Weather Derivative Pricing and the Year-Ahead Forecasting of Surface Air Temperature: A Comparison of Predictions Based on Local and Global Trend Estimates

Abstract

The actuarial pricing of weather swaps often depends on predicting the temperature several months in advance. Such predictions can be made by modelling the trends in historical temperature indices and extrapolating them forward in time. Linear trends are one of the simplest types of trend model one can use. They have two parameters: the mean level, and the trend rate. The mean level is probably best estimated from local data, but it is not obvious how best to estimate the trend rate: in this study we compare the performance of predictions based on local and global trend rate estimates.

Keywords: weather derivatives, trends, detrending, year-ahead forecasting

JEL Classification: G12, G13