Doing some further digging I think I have found a partial...

Doing some further digging I think I have found a partial explanation for the total COGS/t being less than the production cost/t.

In the 2020 HY report MRC states that the material improvement in COGS was "due to a material positive garnet inventory adjustment during the half-year due to settlement of the stockpile areas", and that without this adjustment the Revenue-COGS ratio would have been 1.6x or approx $69.88 per ton. While this makes some sense it still leaves the COGS less than the production costs for the 1H/2020.

If this is correct it raises another question as to why the garnet inventory was so grossly underestimated previously, by a factor of 2.4x. It's hard to imagine how as an experienced sandminer you could look at a stockpile of minerals and make such a mistake...