25035641 24986279 Goldman Sachs Reseach Commodities Gold Oil

8/9/2019 25035641 24986279 Goldman Sachs Reseach Commodities Gold Oil

1/32

For Private WealthManagement Clients

Insight InvestmentStrategy GroupJanuary 2010

Commodities: A Solution inSearch o a Strategy


2/32

Commodities:A Solution in Search o a Strategy

Investing in gold and oil requires a much more

sophisticated approach than many popularcommodity-investing strategies suggest.

Is There a Strategic Role or Commoditiesin a Well-Diversied Portolio?The argument or a strategic allocation tocommodities isnt new. But there is very littleconvincing data to support long-only utures

exposure.

Commodities Have Not

Provided Real Returns Above Ination

Do Commodities Provide

Positive Roll Returns?

Do Commodities Provide

Downside Protection?

Our Strategic Asset Allocation

Recommendation

Are Oil and Gold OpportunisticInvestments Today?While large price fuctuations are inevitable inthe near term, we see no case or a tactical tilttoward gold or oil at this time.

Crude Oil

Peak Oil and the Long-Term Upside

Gold

3

6

7

11

12

13

15

16

19

22


3/32

Insight 3Investment Strategy Group

A Solution

in Search oa Strategy

Sharmin Mossavar-Rahmani

Chie Investment Ofcer

Goldman Sachs Private

Wealth Management

Additional Contributors rom

Goldman Sachs Private

Wealth Management:

Neeti Bhalla

Managing Director

Farshid Asl

Vice President

Milda Darguzaite

Vice President

Thomas Devos

Vice President

Benjamin Ngan

Associate

We do not know o any investment strategy

that can produce outsized returns in both

a strong economic recovery as well as a

signifcant economic downturn. Yet thepreponderance o the recent bullish commentary

on gold is driven by a heads you win, tails youwin investment thesis. The argument rst statesthat i world economies ully recover, infationwill be inevitable, debt will be monetized, thedollar will be debased and gold, as the only storeo real monetary value, will approach or evensurpass its infation-adjusted levels o about$2,400 per troy ounce (toz) seen in early 1980.You win. The argument then continues to assertthat in the alternative case, where the worldeconomies alter and slip back into a deep reces-

sion, all condence in governments and theirmonetary and scal policies will be eroded; peo-ple will abandon at money and take reuge ingold, driving it to or above the same $2,400/tozlevel in the process. You win again!

The incongruity o this win-win argumenttypies the incongruity o the gold mystique thathas existed since the Egyptians rst used goldbars as money as early as 4000 BC. The openingparagraph o the late Peter Bernsteins 2000book, The Power of Gold: The History of AnObsession, captures it well:

At the end o the 19th Century, John Ruskintold the story o a man who boarded a shipcarrying his entire wealth in a large bag ogold coins. A terrible storm came up a ewdays into the voyage and the alarm went oto abandon ship. Strapping the bag aroundhis waist, the man went up on deck, jumpedoverboard, and promptly sank to the bottomo the sea. Asks Ruskin: Now, as he was sink-

ing, had he the gold? Or had the gold him?1

Commodities such as gold and oilhave captured the imagination o manyinvestors. But investing in commoditiesentails a high degree o uncertainty,

and eective exposure in a portolio on either a strategic or tactical basis is much more nuanced than takinga simple long utures position.


4/32

4 Goldman Sachs

In yet another sign o the heightened levelo exuberance, words like peak gold (imply-ing a peak in gold production) and unprec-edented bids or bullion are now entering the

lexicon. Aaron Regent, the President and CEOo Barrick Gold Corp. (the worlds largest pro-ducer o gold), recently expressed his view thatthere is a strong case to be made that we arealready at peak gold,and in November, themining company completed the unwinding oall its hedges o uture production.5 This is therst time since the inception o Barricks hedgingprogram in 1987 that the company has removedall its hedges.

Even academics are joining the stampede

toward gold and, more broadly, commoditiesin general. Ater several research papers,including a 2004 National Bureau o EconomicResearch working paper titled Facts andFantasies about Commodity Futures that madethe case or a strategic allocation to commoditiesthrough the utures market, Gary B. Gortonand K. Keert Rouwenhorst o Yale UniversitysSchool o Management have joined aninvestment management rm to actively managecommodity utures or clients.

It seems to us that the current ascinationwith gold and other commodities stems rom twobroad concerns.

First, should an appropriately diversiedportolio have a strategic allocation tocommodities in order to improve the risk andreward trade-o in the portolio? For example,will such an allocation provide some downsideprotection to the portolio? Or will commoditieshedge the portolio against infation? Will thecommodity returns be uncorrelated with thoseo equities and bonds?

Second, should a portolio have a tacticalallocation to commodities, given the uncertaintyo the current global economic environment?With liquidity spigots wide open and worriesabout high infation, should a portolio containcommodities to protect the real value o theportolio? Shouldnt a portolio invest incommodities to benet rom the rising demandrom many emerging market countries like

China? Wont peak oil and peak gold andpeak-many-other commodities drive pricessubstantially higher?

We recognize that a day doesnt go by with-out some new captivating acts about gold. Sinceits trough in August, 1999, gold has increased333% in nominal prices (a 15.2% compounded

annual growth rate), has outperormed otherprecious metals by 2.7% (in the case o silver)to 0.6% (in the case o platinum), oil by 1.7%,equities by 17.2%, and US home prices nosurprise by 10.6%, all on an annualized basis.2Some might even say that the level o interest ingold is reaching renzied proportions. The year-to-date fow o unds into commodity-relatedexchange traded unds (ETFs) through November2009 was $28 billion, including direct commodityas well as equity-related commodity unds. This

compares to a total net outfow o $42 billionrom US equity unds and infows o $42 billioninto non-US equity unds (both gures includeETFs and actively managed unds).3 The SPDRGold Trust Fund GLD, the massive gold bullionexchange traded und, now holds the worldssixth largest gold reserves at about 1,118 metrictons just less than the reserves held by the US,Germany, the International Monetary Fund, Italyand France, but more than China and Switzerland.Physical gold-backed ETFs, in aggregate, now

hold a total o 1,495 metric tons.Similar fows have been occurring in

commodity hedge unds, prompting some suchunds such as Clive Capital, the worlds largest to close their doors to new investors. Otherdiversied hedge unds have launched gold-onlyunds and several prominent unds have increasedtheir allocations to gold and gold-related equities.Even diversied long-only equity unds havestepped up investments in actual bullion as wellas gold-related stocks, according to bothMorningstar and Financial Research Corp.

In terms o investor interest in physical gold,demand has reached such high levels that in lateNovember 2009, the US Mint had to suspendsales o 2009 American Eagle one-ounce bullioncoins ater its inventory was depleted. O course,storage o all this physical gold has becomean issue; in November, The Wall Street Journalreported that HSBC had asked their retailinvestors to remove their physical holdings o

coins and bullion rom HSBC vaults in NewYork to make room or gold holdings o theirinstitutional clients.4

january 2010


5/32


In this edition o Insight, we explore thesequestions in detail. We begin by analyzing thedata to see the extent to which a strategicallocation to commodities provides diversica-

tion benets as well as a hedge against infation;and as our title suggests, we nd that an eectivestrategic approach to commodities is much morenuanced than simple, long utures exposure.We then examine the current supply and demandactors as well as the nancial market dynamicsaecting gold and oil to see i a tactical alloca-tion to either one is warranted at this time. Whilewe believe that gold could well see continuedprice appreciation in the near term, we do notthink direct, long exposure to either commodity

is suitable or most investors.


