香蕉收入保險訂價之研究

The Pricing of Banana Revenue Insurance in Taiwan

因應氣候變遷與價格波動，作物收入保險已是近年來各國實施農業保險的新趨勢，我國亦開始規劃香蕉收入保險。然因產量與價格資料有限，恐將影響保險之客觀訂價，故應用機率統計方式，模擬量價資料特性及機率分配，以協助保費精算。本文基於歷史模擬收入損失，反映非齊性Poisson過程（non-homogeneous Poisson process，簡稱NHPP）資料特性，並結合布朗運動產生大量價格及產量資料，分別推估損失頻率與損失幅度，以釐訂純保費。實證模擬結果顯示：各地因生產條件及產量差異變化，高屏與非高屏地區之各地保費有別，且非高屏地區各地保費的差異性或發散程度較高屏地區為大，反映非高屏地區的產量變異與價格變異程度均較高的情形。此外，不同保障程度之保費差距，並不具比例關係，以95%與90%保障程度的保費差距，多較其他級距為大，表示大多數的損失均在10%以上；而損失在20%以上雖發生次數較少，但災損幅度更大，表示80%保障程度需要納入承保範圍，以補天災救助之不足。

"In response to climate change and price fiuctuations, crop revenue insurance has been a new trend in the implementation of agricultural insurance in various countries in recent years. Taiwan has also begun to plan banana revenue insurance. However, due to limited production and price data, it may affect the objective pricing of insurance. Therefore, probability statistics methods are used to simulate the characteristics of yield and price data and probability distribution to assist in the actuarial calculation of insurance premiums. Based on historical simulation of revenue loss, this paper refiects the characteristics of non-homogeneous Poisson process data, and combines Brownian motion to generate a large amount of price and yield data to estimate the frequency of loss and the extent of loss respectively to determine the pure premium. The empirical simulation results show that due to changes in production conditions and yield, the premiums of Kaohsiung-Pingtung and non- Kaohsiung-Pingtung regions are different, and the difference or divergence of premiums between regions in non-Kaohsiung-Pingtung regions is greater in Kaohsiung-Pingtung regions, refiecting in non-Kaohsiung-Pingtung regions both the yield variability and the price fiuctuation are relatively high. In addition, the difference in premiums between different protection levels is not proportional. The difference between 95% and 90% protection levels is usually larger than other levels, which means that most of the losses are more than 10%. Although the occurrences in losses more than 20% are less frequent, the disaster losses are greater, which means that 80% of the coverage needs to be included to make up for the less compensation of natural disaster relief."