混合頻率模型在農作物收入保險之應用 ─ 以臺南文旦柚為例

由於臺灣位於季風帶，天然災害對農作物損失影響嚴重，包含颱風、水災、乾旱、低溫等，對於臺灣農作物產量皆有重大之影響，導致農民時常面臨相當大的收入風險。收入保險被視為一種能降低農民面臨氣候造成收入損失風險的避險工具，而保費計算需先審慎評估氣候因素對作物產量之影響。本研究結合混合頻率迴歸（Mixed Data Sampling Regression，以下簡稱MIDAS）、合併平均群體估計法（pooled mean group estimation，以下簡稱PMG）以及移動區塊拔靴法（Moving Block Bootstrap，以下簡稱MBB）等計量方法，研擬一套以計量方法推估氣候變遷下，農作物收入保險之純保費以及農民最高願付保費之計算方式。首先以台南地區文旦柚為例，實證分析所使用的資料包括1981年至2016年臺南地區的累積積溫、平均雨量、平均風速及累積日照時數之月別資料，以及文旦柚每公頃單位面積產量之年別資料。透過MIDAS推估氣候因素對其產量之影響，把高頻率資料之變數轉換成可以對應低頻率資料的變數，以減少因時間頻率整合造成資料訊息流失之問題。以MIDAS之結果顯示，文旦柚於花芽分化期不宜抽梢，因氣候條件若較佳將使其抽冬梢，導致無法開花結果，故於花芽分化期間11月至翌年2月之累積積溫、雨量及累積日照時數對於產量為負面的影響，其中捕捉到雨量影響最大月份為12月；進入春天後開始抽春梢並開花結果，故於開花結果期間3月至7月之累積積溫對於產量為正面的影響；又因文旦柚果實肥大，故於6月至7月風速對於產量為負面的影響。接下來因文旦柚產量變化會造成市場價格變化，進而影響農民之收入，故本研究蒐集全台共計八個農產品批發市場之2006至2016年價量資料，使用PMG估計文旦柚之逆需求函數與價格彈性。估計結果發現文旦柚之需求存在長期穩定之負向價量關係，故可依此估計結果計算文旦柚產量變化造成之價格變化，進而模擬未來氣候變遷下之文旦柚收入變化。本研究應用MBB方法預測未來之可能風速及日照時數資料，並結合科技部「臺灣氣候變遷推估與資訊平台建置」產製之2020年至2100年氣溫及降雨月別推估資料，進一步推估未來價格之變化與農民所面對之收入風險，最後在效用極大化與風險趨避效用函數之假設下，計算文旦柚之純保費以及農民最高願付保費。結果顯示在面臨氣候變遷影響下，文旦柚於2021至2040年、2041至2060年、及 2061至2080年及2081至2100年四段期間之收入保險純保費分別為74,905、77,134、80,430及78,966元；農民之最高願付保費分別為53,530元、54,899元、57,096元及55,905元，兩者的差距可作為獎勵農民參加或提高農民投保率的補貼設計依據。

關鍵字

文旦柚；氣候風險；農作物保險；收入保險；混合頻率模型

並列摘要

Natural disasters have serious negative impacts on Taiwan’s agricultural production in recent decades. Farmers are very likely to face increasing number of disasters and suffer from more severe revenue losses due to climate change. Revenue insurance is regarded as one of the financial tools that reduces the risk from climate risk. Before calculating the insurance premium, the impact of climate risk on crop yields should be estimated. In this study, mixed data sampling model (MIDAS), pooled mean group (PMG), and moving block bootstrap (MBB) are employed to estimate the pure premiums and the maximum premium that risk averse farmers are willing to pay. A case study on Wendan pomelo in Tainan City is conducted using monthly climate data from 1981 to 2016, including growing degree days, average precipitation, average of wind speed, and accumulated sunshine hours, and yearly yield per unit. We use MIDAS to estimate the influence of climatic fact ors on the yield, converting high-frequency variables data into variables that can correspond to low-frequency data. The advantage of MIDAS model is that it can estimate the marginal effect of monthly climate factors, responding the growth status in each growth period. The empirical result of MIDAS shows that during the flower bud initiation and differentiation stage from November to next February, the growing degree days, rainfall and solar radiation accumulation have negative impacts on yields. During the bloom and fruit bearing stage, growing degree days will enhance the yields while the wind speed in June and July lower the yields. Next, considering the changes of market prices will also affect farmers' income, this study uses PMG estimation to estimate the inverse demand function and price elasticity of Wendan pomelo. We use the agricultural product wholesale market price data from 2006 to 2016. The empirical result shows that the demand for Wendan pomelo has a long-term stable negative relationship, so it can be used to calculate the price change caused by the change of production. In order to simulate the change of revenue under the climate change in the future, this study uses MBB method to predict future wind speed and sunshine hour data, and then combines the downscaled prediction data on future temperature and rainfall provided by Taiwan Climate Change Projection and Information Platform (TCCIP) from 2020 to 2100 to predict the revenue risks under climate change. Under the assumption of utility maximization and risk aversion farmers, the results show that for time period 2021-40, 2041-60, 2061-80 and 2081-100 respectively, the pure premium of Wendan pomelo revenue insurance are 74,905, 77,134, 80,430, and 78,966 TWD per hectare per year. The maximum premium farmers willing to pay are 53,530, 54,899, 57,096 and 55,905 TWD per hectare per year. The differences between them can be used to determine the optimal subsidies for the take-up of crop revenue insurance program in Taiwan.

並列關鍵字

Wendan Pemelo ； climate risk ； crop insurance ； revenue insurance ； mixed data sampling model

參考文獻

內政部統計查詢網，2006-2016，「台灣人口總量」。臺北：行政院主計總處。（http://statis.moi.gov.tw/micst/stmain.jsp?sys=100）（2019/2/20）

Google Scholar

王保玲、孫健、江崇光，2017。「我國引入農業收入保險的經濟效應研究」，『保險研究』3卷，5期，71-89。

Google Scholar

王惠正、林明哲、黃琮琪，2014。「先進國家實施農業保險之經驗」，行政院農業委員會，（http://www.rest.org.tw/upload/2014082110154356.pdf）。（2018/10/21）

Google Scholar

江春、張愛民，2010。「安徽省分縣小麥產量保險費率分析」，『安徽農業科學』，38卷，1期，374-377。

Google Scholar

行政院主計總處，2006-2016，「GDP、GNI及NI」。臺北：行政院主計總處。（https://www.dgbas.gov.tw/ct.asp?xItem=33338&ctNode=3099&mp=1）（2019/2/20）

Google Scholar

國際替代計量

混合頻率模型在農作物收入保險之應用 ─ 以臺南文旦柚為例

查找全文

主題瀏覽