of the market factor to stock returns.
The associated policy suggestions include: First, investors, semiconductor companies, and institutional investors must revise their investment decisions period by period according to the change in the QFII shareholdings, and adjust the most proper investment targets and mergers and acquisitions (M&A) objects. Second, in the four-factor PSTR model, there exists a positive spatial spillover effect on the stock returns, indicating that geographical distance actually affects stock returns. In addition, QFII shareholdings represent external monitoring and affect the stock returns. Investors, semiconductor companies, and financial regulators can regard QFII shareholdings as useful external monitoring indicators to assess the corporate governance of the semiconductor companies. Third, holding small growth stocks can earn excess returns as the shareholding of QFII is low. Fourth, in the viewpoint of total premiums, the market premium is the biggest one. Thus, at any level of QFII shareholdings, holding big size stocks is still the best investment strategy.
|
|---|
