When each model was added, and when one was revised and why. A saved business is measured against this register, so this is the page that says whether the answer you were given last month is still the answer.
Split the estimate into three lines rather than one. The 12.5% on casual hours is certain; the single-rate and hours-based sick-leave credits that run the other way are estimated, so the range is now asymmetric instead of pretending all three are equally solid.
Band changed from a chosen 1.9–6.5% to the actual spread of the last six increases, 1.5% to 7.1%. The base now takes the lower of a full-time year at the minimum and a pro-rata share of the real wage bill, because the full-time assumption overstates it for part-timers and in the industries where this rule bites that is most of them.
Model written from the published rates.
Model written from the published rates.
Model written from the published rates.
Model written from the published rates.
Model written from the published rates.
Because the answer is only as current as the register, and a tool that quietly changes its own arithmetic is worse than one that does not exist. Every revision here says what moved and why, so a number you wrote down last quarter can be checked against the model that produced it rather than silently replaced.