COLUMN-What will stop the infernal aluminium premium machine?: Andy Home

(The opinions expressed here are those of the author, a columnist for Reuters.)

By Andy Home

LONDON, Jan 28 (Reuters) - Another quarter and another rise in the physical aluminium premium for frustrated Japanese buyers.

What will serve as the benchmark for physical deals across Asia in the first three months of 2015 (PREM-ALUM-JP) has just been set at $425 per tonne over the price of cash aluminium on the London Metal Exchange (LME).

That's an increase of $5 per tonne from the last quarter of 2014. Which doesn't sound so bad until you consider that the premium was just $255 this time last year and $114 a couple of years ago.

This year's settlement came later than usual as buyers fought a dogged rearguard action. The Japanese argued with some justification that high domestic aluminium stocks and oversupply in other Asian countries were reasons for premiums to fall.

To no avail. Producers could simply counter that premiums elsewhere, particularly in the North American market, were still rising and, in an interconnected global marketplace, that meant that Japanese buyers would have to pay up if they wanted their metal.

Such is the relentless rise of the global aluminium premium machine, which has rent asunder the historical relationship between LME basis price and physical premium.

When will the infernal machine stop?

There is a broad analyst consensus that premiums will peak some time this year, although everyone thought that last year and look what happened.

What should make this year different, many argue, is the implementation next month of the long-delayed new LME warehouse load-out rules.

The stated aim is to reduce the long queues to get aluminium out of the LME sheds at Detroit and Vlissingen. And as more metal flows out of the LME system, the thinking goes, more becomes available to the physical market.

It logically follows that premiums must start falling, right?

QUEUES DOWN...?

Well, maybe.

Firstly, there are a couple of important details when it comes to the nitty-gritty of how the LME's new rules will actually work.

It's quite possible that the queues will lengthen before they start contracting.

Consider, for example, the situation at Vlissingen, where LME warehousing is dominated by Pacorini, the warehousing arm of Glencore (Xetra: A1JAGV - news) . As of the end of December there were 1.215 million tonnes of metal awaiting load-out. The resulting queue to get metal out of the Dutch port stood at 586 days.

Pacorini has been loading out aluminium like clockwork at a rate of 3,000 tonnes per day, in line with the LME's stipulated minimum (or maximum, if you're an LME warehouse operator) rate.

Since July 2013, when the LME first announced its new rules, which link load-in to load-out (LILO) rates, Pacorini has loaded out a total 1.350 million tonnes of metal, mostly aluminium.

Over the same period it has taken in 1.051 million tonnes, again mostly of aluminium.

What the LME calls its "Preliminary Calculation Period" expires this Friday. Load-in and load-out metrics over the interim 19 months will determine whether Pacorini has to load out at a faster rate in the March-May 2015 period.

Quite evidently, Pacorini could warrant more metal before the end of this month without crossing the threshold for faster load-out. Whether it chooses to flex the LILO model in this way we'll get to find out in Monday's LME stocks report.

If it does, there will be more aluminium that could join the load-out queue, if the owners want to move it out of Pacorini sheds.

And even if it doesn't, there are still 536,225 tonnes of non-cancelled aluminium in Vlissingen, which could also still join the load-out queue.

The Vlissingen queue, in other words, will remain dynamic for an as yet uncertain period before it starts definitively decaying.

...DELIVERIES UP?

The aluminium queue at Detroit, by contrast, is now in definitive decay mode. It steadily declined from 702 days at the end of September to 610 days at the end of December.

This is because Metro (Other OTC: MTRAF - news) , the dominant warehouse operator in Motown, stopped taking in fresh deliveries of aluminium many months ago.

Not only has it been starved of fresh inflows since July, most of what was stored there has since been cancelled. There are only 16,075 tonnes of non-cancelled aluminium in Detroit. And even if all of that were in Metro sheds and all of it were cancelled tomorrow, the impact on the Metro load-out queue would amount to no more than six days.

Ironically, this means that the load-out rate from Metro (SES: E1:M01.SI - news) sheds is about to slow, since the total amount of tonnage it holds is fast approaching the 900,000-tonne mark.

Below that level the minimum/maximum (delete according to preference) load-out rate drops from 3,000 tonnes to 2,500 tonnes per day.

Nor, by the way, is there any chance of Metro rebuilding its queue, even if new owners the Rueben Brothers, which bought the company from Goldman Sachs (NYSE: GS-PB - news) in December, were minded to do so.

It just wouldn't make economic sense.

QUEUES EQUAL PREMIUMS?

That's because even while the Detroit load-out queue has been falling, the Midwest U.S. physical premium has been rising.

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Graphic on U.S. physical premiums vs Detroit queue: http://link.reuters.com/fuv83w ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

It's an uncomfortable truth for the many critics of the LME who argued that lengthening queues equalled rising premiums.

However, any correlation between the cost of getting aluminium out of Detroit and the cost of buying freely available metal in the physical market has broken down.

Right now, for example, the "value" of the Detroit queue, meaning the cost of warehousing rent while waiting for a load-out slot plus the actual cost of loading out, stands at $351 per tonne, basis the LME's end-December queue data.

The physical premium for Midwest delivery, as assessed by Platts, leading global energy, metals and petrochemicals information provider, has just risen to a fresh all-time high of 24.25 cents per lb, or $535 per tonne.

For Metro and its new owners, that means there is no way of incentivising fresh deliveries onto warrant without booking an immediate near $200 per tonne loss.

For everyone else, it means...well, quite a lot of collective head-scratching.

The value of the Detroit queue must logically be in the mix, but if it is, it is lying embedded beneath other drivers, most particularly a widening production-demand deficit in the regional North American market.

A QUESTION OF WHEN OR WHAT?

There is another uncomfortable truth for those expecting falling LME queues to translate into falling premiums.

Which is that metal leaving the LME system may not be alleviating physical availability at all.

A total 670,000 tonnes of aluminium left LME warehouses in Detroit last year. Across the system as a whole, the cumulative outflow was 2.13 million tonnes, exceeding even the most optimistic estimates of supply-demand deficit.

It is clear that part, and quite possibly the largest part, of that outflow has simply moved from on-market to off-market storage.

After all, the disparity between the cost of the two is why so much aluminium was queuing to leave the LME system in the first place.

The continued attraction of the cash-and-carry trade, whereby financiers earn the difference between spot and forward prices on the LME curve, is as important a driver as fundamentals of the aluminium premium machine.

Every tonne locked up in financial cold storage is a tonne not available to the physical market place.

And it's here rather than in queue metrics that the infernal machine may yet come unstuck.

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Graphic on LME aluminium stocks and spreads: http://link.reuters.com/guv83w ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Because another by-product of all that metal leaving LME warehouses is the dwindling of non-cancelled stocks in the system. At 1.692 million tonnes they are at their lowest since late 2008.

And LME forward spreads are becoming ever more volatile as a result.

After spending 2013 and the first half of 2014 in becalmed contango, the benchmark cash-to-three-months spread has been experiencing ever more frequent paroxysms of tightness. It's happening again right now.

Making a turn out of financing aluminium isn't yet impossible. But it's getting harder and the margins are getting thinner.

There are currently around 2.4 million tonnes of aluminium waiting to leave the LME system. Whether the cumulative outflow rate from Detroit and Vlissingen will change much after the new LME rules is a highly moot point.

There is probably almost twice that much metal sitting in off-market finance deals. Whether it becomes available to the physical market will depend on LME spreads.

In terms of when the aluminium premium machine will finally stop rising, the key question is what will stop it rising? LME rule change or LME spreads change?

(Editing by Susan Thomas)