由於外人直接投資次數的資料具備間斷且非負整數之特性，若以連續型數量研究方法估計間斷型資料，模型規格設定的錯誤與資料處理上可能產生的重大遺漏，均可能導致最終之實證推論發生偏誤，因而使研究成果之可信度及參考之價值降低。本文應用可數追蹤資料模型針對台灣對外投資的總體影響因素進行實證分析，結果發現開放程度、失業率和經濟成長率與外人直接投資的次數呈正相關，然而通貨膨脹率以及貿易赤字與外人直接投資的次數呈負相關；匯率對外人直接投資的次數則不顯著。 This paper analyzes the determinants of Taiwan Direct Investment using count panel data models. Given the discrete nature of the data, the Poisson and negative binomial regression models are set up as vehicles for our empirical work. Empirical results suggest that openness, unemployment rate and economic growth rate are positively related to FDI occurrences; however, inflation rate and net export are negatively related to FDI occurrences while exchange rate weakly deter new FDI entry.