6/32

6 Goldman Sachs

january 2010

Is There a Strategic Role orCommodities in a Well-DiversiiedPortolio?

Commodities as a viable asset class has beena topic o discussion or many decades. As arback as 1930, John Maynard Keynes stated thatnormal backwardation in the utures priceso commodities (where utures prices are lowerthan expected spot prices or a particularcommodity, as shown in Exhibit 1) wouldprovide risk premiums to those investors whowere, in eect, enabling commodity producersto hedge their risk o uture price fuctuations.

An oil producer, or example, which is intrinsi-cally long crude oil by virtue o its business, canreduce its exposure to oil price fuctuations byselling a series o crude oil utures. The investorwho is willing to buy those oil utures contractsis thereore providing a hedging service to oilproducers and should get paid or that service hence the normal backwardation risk premium.

In 1939, Nicholas Kaldor presented an alter-native and somewhat contradictory theorystating that convenience yield is a source o

risk premium in the commodity utures market.His view was that commodities that are dicultand expensive to store such as crude oil will havegenerally lower inventory levels; thereore, themarket has to incentivize a producer to hold suchinventories given storage costs, oregone interestand uncertain spot prices. This incentive appearsin the orm o a so-called convenience yieldembedded in the dierence between uturesprices and higher expected spot prices. Thisyield is eectively realized by the producer asthe utures prices ride up the backwardationcurve toward the spot price. Where inventoriesare plentiul and storage is relatively easy, onthe other hand, this convenience yield may benon-existent or even negative the produceris eectively losing value as the higher uturesprices slide down to the spot price. Such uturescurves are described as having contango (seeExhibit 2).

In more recent history, there has been a

ocused discussion on the use o commodities ina diversied portolio. At one extreme, a 2000paper by Gerald R. Jensen, Robert R. Johnson

Rolling up from 2nd-to-1st Month(Backwardation) +$3.50 (+17%)

Months to Expiration

$/Barrel

$15

$16

$17

$18

$19

$20

$21

$22

$23

$24

$25

121110987654321

West Texas Intermediate Crude

Exhibit 1: Backwardation

Backwardation occurs when the price o utures contracts rises to

the expected spot price as the contract expiration date approaches.

An extreme example o backwardation happened in oil utures inMarch 1996, when US commercial petroleum inventories ell 10%

below their ive-year seasonal average, driving spot prices up relative

to the orward curve.

Rolling down from 2nd-to-1st Month(Contango) -$8.50 (-20%)

Months to Expiration

$/Barrel

$30

$35

$40

$45

$50

$55

$60

121110987654321

West Texas Intermediate Crude

Exhibit 2: Contango

Contango occurs when the price o utures contracts alls to the

expected spot price as the contract expiration date approaches.

This phenomenon was acutely seen in oil utures in late 2008, when

poor demand sent US inventory levels 10% above average, putting

downward pressure on spot prices.

Data rom December 19, 2008

Source: Investment Strategy Group, Bloomberg

For illustrative purposes only.

Data rom March 19, 1996

Source: Investment Strategy Group, Bloomberg

For illustrative purposes only.


7/32


commodities to an already well-diversiedportolio o bonds, stocks, hedge unds, privateequity and real estate does not improve the risk/return prole o a portolio unless one assumes

a meaningul and in our view, unrealistic riskpremium.

Commodities Have Not Provided Real Returns

Above Ination

Lets rst begin with the real returns ocommodities. The longest publicly available dataset on commodities is The Economists Index oIndustrial Commodity Prices, with guresreaching back to 1845. According to this data,which has also been used or research by the

International Monetary Fund and the WorldBank, real commodity prices have been ona downward trend since 1871. As shown inExhibit 3, real commodity prices have declined lagged infation, in other words by 0.4% a yearover the last 140 years. Real prices or agricul-

and Jerey M. Mercer suggested that investors,depending on their risk tolerance, should investanywhere rom 5% to as much as 36% o theirportolio in commodities using the S&P/GSCI

Total Return Index as the commodities bench-mark. In the more widely quoted 2004 paperby Gary B. Gorton and K. Keert Rouwenhorst(reerenced above), the authors conclude thattheir index o equally weighted commodityutures provides the same risk premium as equities,has less volatility than equities and has negativecorrelation with equities and bonds implying avery signicant added strategic value to a portolio.

Contrast this to a more recent paper byClaude Erb and Campbell Harvey, The Tactical

and Strategic Value of Commodity Futures,published in the Financial Analysts Journalin2006 and winner o the prestigious Grahamand Dodd Award, which shows that the averagereturns o individual commodity utures havebeen indistinguishable rom zero. The authorsurther explain that the dierent returns o thevarious commodity indexes are driven primarilyby the dierent weightings o the utures contractsin the indexes and by dierent rebalancingmethodologies, rather than the perormance o

the contracts themselves. They conclude thatthe returns to the various indexes are thereorereturns o dierent active portolio strategies andnot returns o commodity utures per se.

Having thoroughly analyzed the availabledata and combed through the existing literature,we nd ourselves generally aligned with thisperspective. We have concluded that there is noconsistent and reliable argument or a strategicallocation to a commodity utures index in awell-diversied portolio.

We make our case in three parts below.First, we show that one cannot assume thatcommodities provide a positive expected returnabove infation. Second, we demonstrate thatthe return rom backwardation (the roll upthe utures curve shown in Exhibit 1) is notconsistently and reliably positive; in act, inrecent history and across many commodityutures, the roll return has been negativedue to contango in the utures market. Third,

we show that, unlike high-quality bonds,commodity utures do not provide consistentand reliable downside protection; in act, adding

1871 1881 1891 1901 1911 1921 1931 1941 1951 1961 1971 1981 1991 2001

0

50

100

150

200

250

Annualized Real Return = -0.4%

RealCommodityPrices(1871=

100)

Exhibit 3: Decline in Real Prices o Industrial

Commodities

Data through Year-End 2007

Source: Investment Strategy Group, The Economist, Bureau o Labor Statistics, Robert Shiller (Yale University)


8/32

8 Goldman Sachs

january 2010

tural commodities and livestock have declined,respectively, by 2.1% and 3.3% a year since1947. Gold has actually exceeded infation by0.9% a year since 1871, but as shown in Exhibit

4, prior to the run-up o the last nine years, itessentially matched infation. Oil has shown thestrongest sustained perormance, exceedinginfation by 2.6% a year since 1947. But thisperormance was driven by two distinct geopo-litical events in 1973-74 (the Arab Oil Embargo)and 1978-80 (the Iranian Revolution and theIran-Iraq War). In the three decades since 1980,oil prices have lagged infation by 1.04% a year.

Another important actor to note is that,in the long run, commodities have a positive

correlation with infation o 0.38, but thiscorrelation is not particularly stable. Lookingat a graph o 5-year rolling correlations oquarterly S&P/GSCI versus the U.S. ConsumerPrice Index (CPI) in Exhibit 5, one can see thatcommodities have had correlations as high as0.82 and as low as -0.44. We would expectan eective hedge to have a higher overallcorrelation that is also more stable.

