1.4 Conclusions

LS is the best linear estimate of \(Q(x)\), the theoretical best estimate of the ultimate loss

LS produce more reasonable results when year to year fluctuations are severe

LS does not do well when the variations are due to underlying changes in the payment pattern (due to internal or external changes)

• e.g. When there is a systematic shift in the business

Subject to sampling error when estimating parameters similar to other methods