在過去實證研究中發現，GARCH模型殘差項服從常態分配之假設常常遭到拒絕，因此本篇論文利用殘差項服從調和穩態過程中的基本調和穩態過程以及瞬間調和穩態過程之GARCH模型來做為碳權現貨價格之動態過程。運用Bluenext交易所提供之EUAs碳權現貨價格資料，本研究發現殘差項服從瞬間調和穩態過程之GARCH模型具有極佳的配適能力。 此外，透過數值分析可知，假設EUAs碳權現貨價格動態過程為傳統的GARCH模型將高估EUAs碳權期貨選擇權之價格，且不論運用基本調和穩態過程或是瞬間調和穩態過程，EUAs碳權期貨選擇權之價格差異不大，故當殘差項服從調和穩態過程之GARCH模型，模型風險較低。 Empirical studies show that the hypothesis of normal distribution of residuals was often rejected. Therefore, this paper presents GARCH models with an infinitely divisible distributed innovation, referred to as the classical tempered stable (CTS) GARCH model and the rapidly decreasing tempered stable (RDTS) GARCH model to catch the dynamic process of CO2 emission spot price. This paper compares the performance of normal-GARCH, stable-GARCH, CTS-GARCH, and RDTS-GARCH models using EUAs data obtained from Bluenext environmental exchange and finds that RDTS-GARCH model has a better fitness than others. Our empirical results show the NORMAL-GARCH model tends to overestimate the price of EUAs future options. But the results are virtually similar by using either CTS-GARCH model or RDTS-GARCH model, which means that the model risk of tempered stable-GARCH model is lower.