While commodities do not keep pace withinfation and thereby do not provide any real

returns in the long run, there are periods wherecommodities do substantially outperorm infa-tion. As can be seen in Exhibits 6 and 7, oil andgold dramatically outpaced infation duringthe high infationary periods o the 1970s; butboth lagged signicantly over the subsequenttwo decades, as shown in Exhibits 8 and 9.Such perormance has led to a view that com-modities might be a particularly eective hedgeagainst unexpectedinfation. To test this view,we examined periods where infation exceededexpectations and note that commodities are not,in act, a consistent and reliable hedge againstsuch unanticipated infation.

In Exhibit 10, we see that certain commoditieshedge unanticipated high infation some butnot all o the time. For example, between 1973and 1981, when infation averaged 9.3%, bothoil and gold provided an excellent hedge, sincethis was a time o commodity-induced infation(it included the Arab Oil Embargo, the Iranian

Revolution and the Iran/Iraq War), while in theearly 1950s, neither oil nor gold provided anyhedging against unanticipated infation. In the

Exhibit 4: Real Value o $1 Invested in Gold

Exhibit 5: Correlation o Commodities with Ination

RealGold

Prices(1871=

1.0

0)

0

1

2

3

4

5

6

20011991198119711961195119411931192119111901189118811871

Dollar Devaluation

(Roosevelt)

Collapse of Bretton

Woods (Nixon)

Annualized

Real Return= 0.9%

Data as o November 2009

Source: Investment Strategy Group, Bloomberg, Datastream, Bureau o Labor Statistics, Robert Shiller

(Yale University)

Past perormance is not indicative o uture results.

Data as o Q3 2009

Source: Investment Strategy Group, Datastream, Bureau o Labor Statistics

Rolling

5-

YrCorrelation

-0.6

-0.4

-0.2

-0.0

0.2

0.4

0.6

0.8

1.0

090705030199979593918987858381797775


9/32


Source: Investment Strategy Group, Datastream, Bureau o Labor Statistics Source: Investment Strategy Group, Datastream, Bureau o Labor Statistics

Source: Investment Strategy Group, Datastream, Bloomberg, Bureau o Labor Statistics Source: Investment Strategy Group, Datastream, Bloomberg, Bureau o Labor Statistics

Exhibit 6: Oil Prices Exceed Ination in the 1970s

0

200

400

600

800

1000

1200

8079787776757473

Energy Prices

CPI

Annualized

Energy Price

Change = 37.8%

Annualized CPI Change = 9.2%

Energy

Prices/CPI(Jan

73

=

10

0)

Exhibit 7: Gold Prices Exceed Ination in the 1970s

Gold Prices

CPI

Annualized

Gold Price

Change = 34.6%


0

500

1000

1500

2000

2500

8079787776757473727170

GoldP

rices/CPI(Jan

70

=

100)

Exhibit 8: Oil Prices Lagged Ination in 19801998

Energy Prices

CPI

Annualized

Energy Price

Change = -5.4%


0

50

100

150

200

250

7/987/957/927/897/867/837/80

Energy

Prices/CPI(Jul80

=

100

)

Exhibit 9: Gold Prices Lagged Ination in 19801999

Gold Prices

CPI

Annualized

Gold Price

Change = -5.0%


Gold

Prices/CPI(Jan

80

=

100)

0

50

100

150

200

250

9998979695949392919089888786858483828180

Exhibit 10: Annualized Real Returns During Unanticipated Inationary Periods

Source: Investment Strategy Group, Datastream, Barclays Capital, Ibbotson, CRB, Bureau o Labor Statistics, University o Michigan


Non-Energy

Prices

-10.7%

-5.4%

-2.4%

US Equities

-21.8%

19.7%

-3.1%

6.4%

Energy Price

36.8%

-6.4%

20.9%

3.1%

Gold Prices

-13.9%

-6.6%

11.9%

-6.7%

CPI Inflation

13.9%

6.4%

9.3%

4.6%

Apr 46 to Feb 48

Oct 50 to Jan 52

Mar 73 to Nov 81

Mar 87 to July 91

US

Treasuries

-13.0%

-5.6%

-2.5%

3.2%


10/32

10 Goldman Sachs

january 2010

more recent period o higher-than-expectedinfation between 1987 and 1991, oil was theonly major commodity to exceed infation;however, both equities and bonds outperormed

commodities during that time.Looking at the correlation o quarterly S&P/

GSCI with changes in the CPI, the average long-term correlation is 0.31, with a high o 0.69 anda low o -0.31 on a ve-year basis. I we changethe 5-year rolling correlations to 2-year or10-year rolling correlations and change thequarterly returns to monthly or annual returns,the correlation levels and the stability o thecorrelation change signicantly. Selecting suchdierent time rames produces very divergent

results; this leads us to conclude that the empiricalevidence does not support the argument thatcommodities are an eective hedge againstunanticipated infation.

Even i they were reliably eective in suchcases, we take issue with any hedge that onlyprotects a portolio against unanticipated infa-tion. Investors are aected by and should careabout both expected and unexpected infation.Even i there was an asset class that was aneective hedge against only unexpected infation,

it would be o limited use during periods o highand persistent but not surprising infation,which is just as damaging to a portoliospurchasing power.

One possible explanation or the absence oany consistent hedging benets rom commoditiescan be ound in the composition o the Con-sumer Price Index. Sixty percent o the index iscomprised o services, including medical care,education, communications, transportationservices and housing, while only 40% is repre-sented by commodities. Infation among theseservices has exceeded commodities infationsince 1935, and currently outpaces it by 0.9%.

Based on historical data and the structure othe CPI, we thereore conclude that commoditiesare not a consistent and reliable hedge againsteither expected or unexpected infation.

Exhibit 11: Frequency o Backwardation and

Contango

Exhibit 12: Roll Return rom Backwardation

and Contango


Source: Investment Strategy Group, Datastream





0

10

20

30

40

50

60

70

80

90

100

S&P/GSCI

Commodities

Aggregate

Agriculture Energy Industrial

MetalsLivestock Precious

Metals

%o

fTimeinBackwardation/Contango(M

onthly)

Average Monthly Future Roll Return

-0.03% -0.38% 0.02% -0.10% -0.03% -0.50%

Contango

Backwardation

-25

-20

-15

-10

-5

0

5

10

15

20

25

30

1/081/031/981/9312/87Rollin

g5-YrBackwardation/ContangoRollReturn(%)

S&P/GSCI Commodities Aggregate

Energy

Industrial Metals

Agriculture


11/32


Do Commodities Provide Positive Roll Returns?

Another perceived source o returns tocommodity utures has been the so-called rollreturn as utures converge to the spot price at

their expiration. The roll return is positive whenutures prices are lower than current spot pricesand the shape o the utures curve is in backwar-dation, as shown in Exhibit 1. The roll returnis negative when utures prices are higher thancurrent spot prices and the shape o the uturescurve is in contango, as seen in Exhibit 2.As mentioned earlier, there have been varioustheories about why some commodities that arehard to store, such as oil, should be in backwar-dation while some commodities that are easier

to store, such as gold, should be in contango.But again, here, the empirical evidence regardinga positive return rom the shape o thecommodities utures curve is mixed.

