本論文克服之前文獻之研究設計及方法使用之不足，強化在研究架構及設計上的周延考量，自非線性角度出發進行研究相關議題之實證探究。討論之主題有三大部分，首先將分別以股市及匯市之國際連動以兩個分開之議題分別進行討論，第三部分則將著重於企業現金持有對企業未來獲利之影響探討。 第一部分專注於二十世紀末國際匯市之連動議題探討，以新台幣為主要匯率變數，分三個群組，依組別分別進行研究分析。該部分採Ender & Hurn (1994) 的 「廣義購買力平價」(G-PPP)理論，作多國匯率間互動關係之測試，借以分析新台幣在國際舞台中所扮演的角色，探討新台幣匯率與預設「最佳貨幣區」中各國匯率間的長期均衡互動關係。研究方法為Johansen (1988, 1990, & 1994)多變數誤差修正最大概似共整法。實證發現，所預設的三個多國匯率組合，都成功的建立起「最佳貨幣區」，表示新台幣與預設貨幣區中各國貨幣都存在著高度的互動影響。這其中，尤以「亞洲四小龍」四國間的互動關係最為顯著。文中亦發現，包括台灣在內的泛亞太平洋地區各小國間，也存在著相互依存關係，支持了Ender & Hurn (1994) 的發現。而新台幣與七大工業國強勢貨幣間的匯率共整測試，結果也顯現出新台幣受著此間強勢貨幣的深度影響。 論文第二部分為二十一世紀初受「金融海嘯」影響之國際股市連動之議題探討。使用之研究方法為Enders and Granger (1988)及Enders and Siklos (2001)的不對稱門檻共整合模型，包括TAR與MTAR及進一步的TECM與MTECM，探討美國股市三大指數與亞洲六個主要國家 (地區) 股市間不對稱的長短期均衡與動態關係。實證結果，先於相關性檢定中，發現次級房貸危機確實顯著地提高了美股與亞股間的相關連動，符合世界銀行對「傳染」(contagion)的最嚴格定義。 論文第三部分針對美國跨越世紀間上市且仍活耀交易的668家公司為樣本，探討企業現金持有之數額對企業未來獲利之可能影響。研究方法採以Hansen (1999)的非線性縱橫門檻迴歸(PTR)，探究在跨越世紀間國際金融資產波動加劇及國際資本市場融資難度增加之下，企業現金持有之數額對企業未來獲利之可能影響。先後以包括固定效果及隨機效果的線性迴歸進行分析，發現現金持有與公司價值(企業緯來獲利)間，確有非線性關係之存在。進一步進行的PTR分析，實證所得發現，公司持有現金數越多，對公司價值確有正向的影響；然而，現金持有數會有一定的最適水位，超過門檻後，持有現金水位過高反倒會降低公司價值。研究之發現，建議企業經理人在公司營運上，宜建立適當的最適現金持有之流通性政策，以提升公司價值，創造公司價值最大化。 Three subjects are empirically studied in this research. We first use two parts to investigate the international contagion effects among international stock markets and international exchange rate markets, respectively. The third subject concern the nonlinear testing for the optimal cash holding for a firm to maximize the shareholders’ wealth. In order to conquer the insufficient application of the methodologies from previous literature, the empirical study of our research undertakes both linear and non-linear examination to investigate the ‘style-fact’ of the real life phenomenon. We also reinforce the research design by setting up appropriate grouping in each of our research subjects. In the first subject, we study on the long term equilibrium relationships among international exchange rate markets in each of three group settings (Asian four little dragons, Pacific Rim nations as a whole, and Taiwan together with seven large industrialized countries) basing NTD as the core currency during the period of post-Bretten Woods. The methodologies applied in this part are traditional Johansen (1988, 1990, 1994) multivariate maximum likelihood cointegration test. The theoretic model behind this part is the so-called ‘G-PPP’ elaborated by Enders and Hurn (1994) which related to the concept of OCA by Mundell (1961). The evidence shows that G-PPP holds for all three group settings, while the co-moving style explained by the number of cointegration ranks and the presence of linear trend and quadratic trend is found different for these three group settings. This implies that all three group settings had been successfully established OCA, but they had experienced different time paths of co-movement in the post-Bretten Woods era. The second subject of our study is to examine the long-term equilibrium relationship between the U.S. stock market and Asian stock markets using asymmetric threshold co-integration model of Enders and Granger (1998) and Enders and Siklos (2001). The difference between the long-term equilibrium relationships during the pre- and post- subprime mortgage crisis is further examined. Moreover, this study analyzes the contagion effect of the U.S. stock market on Asian stock markets and asymmetric adjustment of the long-run equilibrium relationship before and after the subprime mortgage crisis. The findings show that the asymmetric long term equilibrium relationships between the US stock markets and Asian stock markets had been largely increased after experiencing the subprime crisis and financial tsunami. Our final subject is trying to find the optimal level of cash holding for maximizing the firm value during the highly volatile period of international financial assets across centuries from 20th Century to 21st Century. All the 668 public listed and still actively trading US corporations from the period 1988 through 2008 are investigated and the methodologies employed is nonlinear non-dynamic panel threshold regression (PTR) elaborated by Hansen (1999). We first find that there is a non-linear relation between cash holding and firm value, and further adopt PTR model to examine the threshold effect. A two-threshold effect of cash holding on firm value is found in our empirical study. The result infers that more cash holding is good for firms but there exists an optimal level. These empirical results show that firm managers should maintain applicable liquidity policy to maximize firm value.