自從1980 年代起，伴隨著貿易與資本帳之開放，各國(包括大部份先 進與少數發展中國家)所得分配不均度(income inequality)呈現逐漸擴大之趨 勢。因此，本計畫旨在探討貿易開放對所得分配不均度的相關議題。既存文獻強 調，貿易引進新技術與知識以及促使政府採取相關制度改革，進而影響所得分 配。相關文獻亦指出新技術與知識的移轉並非自動的(automatic)，一國之經濟 發展程度，例如教育(educational)、技術(technological)與/或基礎建設 (infrastructure)發展，以及金融發展之程度具有顯著左右廠商採行國外技術與 知識的能力。有鑑於此，本計畫分析一國之人力資本，包括質(quality)與量 (quantity)，與金融發展，包括銀行與股市，是否侷限或助長廠商採行國外技術 與知識的能力來檢定貿易開放對所得分配不均度的影響是否呈現非線性效果。利 用Dynamic Panel GMM 與 bias-corrected Least Squares Dummy Variable 估 計方法進行探討貿易開放政策對所得分配之影響是否因人力資本的累積與金融 發展程度不同而不同。藉以評估國家差異性(country heterogeneity)的重要性。 Income inequality is considered as an important determinant of a wide variety of economic outcomes. Numerous studies identify its effects not only on poverty but also on socio-political (in)stability and redistribution, human or physical capital accumulation, the incentive to innovation, fertility, and public policies that determine the long-run economic performance. The question of what are the main determinants of income inequality has thus attracted a great deal of attention and debate in the literature. This project re-evaluates the impact of trade openness on income inequality and intends to contribute to the empirical literature in several dimensions. First, departing from existing empirical literature that focuses on aggregate trade flows, we not only explore the effects of total trade flows, but also disentangle trade flows according to their origin and destination areas to gain a clearer picture of the trade-inequality link. Second, we explore whether the relationship between trade and income inequality varies with the extent of human capital and financial development. Human capital matters because it determines a country’s comparative advantage and affects the adoption of best-practice technologies and learning by doing. Similarly, financial development is critical as it determines a country’s comparative advantage and the ease of accessing capital to finance new global opportunity. Third, the analysis is implemented using a LSDVC estimator proposed by Kiviet (1995) and Bruno (2005) which is particularly suitable for small samples, and is compared with the system GMM estimator of Arellano and Bover (1995) and Blundell and Bond (1998) that is designed for large cross-section units and small time-series observations.