So here is the problem: research quality is a nebulous concept and it takes many years to work out whether someone's output has actually high quality or not.
This is especially hard for non-insiders. The Government is a non-insider that funds academic research in Australia and so would like to know what it is getting for that money.
So here is a typical solution: use the quality of research publications as a proxy for likely high quality in the future and use that to measure performance.
Academics will recognise that solution as it is precisely what they try to employ internally.