Looking at the S&P/GSCI, in aggregate andby its components, we can see that since theinception o the index data in 1970 most o thecommodities spend more calendar months incontango than in backwardation, and thus theaverage monthly roll return has been negative, asshown in Exhibit 11. Precious metals, especially

gold, and agricultural commodities have exhibitedthe most contango both in terms o requencyand magnitude. In other words, uture prices orthese commodities were higher than their spotprices as much as 86% and 69% o the time,respectively, and the average monthly roll returno the utures contract was negative at -0.50%and -0.38%.In energy, however, the uturescontracts were in contango much less, at 53% othe time, and the roll return has been marginallypositive at 0.02%.

As with all data, the averages can mask theunderlying trends. In Exhibit 12, one can see therolling ve-year roll returns o the S&P/GSCIand its three key components. Since the early1990s, the magnitude and requency o contangoseems to have increased, resulting in negative rollreturns. This is most pronounced in the agricul-tural and energy sectors. Exhibits 1 and 2,shown earlier, show an excellent example o ashit in the oil utures contract rom extreme

backwardation to extreme contango.

Three arguments have been put orth toexplain the declining trend in the roll return.The rst is that producers, especially oilcompanies, are not hedging their uture oil

production as much as in the past because theyare not expecting any signicant decline inprices; as such, the Keynesian normalbackwardation would no longer exist and therisk premium thereore disappears. The secondargument is that the increase in the allocation tovarious commodity utures indexes by institu-tional investors has taken away the risk premiumas many investors try to capture the same pre-mium. In their 2006 paper mentioned earlier, Erband Harvey suggest a third possibility: that the

declining trend might just be statistical noise.To address this increased contango in the

ront end o many commodity utures, institu-tions have begun to oer enhanced indexes thatroll their contracts urther out on the curve. Wethink o such enhancements as the evolution oan active management strategy to generatingexcess returns and not an investment rationaleor commodities as an asset class.

Based on historical data and the absenceo a theoretically convincing argument, we

thereore conclude that one should not investin commodities as an asset class in anticipationo positive roll returns.


12/32

9.12% 21.74% -10.95% -29.25% 2.63%

139.46% 183.99% 139.46% 137.47% -42.63% 6.06%

15.72% 6.83% 15.72% 33.33% -14.13% 2.97%

-19.70% -5.11% -19.70% -44.84% -16.91% 24.02%

1.78% -0.25% 3.87% 8.39% -29.53% 2.36%

43.51% 96.45% -2.95% 3.45% -14.69% 4.05%

-13.30% -15.60% -11.07% -6.44% -15.37% 2.88%

-7.88% -8.62% -8.55% 15.28% -44.73% 24.84%

-53.40% -60.33% -35.32% 18.51% -50.95% 13.30%

Exhibit 14: Perormance o Commodities During Periods o Signiicant Equity Drawdowns

1Commodities perormance represents S&P/GSCI Total Returns whenever applicable; prior to index inceptions (Jan 83 or S&P/GSCI Energy and Jan-70 or S&P/GSCI Non-Energy), commodities perormance is

represented by price returns.

Source: Investment Strategy Group, Datastream, Bloomberg, Barclays Capital, Ibbotson, CRB, Bureau o Labor StatisticsPast perormance is not indicative o uture results. Indices are unmanaged. The igures or the index relect the reinvestment o dividends but do not relect the deduction o any ees or expenses which would

reduce returns. Investors cannot invest directly in indices.

S&P/GSCITotal Return Energy1 Non-Energy1 Gold US Equities US Treasuries

1968 Recession

1972 Oil Embargo

1976 Oil Embargo (Cont'd)

1980 Stagflation

1987 Black Monday

1990 Gulf War

1998 Russian Financial Crisis

2000 Dot-Com Bubble

2007 Financial Crisis

Date

Nov 68 to Jun 70

Dec 72 to Sep 74

Dec 76 to Feb 78

Nov 80 to Jul 82

Aug 87 to Nov 87

May 90 to Oct 90

Jun 98 to Aug 98

Aug 00 to Sep 02

Oct 07 to Feb 09

Cumulative Return During Equity Drawdowns

12 Goldman Sachs

january 2010

Do Commodities Provide Downside Protection?

While commodities may not deliver a positiverisk premium, it is possible that the low correla-tion o commodities with a balanced portolio o

stocks and bonds provides some diversicationbenet and improves the risk and return proleo a portolio. Indeed, the rolling 2-year correla-tion o the S&P/GSCI with a 50% stock/50%bond portolio since 1970 is relatively low at0.02, with a high o 0.59 and a low o -0.65, asshown in Exhibit 13. And the historical risk/return prole o a 50% stock/50% bondportolio does improve with an allocation tocommodities. Between 1970 and November2009, a 10% allocation to commodities unded

out o stocks and bonds improved returns by0.37% per year while holding annualizedvolatility at the same level as the 50% stock/50%bond portolio.

In periods o signicant equity marketdrawdowns, however, high quality bonds area more reliable source o diversication. Asshown in Exhibit 14, we have examined thepost-World War II periods with signicantmarket dislocations and geopolitical events.In some cases, such as during the 1973-1974

bear market that was partly triggered by theArab Oil Embargo and the tripling o crudeoil prices, commodities provided tremendous

downside protection. In other cases, such asin the stagfation period between 1980-1982,commodities underperormed equities. In the recentnancial crisis, commodities again underperormedequities, and the 12-month rolling correlationbetween commodities and equities reached 0.76 inAugust 2009. Commodities did not provide anydiversication benets in the biggest equity marketdowndrat since the Great Depression, when it was

needed most.The results are quite mixed, in other words.

And i we look at the biggest historical draw-

Exhibit 13: Correlation o S&P/GSCI with a

50% Stock/50% Bond Portolio



-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

08060402009896949290888684828078767472

Rolling

2-

YrCorrelation


13/32


down o a 50% stock/50% bond portolio andcompare it to a similar-risk portolio containingcommodities, we nd that the maximum draw-down o the stock/bond portolio was -27.1%

while that o the portolio with stocks, bondsand commodities was -32.7%.

In a more orward-looking analysis, we ndthat a well-diversied portolio comprised ostocks and bonds as well as hedge unds, privateequity and real estate does not benet rom anallocation to commodities even with such lowcorrelations, unless we assume a non-negligiblepositive risk premium. In the absence o a riskpremium, the addition o commodities does notimprove the Sharpe ratio (i.e. the return per unit

o risk where risk is dened as the overallvolatility o the portolio) o a well-diversiedportolio.

Our Strategic Asset Allocation Recommendation

Based on the absence o an identiable riskpremium, a reliable infation hedge or even aclear diversication benet, we do not recom-mend a strategic allocation to a commodityutures index. However, that is not to saythat investors should have no exposure to

commodities; in act, most already have signi-cant exposure through their equity holdings.As shown in Exhibit 15, commodity-relatedstocks account or 19.5% o developed equitiesand 29.3% o emerging market equities.

When investors think o commodity-relatedstocks, however, it is important to bear in mindthat a bigger portion o the returns o manysuch stocks comes rom their overall equitymarket exposure rather than rom their exposureto commodities. Take energy stocks. At 10.9%o developed equities and 15.9% o emergingmarket equities, energy stocks account or thelargest portion o commodity-related stocks (inthe S&P/GSCI, a production-weighted index thatis re-weighted annually, energy commodities arealso the largest component o the index at 70%).When we decompose the return o energy stocks,we nd that we can attribute 36% o the returnsto broad equity market returns (as measured bythe S&P 500) and about 28% to energy

commodities (as measured by the S&P/GSCI energycommodities). The remaining 36% is attributableto the idiosyncratic risk o the energy stocks.

