以因素分析法探討放空限制對股市之影響

Abstract

本論文致力探討放空限制對於台灣股市的影響；討論的層面包含自1998年至2008年共7次的政策轉折，使用的工具包括因素分析法以及聚類分析，目的是避免由表面的股票變量做相關性推論，而是從更本質的經濟潛在因子著手，分析它們的行為於放空限制前、後的變化，進而得出最一般化的結論。本論文首先使用因素分析法來確認四個主導股票市場行為的經濟因子：投資人意見分歧程度、市場交易活絡程度、散戶之知訊交易程度、價格訊息揭露程度，而各支股票於這四個因子的表現則使用加總分數來估計。在得到放空限制前、後的因子分數之後，本研究利用Hotelling's T2檢定以及MANOVA來推論經濟因子的行為是否被顯著地改變；由檢定結果可發現放空限制的效果並非獨立於時空背景的常數，它的影響會隨著不同時期的經濟背景而有所變化。當經濟局勢較為動盪時，實施放空限制能顯著地改變價格訊息揭露程度，促使市場空頭被抑制，進而達到遏止崩盤的效果。但是當金融環境相對穩定時，解除或實施放空限制都不會對股市有顯著的影響。因此，本論文不支持「放空限制最終會加速市場崩盤」的論點；本論文贊成放空限制能在危急時有效地挽救股市，付出的代價是價格訊息揭露程度下降，而承平時期的放空限制則不會有顯著的影響。

This thesis investigates the impacts of short-sales constraints by factor analysis models. Factor analysis models aim to find the latent factors behind the stock variables, and these latent factors provide more essential information since they are cardinal causes to the stock variables. Four latent factors are unearthed, including dispersion of investment opinion, thriving level on market trade, the ratio of informed trading, and the information content of market price. The deductions by factor analysis models are also verified by cluster analysis, and they agree very well. After that, the factor score of each latent factor is estimated, and the impact of short-sales constraints are statistically inferred by the changes of these factor scores. Both Hotelling's T2 test and multivariate analysis of variance (MANOVA) are applied to infer the impacts of short-sales constraints, and the results show that the impacts of short-sales constraints are dependent on the international economic environment. As financial crisis strikes economic environment, short-sales constraints significantly influence the information content of market price. Only optimistic opinions are revealed by the stock price, thereby preventing the escalation of stock market crash. On the other hand, when the economic environment is relatively stable, it appears that short-sales constraints have no significant influence on the stock market.