Tariffing a Nuclear Cover

In many actuarial calculations today, it is no longer sufficient to simply derive best estimates; instead, it is expected that the possible fluctuations in the outcome beyond these best estimates are also quantified and account for. This certainly applies when carrying out a solvency exercise, but also when deriving the risk margin of loss reserves or determining the cost of capital of a tariff.  The parameter risk is an often-neglected component of these deviations from the best estimate. Of course, this risk component can be small if the data sample is large and of good quality, but it may be large or indeed dominate when there is little data or data is inconsistent. Possibly a large parameter error may even be symptomatic of an inadequate model.  We present here a straightforward and general method to account for the uncertainties that arise when calibrating the parameters of any actuarial model, by sampling the probability distribution of the parameters, which arises implicitly from a chi-squared fit.