
202 Pricing in General Insurance
The considerations above lead us to revise Version 0 of the decision process like this:
Version 1 of the decision method still has some problems. First of all, if the exposure is
not constant, the variance/mean distribution may be different, even after adjusting the
individual years bringing them to current term’s exposure. This is not a major difculty
because one can simply produce simulations for this case as well, sampling for each year
from a different Poisson distribution whose rate is proportional to the exposure.
Second, this method implicitly assumes that the default distribution is the Poisson dis-
tr