本文著重於溫度風險及探討何種分配最能捕捉臺灣溫度之動態行為。我們採用Campbell 與Diebold (2005) 模型捕捉臺灣溫度之特性及探討在不同機率分配之影響。我們發現標準 Gumbel 分配在樣本內外皆提供良好的配適與預測能力。此外，我們延伸Cao 與Wei (2004) 之 評價方法並求得HDD 與CDD 之價格。最後，我們發現在不同機率分配假設下其對溫度衍生 性商品影響十分顯著。

This research focuses on the temperature risk and attempts to investigate which distribution is most appropriate for capturing the Taiwan’s temperature dynamics. We adopt the Campbell and Diebold (2005) model to describe the temperature characteristics and examine a variety of distributions. We find that the standard Gumbel distribution provides the best fit for both in-sample and out-of-sample performance. Further, we extend Cao and Wei’s (2004) approach to obtain the valuation framework for HDD and CDD contracts. Finally, we observe that the effects of different distributions on the value of the temperature derivatives are very significant.