London, 18 April 2011

Goldman Sachs analysts' decision to withdraw their recommended long position in crude and some other commodities last week, triggering a sharp sell-off, has drawn a firestorm of criticism. The world's most powerful commodity trading bank sees continued upside for oil and many other commodity prices over a 12-month horizon. But near-term risks are more symmetric, and the risk-reward ratio is no longer appealing.

Goldman analysts fear oil prices have pushed ahead of fundamentals, citing record speculative positions and early signs of demand destruction. The bank makes no formal trading recommendations in energy and base metals but suggests investors adopt an underweight allocation to commodities on a three-to-six-month horizon, while expressing confidence the long-term bull trend remains intact.

Only a few weeks ago, the bank insisted the balance of risks in oil was tilted towards the upside. Its abrupt turnaround drew fierce condemnation from an unusually wide range of traders, commentators and analysts - some clearly wrong-footed by the subsequent price drop.

In a thinly veiled response, researchers at Barclays Capital, the oil market's other leading research team, restated their view the "market is currently massively out of equilibrium" and "the highs for oil prices this year are not yet in".

"On top of that the underlying political and economic uncertainties are so great that were we to call for a top to prices at this point, it would probably appear more than simplistic. Indeed, it might even look opportunistic as it would perhaps guarantee a few short-term headlines and some more headlines later when that view was reversed," Barclays wrote in a note Friday.

They went on to argue, "If analysis were to be judged solely in terms of the weight of headlines generated and their impact on the petroleum paparazzi, then following a route of frequent turns in a basic view might well be the best way to proceed."

But they warned, "Analysis that attempts to ground itself in reference to actual data does not have enough to call for a top". Both banks are fundamentally bullish and so are many investors, but a bitter argument has broken out over strategy and tactics in interpreting the balance of risks over differing horizons.

Divisions between the banking world's two most prominent teams highlight the complicated relationship between research, price forecasts, trading and the media.

RESEARCH FUNCTIONS

Research matters because it has the potential to move prices, though not all teams are equally influential. Goldman research remains unrivalled in its capacity to have a price impact.

The bank's warning unrest in Libya had halved OPEC's spare capacity and that the market could not absorb another disruption, creating significant upside risk, was linked with a $10 surge in spot Brent prices on Feb. 24. Its decision to close the long oil recommendation has been cited as causing a $4 drop in prices on April 11 and breaking the uptrend in prices.

Research is more than a passive activity. Forecasters, like the media, are participants as well as observers in the market: interpreting data, framing issues, shaping perceptions and expectations about the future, assembling coalitions of hedgers and investors, and mobilising large blocks of capital, sometimes creating self-validating expectations and price movements.

The idea that research plays any role in price formation is wholly alien to classical theories that insist prices are driven strictly by fundamentals. But it easily fits into behavioural approaches stressing the role of sentiment. The power of analysts and the media to shape perceptions, and ultimately fundamentals themselves, sometimes propagating bubbles, has been explored many times, including by Professor Robert Shiller in "Irrational Exuberance" and the theory of reflexivity propounded by legendary financier George Soros.

It is the reason regulators increasingly insist on a separation of researchers from sales and trading activity and journalists are generally forbidden from trading in assets about which they write. So research matters, though not always in the way analysts believe.

TURNCOAT ANGER

Research performs a variety of functions for banks. At the most basic, it serves as loss-leading advertising and brand building. Beyond that it provides sales and trading staff with a reason to touch base with clients regularly and can stimulate ideas and interest in trading and turnover. Research can give both hedgers and investors a reason to trade. It has occasionally led to criticism that analysts overdramatise potential upside/downside factors to stimulate business, since predictions of stable prices gives no one a reason to trade.

At its best, research can offer genuine insights and specialist expertise that at least some clients lack. But research also performs a powerful role in transmitting ideas and waves of bullish and bearish sentiment through the market eco-system. Like industry and investment conferences, media stories and occasional speeches, it helps establish and maintain dominant narratives that drive big price movements.

As big price movements become more extended, and market participants begin to worry about the possibility of a setback or sudden liquidation, research becomes increasingly important as a way of holding the dominant narrative bloc together and providing reassurance that key players are not defecting. Research can be a signal that everyone remains on the same side and the narrative still holds good, soothing fears about a sudden rush for the exits.

Researchers' role in building and sustaining influential coalitions of investors and hedgers explains why Goldman's decision to drop its long oil recommendation generated not just perplexity but real anger and bitterness in parts of the market. It was a decision to abandon the prevailing consensus in a way that threatened to trigger a rush for the exits that could harm those who remained bullish.

Goldman's influence in commodities has long been known. Its research, more than that of any other bank, is watched because of its intrinsic quality, its known influence with many hedgers and investors, and because it can (and demonstrably has) affected market prices. It becomes self-fulfilling. Research is influential because it is influential.

Defection by such a prominent and powerful participant poses a real danger to the remaining bulls, as shown by the sharp drop in prices and end of the uptrend after publication, though prices have since stabilized.

TACTICS OR TREND

Barclays' critique also raises the question of how research should relate to short and medium-term market movements. Most research teams have come under pressure to prove their value-added by publishing recommendations to counter the objection: "If you are so smart, how come you don't trade?"

But markets rarely rise or fall in straight lines. Should research recommendations focus on the medium-term trend or adopt a tactical overlay? Analysts who focus on the trend risk periodic embarrassment (like drawdowns for traders). Tactical overlays avoid that problem but risk exposing a tension between fundamental and sentiment components in the analyst's forecast.

Traders have learned to live with these ambiguities and the multiple factors acting on a market, not always consistently. But for analysts who are expected to rationalise movements, it can be harder.

Adding to the complication, traders are taught to run profits and cut losses. There is no shame in giving back a little at the top. But for analysts worried about reputational risk, there are strong incentives not to be caught still bullish at the top or bearish when the market hits a trough.

To protect reputation, analysts need to defect early, changing their calls before the market turns. But to be useful promoters and sustainers of consensus, they must only change once some or all of the members of the coalition have already exited.Most traders can relate to the tactical thinking behind Goldman's call. That doesn't mean to say they have to like it.

Ends --

By John Kemp, Reuters market analyst – for Commodities Now.

The views expressed here are his own.