向分量迴歸模式之建構──二氧化碳排放與經濟發展關係再檢視

Construction of Double-Quantile Regression: Revisit the Relationship between Emission of CO2 and Economic Development

本文的主要目的首要針對全世界104個國家1990年至2005年，每人每年CO2的不同排放水準，進行一個面向的分量迴歸，同時也依據各國每人每年GDP劃分不同經濟發展程度，進行另一個面向的分量迴歸，並進而將二者結合建構出一個雙向分量迴歸模型，期望在綜合考量不同CO2排放量及不同經濟發展情況下，能更準確的連結CO2的排放量與經濟發展的關係。此外，在長期追蹤資料的使用下，進而以共同邊界技術比（meta-technicalratio，MTR）為代理變數以反應國家效果與時間效果的影響。最後則檢視在雙向分量與傳統單向分量迴歸模型，在不同經濟發展程度及不同CO2排放程度之國家，比較GDP每變動百分之一時，CO2排放量變動百分比之差異。實證結果顯示，不論那一類型經濟發展水準之國家，每人每年GDP與每人每年CO2排放量並不存在倒U字型的關係。此外，利用雙向分量迴歸得到的估計結果顯示，經濟發展水準較低的國家，每增加1％之每人每年GDP，必須承受較大的每人每年CO2排放量變動百分比之增加，然隨著經濟發展程度的提昇，每人每年GDP與每人每年CO2排放量關係的曲線則漸為平坦。具體而言，高所得高排放量的國家，每1％GDP變動帶來CO2排放量變動的所得CO2彈性值最小，其次則為高所得低排放量群組的國家，此表示對於這些最高所得國家，其所得已高至可以或是願意付出代價控制CO2的排放。相對的，低所得、任何排放水準的國家，在面對經濟發展的壓力下，每增加1％的每人每年GDP帶來的每人每年CO2之排放量，均比對應的相同排放水準而所得水準較高的群組國家為高。此一結果顯示經濟發展與對CO2排放量的控制，對低所得水準國家的衝突更加明顯，決策者在二者間的權衡拿捏將面對更大的壓力。 The purpose of this study is to construct a double-quantile regression model for 104 countries in the year of 1990-2005. This is a model by combining quantify one dimension for different levels of CO2 emissions and quantify the other dimension for different levels of GDP. It is expected to have better portrait between the emissions of CO2 and economic development while emissions of CO2 and developments of economy are accounted for simultaneously. Due to the characteristics of data, meta-technology ratio (MTR) is appropriate to catch country and time effect. The income elasticity of CO2 emission, i.e. every 1% change of GDP to the impact on the emission of CO2 will then be compared both for double-quantile regression model and traditional quantile regression model. The results indicate that a positive relationship exists between annual CO2 emission per capita and annual GDP per capita. That is, for either type of the country with any CO2 emission level, there is no inversed-U shaped between annual CO2 emission per capita and annual GDP per capita. The results from double-quantile regression model show that every 1% increase of annual GDP per capita will generate much higher annual CO2 emission per capita for low-income countries than that for high-income countries. Specifically, the income elasticities of CO2 emission for high-income and high-emission countries from the double-quantile regression model are the lowest among all countries at all income and emission levels. This implies that these countries are rich enough and willing to control the emission of CO2. On the contrary, since income elasticities of CO2 emission for low income countries are higher than those for high income countries this has led to a much higher conflict between the development of economy and control of CO2 emission for low development countries as these countries have significant pressure of economic development.