Alternatively, we can look at the correlationo energy stocks with the broader market versustheir correlation with energy commodities. Thecorrelation with equities between January 1983

and November 2009 is 0.60 and has rangedbetween 0.22 and 0.87 on a rolling two-yearbasis, whereas the correlation with energycommodities is lower at 0.48, with a low o 0.01and high o 0.79 on a rolling two-year basis.We start with 1983 because that is the inceptiondate o the energy component o the S&P/GSCI.Thereore, it is important to recognize that byinvesting in commodity-related stocks, an inves-tor, in general, is assuming more equity exposurethan commodity exposure, but the commodity

exposure is statistically signicant and meaningul.

Global Equity Sectors

Energy

Integrated Oil & Gas

Exploration & Production

Oil Equipment & Services

Coal

Other Energy

Metals & Mining

General Mining

Iron & Steel

Gold Mining

Platinum & Precious Metals

Other Metals & Mining

Others

Specialty ChemicalsCommodity Chemicals

Building Fixtures & Materials

Farming & Fishing

Other Commodities

Total

% of Developed

World Market

Cap

6.10%

2.94%1.25%

0.33%

0.24%

1.61%

1.60%

0.81%

0.17%

0.13%

1.55%1.21%

1.00%

0.28%

0.23%

19.45%

% of Emerging

Markets Market

Cap

9.78%

4.86%0.13%

1.14%

-

0.25%

6.49%

1.02%

0.66%

0.15%

0.70%

1.67%

1.55%

0.44%

0.45%

29.28%

Exhibit 15: Commodity-Related Sectors in

Equity Markets

Data as o December 2009



14/32

14 Goldman Sachs

january 2010

Given the tremendous interest in gold andgold stocks, we perormed the same analysis onBarrick Gold. We selected Barrick Gold or tworeasons: It is the largest gold producer in the

world, and it is more o a gold mining companythan a general precious metals or industrialmetals mining company. When we decomposethe returns o Barrick Gold, we see a dierent

do not make strategic asset allocation recom-mendations based on portolio strategies.

To take advantage o active commoditystrategies, we recommend an allocation to two

types o hedge unds: 1) macro-tactical tradinghedge unds where the hedge und managersgoal is to provide uncorrelated returns throughinvesting in a broad range o assets includingcommodity-related assets based on macroeco-nomic themes; and 2) commodity-ocused hedgeunds where the hedge und managers goal is toprovide uncorrelated returns through investingexclusively in a broad range o commodity-related assets (e.g., commodity utures, commodityoptions, commodity equities and in some cases,

direct physical commodities).6Commodity hedge unds can also use an array

o strategies to take advantage o the high levelo volatility in commodities. Annualized volatil-ity in commodities ranges rom a high o 56.4%in natural gas, to 34.3% or crude oil, to 19.7%or gold to 15.1% or eeder cattle.7 The aver-age o the volatility o the 24 commodities inthe S&P/GSCI is 29.9% and the volatility o theS&P/GSCI itsel is 20.0%, due to the diversica-tion benet o a basket o commodities (this is

similar to the diversication benet o a basketo stocks: the average o the volatility o thestocks in the S&P 500 is higher than thevolatility o the S&P itsel).

These hedge unds use strategies includingrelative value trading based on historical rela-tionships, price momentum, shape o the uturescurves and undamental research into supply,demand and inventory levels. But just as highvolatility creates investment opportunities orsavvy managers, it can also lead to signicantlosses. Silver in 1980 and oil in the mid-80sdrove the Hunt brothers (sons o legendaryoil tycoon H.L. Hunt) to eventual bankruptcy.Copper was responsible or a $1.8 billion lossaccumulated over 10 years that was uncovered atSumitomo in 1996. Natural gas was responsibleor the collapse o Amaranth Advisors LLC andMotherRock LP in 2006. Clearly, an investorshould proceed with caution and signicant duediligence when selecting commodity-oriented

hedge unds.

story than we did with energy: 20% o thereturns is attributable to the equity market and40% is attributable to changes in the price ogold. In this case, the gold stock is more o acommodity play than a broad equity play.Similarly, the correlation o the stock with goldis 0.59, much higher than its 0.21 correlationwith US equities.

In addition to exposure through equities,we believe that active strategies within the

commodity utures complex provide a signi-cant opportunity to investors. For example, theequal-weighted index constructed by Gortonand Rouwenhorst produced substantial returnsabove both infation and the spot returns o com-modities. In our view, their index does not pointtoward a passive allocation to commodities; itactually demonstrates the historical returns o anactive management strategy where the strategy isto equally weight a basket o commodity uturesand rebalance the positions on a monthly orannual basis.

We should note that Erb and Harvey alsobelieve that commodity indexes are actuallydierent portolio strategies: Unlike equityindexes, none o the commodity indexes arebased on market capitalization since there is nomarket capitalization in the commodity uturesmarket (or every investor that owns a longutures contract, there is a correspondinginvestor that is short the same utures contract;

the net exposure adds up to zero!). Thereore,the design o each commodity index is a portoliostrategy around a basket o commodities. We

We believe that active strategieswithin the commodity utures complexprovide a signiicant opportunity toinvestors.


15/32


Are Oil and Gold OpportunisticInvestments Today?

As we set orth our tactical views on oil and gold

the two most widely discussed commodities inthe last year or so we are reminded o a com-ment by Rex Tillerson, the CEO o Exxon MobilCorp. in March 2008. As crude oil prices movedabove $100/barrel on their way to a peak innominal prices o $147/barrel, Tillerson reerredto the continued increases in the price o oil aspretty crazy; he attributed one-third o the risein prices at the time to a weak dollar, one-thirdto geopolitical uncertainty and the remainingthird to market speculation.8 More recently, the

head o Saudi Aramco, Khalid Al-Falih, statedthat with so many nancial investors andplayers who are not industry participantsyou can never predict what the oil price is.9

The insight o these two chie executivesshould not be underestimated. Exxon Mobil isthe largest publicly traded non-state controlledoil company in the world and ranks third in thePetroleum Intelligence Weekly ranking o the top50 largest oil companies in the world (includingnational oil companies). Saudi Aramco, Saudi

Arabias national oil company, is ranked numberone with the worlds largest oil reserves and theworlds largest estimated production capacityo 12.5 million barrels per day. When chieexecutives o two such oil companies reer to theimpact o nancial investors on market volatilityand oil prices, we take heed and recognize theimmense uncertainty in orecasting commodityprices in this environment.

