Bayesian prediction of disability insurance frequencies using economic indicators

We use economic indicators to improve the prediction of the number
of incurred but not recorded disability insurance claims, assuming that
there is a link between the number of claims and the chosen economic
indicators. We propose a Bayesian model where we model the claims
development in three directions: along incurred periods, recording lag periods and calendar periods. A stochastic model of the economic indicators
is incorporated into the calendar period development direction. Thus we
allow for the impact of the economic environment on the number of claims.
Applying the proposed model to data, we illustrate how the inclusion of
economic indicators effects the prediction of the number of incurred but
not recorded disability claims.