COLUMN-The Goldman Sachs guide to LME warehousing: Andy Home

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

By Andy Home

LONDON, Nov 25 (Reuters) - September 17, 2010 would turn out to be a fateful day for the global aluminium market and for the London Metal Exchange (LME).

That was the day 100,000 tonnes of metal being stored in LME-registered sheds in Detroit were cancelled prior to physical load-out.

It was the largest single-day cancellation the LME system had experienced.

And quite evidently this wasn't a manufacturer looking for emergency top-up metal. It was, in fact, Deutsche Bank (Xetra: 514000 - news) engaged in a "cash-and-carry" trade, using the LME forward curve to generate a level of return that had become increasingly elusive in a time of quantitative easing.

The cancellations created an instantaneous bottleneck at Metro International, the dominant LME warehouse operator in Detroit.

The LME at that stage stipulated that warehousers had to load out a minimum 1,500 tonnes per day per location, even that requirement being diminished by the common practice of treating the minimum as a maximum requirement.

A relatively benign load-out queue of just over 20 days at Detroit mushroomed to a monster queue of 120 days.

That was the "big bang" moment for the aluminium market, the start of an evolution that would see load-out queues lengthen to years and the industry's pricing model splinter.

There was another unique aspect to those cancellations. The metal never really left Metro (Other OTC: MTRAF - news) 's warehouses. Most of it was trucked to another Metro shed in Detroit, eventually to be put back on to LME warrant.

It marked the start of a revolution in LME warehousing. It also marked the arrival in the metals storage business of Goldman Sachs (NYSE: GS-PB - news) , which had bought Metro earlier that year.

THE WAREHOUSER'S DILEMMA

Metro and Goldman executives were last week grilled by a U.S. senate investigative subcommittee. They stand accused of manipulating the aluminium price to the detriment of manufacturers by aggressively building up load-out queues at Detroit.

Neither Metro nor Goldman probably had any thoughts about the aluminium price when they offered Deutsche Bank their unusual deal back in 2010.

Rather, in all likelihood, it must have seem an innovative way of dealing with every warehouse operator's dilemma, which is how to stop metal, and the associated rental revenue stream, constantly flowing out of its sheds.

Historically, LME warehouse operators have offered rent discounts to keep metal locked in storage.

Metro was already playing with a different template by the time Goldman Sachs swooped on the company in February 2010.

At the end of January 2010 Metro Detroit was storing 893,000 tonnes of aluminium, largely because it was in the right place, the nearest LME good-delivery point to the Canadian smelter network, at the right time, one of massive surplus.

The only problem was that people were increasingly looking to move their metal out of LME warehouses to cheaper non-exchange storage, maximising the profitability of the cash-and-carry trade.

Metro itself profited from the resulting exit queue because it could charge full rent on cancelled metal and use that revenue stream to bid for fresh deliveries.

But it was still "losing" metal every day.

MERRY-GO-ROUND

So when Deutsche Bank asked for a rent discount for its 100,000 tonnes, Metro, and the idea does seem to have originated with the warehousing operator although it was approved by Goldman executives, came up with its smart idea.

Deutsche would get its discount while the metal was still in the LME system and it would get an even better discount deal for off-market storage as well.

All it had to do was cancel the metal, apply for the maximum number of load-out slots, transport the metal to another Metro (SES: E1:M01.SI - news) warehouse in Detroit and, once its financing deal had expired, re-warrant the metal back into the LME system.

If it did, its transport costs of $42.95 per tonne would be reimbursed when the metal was re-warranted. If it didn't, it would have to pay Metro a $65 break fee.

It must have looked like a win-win deal. Deutsche got its discount. Metro kept the metal in its system and pumped up the load-out queue, which it was already using to finance incentives. And Detroit-area truckers would earn richly from transporting thousands of tonnes of metal from one warehouse to another.

It was, according to Senator Carl Levin, chairman of the Senate subcommittee, a "merry-go-round" deal. It was, countered Metro chief executive Chris Wibbelman and chairman Jacques Gabillon, head of the bank's Global Commodities Principal Investment Group, an "off-warrant transaction".

And there would be plenty more such "off-warrant transactions" over the next few years.

The subcommittee's report itemises five other such deals, four of them conducted with UK metals hedge fund Red Kite and one with Glencore (Xetra: A1JAGV - news) , although the latter was structured differently because Glencore was already sitting in the Detroit queue.

Between February 2010 and January 2014 more than 625,000 tonnes of aluminium went on the Detroit merry-go-round.

Each one of those tonnes acted to maintain or increase the load-out queue.

AMBIGUITY

Extraordinary though it may seem, such a circular movement of metal doesn't appear to have breached any LME rules.

In the wake of the media blitz that followed aluminium consumers' complaints at an earlier session of the subcommittee, the exchange notified Metro on Dec 4, 2013 it was opening a formal investigation into its "off-warrant transactions".