We have thus proceeded with due cautionand have concluded that the broad-based bullishview on commodities is not warranted at thistime. In our view, the commodity markets willremain volatile or the oreseeable uture;however, we are not expecting a sustainedincrease in commodity prices over the next yearor so. Several actors contribute to this view:supply and demand orces in the oil market,the shape o the utures curve across manycommodities, our generally avorable outlookor the dollar in the next 12-24 months and our

view o technological innovation as an antidoteto predicted shortages. We recognize that theconsensus view tends toward higher returns or

the next several years due to strong demandrom emerging markets (particularly China),continued weakness o the dollar and infation;it is not surprising that at a recent Barclays

Capital Commodities Investor Conerence,59% o those surveyed had increased theircommodities exposure over the last 12 monthsand 63% expected to initiate or increase theirexposure.10

In analyzing commodities and ormulatingour tactical views, we know that we cannot relyon one consistent ramework to guide us in allenvironments. At times, undamental supply anddemand actors and various valuationmetrics dominate the discussion; but at other

times, much more unpredictable orces such asgeopolitical events, weather, central bankdecisions and the proverbial ear and greeddrive prices. In good humor and an apt choice owords, The Wall Street Journalreerred to someo these actors as cosmic orces!11

Beore we jump into our tactical views on oiland gold, we should note that while people talkabout bullish and bearish views on commoditiesor total return targets or the broad commoditiesindexes, we believe that we have to look at

individual commodities separately. Commoditiesare less homogenous and correlated to each otherthan, or example, stocks in a typical equityindex. In 2009 through Dec. 22, the overallS&P/GSCI spot return was up 42.6% (thetotal return index was up only 7.7% due to thecontango in the shape o the curve that draggeddown returns). The spot returns are comprisedo a positive 65.3% in crude oil but only a 1.6%return in natural gas within the energy sector, apositive 2.9% in soybeans but a negative 14.9%in wheat within the agricultural sector, and apositive 5.5% in lean hogs versus a negative0.6% in live cattle within the livestock sector.

This dispersion o returns is not atypical incommodities. In 2007, copper was up 6% butzinc was down 45%; in 2001, coee was down30% but cocoa was up 73%. We measure thisdispersion with more analytical rigor by lookingat cross correlations. The cross correlation o theve commodity sectors (energy, industrial metals,

precious metals, agriculture and livestock) inthe S&P/GSCI Index is 0.18, while that o the10 broad equity sectors (technology, materials,


16/32

16 Goldman Sachs

january 2010

consumer discretionary, industrials, healthcare,nancials, consumer staples, energy, utilities andtelecom services) in the S&P 500 is 0.58. There-ore, we think it is not as useul to talk about a

broad view on commodities in the same way thatwe talk about our general views on US equities,or example, or developed market equities orhigh yield debt.

In this report, we have ocused our tacticalviews on oil and gold: oil because it is the largestcommodity produced annually as measured bytotal dollar value o annual production and asa percentage o the production-weighted S&P/GSCI (it is 65% o the index); and gold becauseit is the commodity du jour, recommended

by many as the win-win investment strategy o2010.

Crude Oil

From its trough on Dec. 22, 2008 through Dec.22, 2009, crude oil spot prices, as refected in theWest Texas Intermediate price and reported byBloomberg, were up 135%. But that impressivereturn bears qualication. First, it was onlyavailable to those who bought a physical barrelo oil at the Cushing, Oklahoma terminal. For

those investing in oil utures, as measured by theS&P/GSCI total return, oil was up only 13%over that period (the dierence, as mentionedearlier, is due to the contango o the uturescurve in oil). It was also a feeting opportunity; iwe use the price on Dec. 31, 2008 as our startingpoint, the crude oil price increase was less thanhal, at 66%, highlighting the immense volatilityin commodities.

One can point to ve actors as plausibleexplanations or the increase in prices to thecurrent level o $74/barrel. First and oremost,we believe that a signicant part o the increaseis driven by the virtual elimination o the risko a Great Depression-style global economicmeltdown. Other possible actors are a decreasein the value o the dollar, ears o high infation,expected increases in demand rom the emergingmarket countries and resumed talk o underin-vestment in the energy sector and peak oil. It isalso important to remember that while oil has

recovered rom its December 2008 lows, it is stillabout 50% below its July 2008 peak.

Lets rst look at the relationship betweenoil and the dollar. The dollar is, in act, slightlynegatively correlated with oil with an aver-age 12-month rolling correlation o -0.11. The

latest 12-month correlation stands at -0.47. Asoil prices have increased rom their lows in lateDecember 2008, the dollar has depreciated by5.2%. Since its peak level during the nancialcrisis in March 2009, the dollar has depreciatedby 12.8%.

While the current increase in the price ooil has coincided with a all in the dollar, thisrelationship is not very stable; there are manyperiods where oil and the dollar have been posi-tively correlated. From November 1985 through

While the current increase in theprice o oil has coincided with a allin the dollar, this relationship is notvery stable.

April 1987, or example, the dollar depreciatedby 21% and oil dropped by 37%. No negativecorrelation there. Similarly, between December

1998 and February 2002, the dollar rallied 17%against a trade-weighted basket o currenciesand oil rallied 80%! The explanation or thisshiting correlation is that many other actorsdrive changes in the price o oil. In act, onaverage, movements in the dollar explain only16% o the changes in the price o oil.

We do not recommend overweighting crudeoil as a hedge against urther deterioration inthe dollar. First, the dollar-oil relationship is notstable enough to be a primary driver o returns.Furthermore, we believe that the dollar is, inact, cheap, and our long-term view o it is avor-able. While the dollar might depreciate urther,we believe that the US economy will outperormEurope, the UK and Japan, and its currency willadjust accordingly sometime over the next 12months.

Another explanation or the rally in oil hasbeen an expectation o higher infation. With anexpected debt/GDP ratio o over 70%, a Federal

Reserve balance sheet o $2.2 trillion as o year-end 2009 and a near-zero interest rate policy,


17/32


some have argued that infation is inevitableand commodities both oil and gold are aneective hedge against it.

As shown in our discussion above, commodi-

ties are not an eective hedge against generalizedinfation; they are an eective hedge againstcommodity-induced infation as witnessed in the1970s during the Arab Oil Embargo, the IranianRevolution and the early part o the Iran-IraqWar. So buying commodities to hedge againstgeneralized infation brought about by extensivescal stimulus and loose monetary policy is noteective in our view. Furthermore, we are notworried about infation in the current environ-ment. As detailed in our 2010 Outlook, there is

signicant excess capacity globally, rom labor toindustry to most services. Infation, as measuredby the Headline Consumer Price Index, has been1.8% over the last year; consensus or the next12 months is 2.0%. In our view, a muted infa-tion level o 2-3% will not drive up oil prices inthe next year or two.

Turning to supply and demand or oil overthe next 12-24 months, it seems that most othe data points to stable or lower prices in thenear term. Take inventory levels; at 1.06 billion

barrels, commercial inventories in the US(which exclude the 726 million barrels stored inthe Strategic Petroleum Reserves) are above thehigh end o the range seen in the last ve years,as shown in Exhibit 16. US distillates, includingdiesel and heating oil, are 26% above theirve-year average (Exhibit 17).

Looking globally, ocial Organisation orEconomic Co-operation and Development(OECD) inventories stand at a strong 2.73billion barrels, as shown in Exhibit 18. Accord-ing to the International Energy Agency (IEA),there is another 150 million or so barrels ocrude oil and products stored in tankers in thehigh seas; you may have noticed a recent articlein The Wall Street Journalabout how the oiltankers used as foating storage o the coasto southeast England have raised concerns abouta possible oil spill that could ruin the areasbeaches!12

At these levels, inventories cover 59.4 days o

OECD consumption (reliable inventory data onall emerging countries is not readily available).

