Bridging meteorology and finance: Italian temperatures for weather derivatives

This thesis aims to present a statistical modeling of Italian daily mean temperatures with the aim of applying it to temperature-based climate derivatives. To this end, nine cities were selected in order to best represent the main climate zones of Italy. The first phase consists of an in-depth analysis of the mean, maximum and minimum temperatures observed over the last twenty years. Two different modeling approaches were then developed and compared: a continuous-time model based on an Ornstein–Uhlenbeck process modified to obtain a time-dependent mean and a discrete model, based on an ARMA–GARCH combination applied to the seasonally adjusted residuals. Both models were calibrated on daily historical data and used to simulate, through the Monte Carlo method, the future evolution of the most used temperature indices in the financial field: the Heating Degree Days (HDD) and the Cooling Degree Days (CDD). The simulations obtained were compared with the observed data, evaluating their predictive capacity. Finally, the results were compared with those deriving from the classical burn analysis approach, in order to highlight its advantages and limitations from an application perspective.