Forecasting techniques for insurance company production planning and recommendations

Forecasting

The type of forecasting that should be in place at an insurance company is time series analysis, as it is through this approach to forecasting that prior demands are used to predict future demands (Chase et al. 2005). At the particular insurance corporation in question, this is precisely the type of forecasting that is in place; the number of claims expected in a given period of time is based on the number of claims (in relation to the number of clients covered by insurance policies) in previous comparable time periods, and perhaps more importantly the specific time demands of any given claim are predicted based on the average times of claims handled by the organization. For firms like insurance companies where per-unit "production" times and resource demands vary, time series analysis is an essential means of forecasting to ensure proper human and other resources are available (Armstrong 2001).

Time series analysis forecasting methods are not always incredibly successful, as they depend on consistency in the environmental factors influencing demand and affecting supply capabilities, or at least the ability to accurately predict and account for them (Makridakis et al. 2008). For insurance companies, however -- for the particular insurance company in question, at least, and presumably for all insurance companies that manage to meet their growth and profitability targets -- this method of forecasting is quite effective. The averages used to make forecasts are derived from large and ever-growing data populations, increasing their accuracy, and the accuracy of the data used is one of the key factors in determining the accuracy of forecasts and predictions (Chase et al. 2005; Armstrong 2001).

The adoption of a weighted moving average practice for determining claim numbers might be advisable to account for changing overall trends in the number of claims filed, enhancing still further the time series analysis techniques already in place at the company. This should not take place in a direct weighting of more recent data as heavier than data in the more distant past, as is a standard method of incorporating weighted moving averages, but rather specific time periods or months should be compared to the same period or month in previous years, with heavier weight given to more recent years, but without the standard continuous linking of discrete time periods as advocated in traditional weighted moving average calculations and forecasts (Chase et al. 2005). More straightforward and un-weighted methods of time series analysis forecasting should also be maintained, providing the company with ore comprehensive and detailed forecasts (Armstrong 2001).