Exhibit 16: US Commercial Petroleum Inventories

800

850

900

950

1,000

1,050

1,100

1,150

MillionBarrels

5-yr range (04-08)

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov

2008 2009 5-yr Average

Data through December 18, 2009

Source: Energy Inormation Administration, Investment Strategy Group

Exhibit 18: OECD Commercial Petroleum Inventories

MillionBarrels

5-yr range (04-08)

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

2008 2009 5-yr Average

2,300

2,400

2,500

2,600

2,700

2,800

Data through October 31, 2009

Source: International Energy Agency, Investment Strategy Group

Exhibit 17: US Inventories Relative to Their 5-Year

Averages

100=

5-

YrAverage

1/2/09 3/2/09 5/2/09 7/2/09 9/2/09 11/2/09

80

90

100

110

120

130

140Distil lates Gasoline Total petroleumCrude

Data through December 18, 2009

Source: Energy Inormation Administration, Investment Strategy Group


18/32

18 Goldman Sachs

january 2010

As can be seen in Exhibit 19, this coverage ratiois substantially above the 20-year low o 49.3days and not ar rom the 20-year high o 64.6in 1990. Such high levels o inventory coincide

with lower levels o demand. In the US, totalpetroleum product demand is 8.9% below itsve-year average.

With such high levels o inventory as abackdrop, small supply and demand imbalancesbecome a little less critical. And as is typical inthe oil industry, there is a wide range o esti-mates on supply, demand and excess capacity. Itis clear that the rebound in global growth willreverse the drop in demand we saw in 2009. IEAestimates that oil demand will increase by about

1.5 million barrels per day (b/d) (Exhibit 20).At the low end o the range, OPEC estimates anincrease o about 800 thousand b/d, while ourcolleagues at Goldman Sachs Global Econom-ics, Commodities, and Strategy (ECS) Researchestimate 2.2 million b/d. The average o eighthighly-regarded orecasters is about 1.3 millionb/d. This incremental demand can be met romboth OPEC and non-OPEC sources.

On the supply side, the IEA estimates anincrease in non-OPEC supply o about 400,000

b/d and an increase in OPEC Natural GasLiquids (NGLs) o about 900,000 b/d. Thecall on OPEC supply is estimated to increaseby about 230,000 b/d. In the most optimisticscenario, FACTS Global Energy estimates supplyexceeding demand by about 1.5 million b/d; ina more pessimistic scenario, our colleagues atGoldman Sachss ECS Research expect a short-age o about 580,000 b/d. We should note that inspite o repeated warnings o depleting oil wells,non-OPEC production has actually surprised tothe upside in 2009, as shown in Exhibit 21, byabout 990,000 b/d.

It seems to us that with a combination ohigh inventory levels, increased production innon-OPEC countries, marginal increases inOPEC crude production and increases in OPECNGL production, prices do not have muchurther upside rom current levels, barring amajor geopolitical event. Even i economicgrowth exceeds our expectations and drives up

consumption, OPEC spare capacity is quitesignicant. As shown in Exhibit 22, year-end2009 spare capacity is estimated at 5.5 million

Exhibit 19: Industry Inventories Days o Forward Cover

DaysofDemand

89 91 93 95 97 99 01 03 05 07 09

45

47

49

51

53

55

57

59

61

63

65

Data as o December 11, 2009




Million b/d 2008 2009 (e) 2010 (e)

Exhibit 20: IEA Supply and Demand Balance 2008-2010

Demand

Total OECD 47.6 45.5 45.5

US 19.5 18.8 18.9

Europe 15.3 14.6 14.6

Japan 4.8 4.3 4.1Other 8.0 7.8 7.9

Total non-OECD 38.7 39.3 40.8

China 7.9 8.3 8.7

Middle East 7.1 7.3 7.6

Other 23.7 23.8 24.5

Total Demand 86.2 84.9 86.3

Supply

OPEC NGLs 4.5 4.9 5.7

Non-OPEC 50.7 51.3 51.7


19/32


b/d, which is 3.7 million b/d above the low levelsseen in 2004 and 2005. This spare capacity isexpected to increase by up to 1 million b/d in2010 and excludes any growth in Iraqs produc-tion capacity. While Iraqs uture productioncapacity is highly uncertain given the structuraland political problems the country aces, it isreasonable to assume that the increase in produc-

tion will be over 2-3 million b/d in the next ewyears. In the last year, Iraq has signed deals withseveral major oil companies including ExxonMobil, BP, Royal Dutch Shell, China NationalPetroleum Corporation and Eni with incentivesto increase production quickly. In act, somehave estimated an increase in productioncapacity by as much as 10 million b/d by 2020.

For the near term, we estimate prices willrange between $60-85 per barrel. The low endo the range among major contributors to theBloomberg consensus survey or year-end 2010 is$60. Goldman Sachs research is at the high endat $95 per barrel. Some OPEC members haveindicated a preerence or stable prices in the$70-80 range; the utures curve or end o 2010is $80 per barrel. With current prices at about$74 per barrel and six-month realized volatilityat 36%, we do not believe that a tactical over-weight to oil is warranted at this time. While werealize that nancial buyers may drive short-

term prices higher, we think that the supply anddemand actors point toward stable or slightlydeclining prices. Furthermore, our view o muted

infation or the next two years does not provideany tailwind and our avorable view o thedollar might actually provide a slight headwindto sustained oil price increases.

Peak Oil and the Long-Term Upside

One o the common themes that pervades mostoil discussions is the impending shortage o

supply to meet the increasing demand romemerging markets. People conjure images ohundreds o millions o cars in China and Indiajuxtaposed with declining production in non-OPEC countries, attacks and kidnappings inNigerias Niger Delta and 24,000-oot deepwells in the oshore Tupi eld in Brazil. Suchdiscussions end with a conclusion that oil pricesare on a signicant upward trajectory. At theextreme, the discussion ultimately leads to peakoil a point in time when the maximum rateo global production is reached, and ater whichproduction declines. In 1956, M. King Hubbertpredicted peak oil production in the US between1965 and 1970 (US peak production was, in act,reached in 1970) and globally by 2000 (thisobviously has not occurred, and we do notexpect it to in the oreseeable uture). Peak oiltheorists generally believe that supply shortageswill be very disruptive to global economies.

We do not believe in peak oil. First, improve-

ments in energy eciency and concerns aboutcarbon emissions have led to a decline in globaloil consumption per capita per year since 1979,

Exhibit 22: OPEC Spare Capacity is HighExhibit 21: Recent non-OPEC Upside Production

Surprises

Millionb/d

Russia US Europe

(North Sea)

Other

0.0

0.2

0.4

0.6

0.8

1.0

1.2

Total

Non-OPEC

+0.34

+0.23+0.30

+0.12

+0.99

Average 2009 production through Q3 2009 vs.

the Jan 2009 IEA forecast for the same period.

Millionb/datYearEnd

98 99 00 01 02 03 04 05 06 07 08 09

6.1 6.0

5.5

1.8

0

1

2

3

4

5

6

7



Source: Bloomberg, Investment Strategy Group


20/32

20 Goldman Sachs

january 2010

as shown in Exhibit 23. Even more amazing tous is the act that the ratio o proven global oilreserves to consumption has held quite steadynear the 40-year level; in other words, in spite

o increasing overall consumption and depletiono reserves o many elds, proven global oilreserves have steadily increased with consump-tion to maintain this reserve-to-consumptionratio, as shown in Exhibit 24.

