There would appear to be a modelling error in UBS's historical...

There would appear to be a modelling error in UBS's historical figures. Does anyone know if those historical errors have been corrected within the forecasts UBS are now making of an extended period of over supply?

The flow sheet that exists for almost all the hard rock production is that it leaves the mine site in Australia, Canada, Brazil and Zimbabwe. From spodumene concentrate goes to China where it is refined into Lithium Hydroxide or Lithium Carbonate. Refining losses occur at this step. While I'm not sure exactly how large these refining losses are, recovery rates around 90% appear or can be calculated within western hydroxide/carbonate business cases. Assuming a 60/40 split of mine site supply to Hard rock/Brine (and no similar issues with brine), any world LCE supply estimate should be ~6% or below the aggregate totals of mine site production. This difference would be larger if recovery rates from spodumene were below 90%.

The USGS reporting of production volumes is on a mine site basis. The totals reported are the contained lithium equivalents, not what may ultimately be recoverable from that material. Allowing for Silver peak production that is not part of the USGS totals and as being 1kt of pure lithium equivalent and converting USGS pure lithium equivalents into LCE (5.323x), the last 3 years USGS prior year estimates (2021, 2022 and 2023) were 575kt, 782kt and 1,091kt. The equivalent UBS global supply estimates were 588kt, 788kt and 1,100kt. They are closely aligned (although UBS is fractionally higher).

Rather than estimating world LCE supply at perhaps 95% of mine site supply to reflect the recovery losses that will be occurring from further processing to get to the LC or Hydroxide end product, UBS has simply used the LCE calculations that come from multiplying mine site production volumes and reported grades. Recent UBS figures therefore remain aligned to USGS (well ~1% above) when they should actually be closer to 6% below USGS figures. The 1% could potentially be explained by differences like DSO that USGS didn't capture or timing differences around mine site vs port volumes.

Overall however, if you don't make the Spodumene to Hydroxide recovery loss adjustment you will have an overstated version of supply. At 1% over vs perhaps 6% under, this would create supply estimates with a ~7% (and perhaps more) overstatement bias. If UBS have done this in the historical actual results, its likely this same error has crept into their forecasts. While price dynamics indicate there is a supply surplus at the moment, it may not be anywhere near as bad as portrayed by UBS if they haven't fixed this historical error.

Haplo - With your skills are you able to double check this issue and confirm you agree?