Chapter 16 page 9 states that a "range of best estimates" reflects parameter uncertainty and model error alone. However on page 21 it states that a "range of reasonable best estimates" illustrates the parameter uncertainty (alone). ??
In page 9, we are given an example of 100 actuaries using 100 'different' models. Some of the models may be more similar. Therefore the differences between their best estimates can be either due to choosing in correct parameters or model error. On the other hand on page 21, we are referring to a range of reasonable best estimates of one model. In this case, the range shows how far the median/mode/mean can lie from the actual (unknown) measure of central tendency; resulting in an illustration of parameter uncertainty. The uncertainty in model selection is not examined here since only one model is being examined and it is not compared to others. P.S. I found this article to be very helpful.
Is the range of best-estimates for one model simply a confidence interval? By "an estimate with a particular likelihood of exceeding the outcome" are we referring to a percentile? Furthermore does chapter 16 indicate that Best-estimate = mean ⊆ {Measures of Central Tendency: mean, mode, median} ⊆ Point estimates (this includes percentiles)