But in communications with the committee, the LME conceded that there was ambiguity in its own rule-book concerning what constituted a "delivery out". Up until then metal simply had to physically leave a warehouse to count as a delivery. Moreover, once it did, it moved beyond the LME's regulatory purview.

"Without addressing any specific actual conduct" and "on a hypothetical basis," the LME told the subcommittee such a merry-go-round deal would be "inconsistent with the 'spirit'" of the rules but might not "violate the 'letter'" of them.

Even Metro personnel were uncomfortable with the idea of offering incentives to those already holding metal in the company's warehouses rather than as a way of attracting fresh inflows.

"I remain concerned, as I have expressed from start, regarding 'Q management'," wrote Mark Askew, VP Marketing, in a Dec. 4, 2010 email to Metro CEO Chris Wibbelman.

He was reassured, then, and again in February 2012, when Red Kite was taking 250,000 tonnes on a circular tour of Detroit, that Metro was not in the business of managing queues, but were merely offering an "off warrant storage option (...) that doesn't in any way violate the rules of the LME."

In truth, no-one at the LME ever seems to have thought about such an eventuality.

But then no-one was expecting five million tonnes of aluminium to flood the LME system either.

BLURRING THE LINES

From that ambiguous start, the Goldman-Metro warehousing model became ever more ambivalent in terms of the "spirit" of the rules.

By 2012 when "off-market transactions" were being offered to Red Kite and Glencore, the linkage between Detroit load-out queue and physical Midwest aluminium premium was becoming increasingly evident.

Indeed, according to the subcommittee's report, "Goldman disclosed that, from 2010 through 2014, in at least 13 arrangements, Metro received payments from some warehouse clients of amounts there were directly or indirectly tied to the Midwest premium."

In other words, Metro had moved from being pure logistics supplier to the aluminium market to becoming a protagonist and participant in that market.

Goldman, meanwhile, was itself building up an ever bigger physical aluminium book. The bank's physical holdings grew from zero at the time of the Metro acquisition to 95,000 tonnes at the end of that year and to 277,000 tonnes in March 2012.

It then embarked on two major warrant-sifting exercises on the LME, ironically, it seems, to get its hands on metal not stuck behind load-out queues. Its aluminium position peaked at 1.5 million tonnes towards the end of 2012.

In the course of those two LME gambits, Goldman itself cancelled just over 320,000 tonnes of aluminium at Detroit, the largest tranche coming over three days in December of that year.

The Detroit queue flexed out from 115 days at the start of 2012 to 370 by the end of the year.

Key drivers of that process were the Red Kite cancellations, the Glencore deal, two large-volume cancellations by JP Morgan (looking to move the metal to its own warehouses in Baltimore) and, of course, the Goldman cancellations.

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Graphic on key moments in the Detroit aluminium queue: http://link.reuters.com/fyh53w ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

CONFLICTED?

There is, by the way, no evidence whatsoever that Goldman was abusing any informational advantage it had through its ownership of Metro.

Goldman's Gabillon was at pains in last week's hearings to stress the efficiency of the Chinese walls separating the two businesses.

Indeed, the relationship between the warehousing arm and the trading arm was often tense with Metro officers resisting pressure from Goldman traders for free or discounted rent. Several left, including Mark Askew and Michael Whelan, head of business development, departures that undermined Goldman's desire to keep Metro independently managed.

Everyone concerned seemed fully aware of avoiding the potential conflicts of interest.

Yet, the conflict of interest was arguably hard-wired into a situation when a bank with multiple interests in the aluminium market was owning a warehouse operator, whose aggressive incentive schemes had made it a core driver of that market.

MONSTER

Does that mean that Metro and Goldman conspired to fix the price of aluminium?

It would be a hard case to prove, given we can't possibly know what would have happened to the price without the evolution of the Detroit queue.

What we do know is that the original Deutsche Bank merry-go-round started an evolutionary process that has fractured the "all-in" aluminium price into two parts, LME basis price and physical premium. It may never recover.

Metro and Goldman certainly created the queue monster but genuinely seem to have seen it only as a way of maximising the LME warehousing model, to begin with at least.

On paper it's been a profitable monster for Goldman. It paid $450m for Metro and, according to internal documents submitted to the subcommittee, Metro generated pre-tax profits (EBITDA) of $156m in 2010, $198m in 2011 and $211m in 2012.

However, Metro has also become a regulatory and media monster that has done nothing to improve Goldman's already tarnished public image.

Metro is officially for sale and there may well be a few Goldman executives who will be happy to see it go.

That, though, raises an even thornier question.

We may not like banks such as Goldman getting involved in the real-world business of running warehouse companies. But at least they can be grilled by U.S. politicians, as they were last week.

Will the next owner of Metro be so easily held to account, if it's not a U.S. bank? (Editing by William Hardy)