Skip to content

Technical Pricing Framework (for Risk Carriers)

Data

10-30 years of historical weather data for the selected weather variable (e.g. precipitation) is obtained for the nearest grid point to the insured location and the following operations are performed.

  • Data cleaning and validation: Handling of missing values, reconciliation with external sources, correction of outliers.
  • Detrending: Removal of any long-term trends in the historical data using appropriate detrending methodology such as linear detrending, time series decomposition etc.
  • Index Calculation: Selected choice of the weather index is calculated for each historical year

Linear Detrending

Modelling

Experience Rating

  • Burn Cost methodology is used for calculation of risk premium where current level payouts are assigned to historical calculated indices and averaged to arrive at experience based risk premium
  • As an alternative to burn cost methodology, statistical distributions are fitted to the indices / data points and Monte Carlo simulation is performed to calculate various risk measures such as x% VaR

LEC

Exposure Rating

  • In order to reflect the future trends appropriate, forecast data for an appropriate time horizon is used to calculate expected payouts
  • Forecast Ensembles data is used to calculate future index values and assign payout to each ensemble (scenario), these are then probability weighted to arrive at expected payout

Ensemble Forecasting

Risk Premium

Risk Premium is then a credibility weighted average of the experience and exposure based risk premium