台灣地區能源需求與CO2排放預測──狀態空間模型之應用

Energy Demand and CO2 Emission Forecasting Using State Space Models

能源為生產活動的重要投入，也是一般家計單位維持民生生計的必需品，而溫室氣體的排放量與能源需求的數量也息息相關。本文採用狀態空間模型（State Space Model）結合卡曼濾波（Kalman Filter）方法，針對煤炭、天然氣、液化天然氣、汽油、柴油、航空燃油、燃料油、焦炭、煤油、煉油氣、煤氣與液化石油氣等能源未來17年（1999年至2015年）的需求情況分別做預測，並同時將預測結果經過運算與轉換後，估算出CO2未來的排放量。另一方面，本文亦使用黃宗煌、李秉正、徐世勳、林師模與劉錦龍等人所開發出來的台灣可計算一般均衡動態模型（TAIGEM-D），來預測各年度能源的需求量，再用以推估二氧化碳的排放量，並進一步與狀態空間模型預測之結果做比較。本文研究結果發現，兩個模型估計的二氧化碳排放量結果相當接近，不過就各種能源排放二氧化碳的結構而言，兩個模型之間是有一些明顯的差異，而這些差異的來源主要是來自二者在模型結構以及資料應用上的差別。 The purpose of this paper is to forecast energy demand and CO2 emissions for Taiwan over the period 1999-2015. Two models are constructed for this purpose: one is a state space model and the other is a dynamic computable general equilibrium (CGE) model. The state space model is built based on detailed time series data on energy consumption spanning from 1961 to 1998, and the estimation is based on Kalman filter techniques. The dynamic CGE model used is the TAIGEM-D model, a multisectoral model of the Taiwan economy developed specifically to analyze climate change response issues. Total CO2 emission forecasts are computed using all the energy demand forecasts, with some adjustments and transformation. Results show that both models generate quite similar trends in total CO2 emissions. Discrepancies, however, are found to exist in the structure of energy demand between the two models.