巨災債券利差，氣象指標與巨災預測

Abstract

由於氣象學家預測巨災時更多地集中於氣象數據的分析，缺乏對人群活動等對氣候動態影響的考量，加上巨災相關的地質氣象指標通常每月或者每週才更新一次，因此傳統的巨災預測通常具有一定的滯後性。而在金融市場上，證券公司通常會聘用氣象學家等為其氣候衍生品提供諮詢服務，並結合各方面的資訊進行的發行或交易,進而決定衍生品的利差情況。基於此，本文主要對巨災債券的利差能否為巨災發生次數和巨災的預測提供額外的資訊進行了探討。我們使用負二項回歸或卜瓦松回歸去分析巨災債券的利差能否為巨災發生次數的預測提供額外資訊；而在巨災債券的利差能否為巨災損失的預測提供額外資訊的研究中，我們採用了指數回歸進行分析。在分析中，我們發現對於不同的情況而言，巨災債券的利差能否提供額外資訊的結果是不一樣的：巨災債券的利差能夠為預測地震的發生次數及相應的損失，風災發生所帶來的損失提供較為顯著和有效的資訊，對於風災和地震的總損失雖然能夠提供額外的資訊，但是卻並不是非常明顯。而對於風災的發生次數，風災和地震發生的總次數而言，巨災債券的利差則不能提供額外的資訊。

Due to the fact that meteorologists’ catastrophe forecast lacks the dynamic influence of human activities, plus the meteorological data updated monthly or weekly, traditional catastrophe forecast method may be inaccurate in forecasting catastrophe. Meanwhile, financial institutions usually hire meteorologists advise meteorological derivatives issuance and trading, and then make decisions according to various information. Therefore, we want to survey whether the spread of catastrophe bonds can provide extra useful information in catastrophe forecast. We use Poisson Regression or Negative Binomial Regression to analyze whether catastrophe bonds’ spread can provide extra information in forecasting catastrophe frequency. Likewise, we use Exponential Regression to explore the whether CAT bonds’ spread can provide extra information in forecasting catastrophe severity. According to our research, we find spread can remarkably provide extra information in forecasting earthquake frequency and financial losses of wind and earthquake. Besides, spread has trivial impact on forecasting total loss of wind an-d earthquake. However，spread fails to forecast the frequency of wind and the total frequency of wind and earthquake.