In addition, technological improvements inexploration and production have increasedrecoverable reserves rom existing elds andmade deep-water elds such as Tupi accessible.In spite o earlier predictions that no majornds were going to occur, these discoveries have

surprised to the upside, as shown in Exhibit 25.For example, in 2008, the US Geological Surveyreported 3.65 billion barrels o undiscoveredtechnically recoverable oil in the Bakken orma-tion in North Dakota. While production in theBakken dates back to 1970, improvements inhorizontal drilling and hydraulic racturing andthe higher prices that make such techniqueseconomical have increased the recoverablereserves substantially. In another example,Chevron has been trying to extend the lie o one

o the worlds oldest oil elds, the Kern River,by using enhanced recovery techniques involvinghigh-technology temperature sensors to monitorproduction.13 It is worth noting, incidentally,that not all the discoveries will require higherproduction costs.

Outside the United States, the potential orsignicant discoveries is also high. As shown inExhibit 26, with the exception o North Americaand Latin America, most o the drilling activity issmall relative to the current estimates o provenworld reserves: The Middle East holds 60% othe worlds proven reserves, yet accounts or only15% o the worlds drilling activity; Arica has10% o the proven reserves, yet accounts or just4% o drilling activity.

In our view, the peak oil theories are alarmistand unwarranted. Ater steady price increasesin the 1970s rom a low o $1.80 per barrel toa high o $44 per barrel (which is equivalent to$109 per barrel in todays dollars), it was

common to hear about production peaks andlong term oil shortages. Some 30 years later, themessage is being repeated. However, we go back

Exhibit 23: Global Oil Consumption Per Capita

Per Year

Exhibit 24: Ratio o Proven Global Oil Reserves to

Annual Consumption

Barrels

perCapita

perYear

0

1

2

3

4

5

6

059789817365

5.4

4.6

GlobalProven

OilReserves(Years)

40.8

25

27

29

31

33

35

37

39

41

43

45

07040198959289868380


Source: BP Statistical Review o World Energy 2009, World Bank, Investment Strategy Group

Data through 2008

Source: BP Statistical Review o World Energy 2009, Investment Strategy Group


21/32


to Economics 101: Higher prices beget highersupply, albeit with some lag and at higher mar-ginal costs. So while we dont believe in oil short-ages in the oreseeable uture, we do think that,on average, marginal costs will be higher, thereby

putting a foor on prices and supporting asustainable increase in prices toward the $80-$100 range over the next several years.

Our expectation o broadly higher pricesrepresents only a modest and gradual increaseover current levels, and it is not signicant in thecontext o oils volatility level (which averagedabout 40% over the past decade). As such, wedont believe that investing in oil utures isan optimal long-term solution at this time.Furthermore, contango will be a persistent dragon returns and short-term downside risk, givenoils volatility, is signicant.

For those who have a strong bullish view,we recommend investing through the publicor private equity market, where an eectiveoperator can add value in the long run irrespec-tive o the direction o prices and provide somedownside protection should oil prices decline tothe low end o our 12-month range (which, asnoted above, is $60-85 per barrel). Exhibit 27

shows how a major oil company perormed wellover an 18-year period where oil prices declinedover 70% in nominal terms.

Exhibit 27: Oil Companies Can Outperorm Even When

Oil Prices Decline

0

500

1000

1500

2000

2500

3000

7/987/967/947/927/907/887/867/847/827/80

WTI Oil Prices

US Energy Equity

US Equity

Major Oil Company

19.0%

17.7%

11.6%

-6.6%CumulativeReturn(Jul80=

$100)


Source: Datastream, Empirical Research, Ibbotson, Investment Strategy Group

Numbers listed are annualized returns

Exhibit 26: Oil Rig Count Versus Proven Reserves by Region

Data through December 2008

Source: Baker Hughes, BP Statistical Review o World Energy 2009, Investment Strategy Group

Field Country Discovered Recoverable Reserves Estimated Production Cost

(Billion Barrels) ($/barrel)

Jubilee Ghana 2007 1.2 $29

Miran West Iraq 2009 1.1-2.5 $30

Jidong Nanpu China 2007 1.7 $35

Blocks 1, 2 & 3 Uganda 2009 1 $35

Santos Basin Brazil $38-50

Vesuvio 2009 1

Guar 2009 1.25

Iara 2008 3-4

Tupi 2007 5-8

Bakken formation US 2008 3.7 $60

Block 31 Angola 2006 1.8 $60-70

Block 32 Angola 2005 1.6 $65

Korchagina & Filanovskogo Russia 2006 2 $85

Kashagan Kazakhstan 2000 10-15 $100-110

Ferdows/Mound/Zagheh Iran 2003 7-9 n/aAzadegan Iran 2003 5-9 n/a

Exhibit 25: Major Oil Finds o the Past 10 Years


Source: Investment Strategy Group, Goldman Sachs Global Economics, Commodities and Strategy Research, company reports

Reserves (YE 2008)

Billion brls % of World

71 6%

123 10%

754 60%

42 3%

62 5%

125 10%

81 6%

1258 100%

Drilling (2008)

Oil Rigs % of World

506 40%

266 21%

194 15%

175 14%

68 5%

45 4%

NA NA

254 100%

North America

Latin America

Middle East

Far East

Europe

Arica

Others

World


22/32

22 Goldman Sachs

january 2010

dollar and real interest rates aect gold prices.I there is a strong relationship, then our viewo the dollar and interest rates might inorm ourview o gold. As many have postulated, gold is

negatively correlated with the dollar and withreal interest rates. Looking at 12-month rollingcorrelations, gold is more negatively correlatedwith the dollar than with real interest rates.Its negative correlation with the dollar is -0.34and the range is quite wide, with a low o -0.89and a high o 0.62. Its negative correlation withreal rates, as measured by the Federal FundsRate minus 12-month CPI, is only -0.11, and therange is also very wide with a low o -0.78 anda high o 0.73.

Analyzing this urther, we look at the rela-tionship o changes in gold prices to changes inboth the dollar and in real rates. Here, we ndthat changes in the dollar explain 26% o thechanges in gold and changes in real rates explain10% o the changes in gold. So a view on thedollar is somewhat but not critically impor-tant to our view on gold and our view on interestrates, while still relevant, is much less important.And neither actor alone or in combination explains more than 36% o the changes in gold

prices. Interestingly, the impact o rates becomesmore signicant when real rates are less than 1%.

On the surace, our avorable view o thedollar and our avorable view o a reasonableeconomic recovery with a modest increase in realrates by late 2010 would imply a negative viewo gold. However, these two actors only explaina little over a third o the changes in gold prices.Furthermore, there are many periods where thecorrelations have been positive, indicating thatgold moved more in line with the dollar andreal rates. From August 1987 to May 1993, orexample, the dollar depreciated 11.1% againsta trade-weighted basket o currencies, but goldalso declined by 17.1%. In act, upon urtherexamination o past monthly data, we note thatabout 36% o the time since 1970, gold and thedollar have moved in the same direction. Sowhile our avorable view on the dollar leads usto be cautious about gold, our expectations ocontinued low real rates have a much smaller

albeit slightly positive impact on our view ongold.

Gold

For thousands o years, gold has held a specialstatus as a perceived store o value and asymbol o wealth. The oldest large stash o gold

was ound in a cemetery in Bulgaria datingback some 6000 years. The oldest coin that hasbeen discovered is rom 2700 years ago, and isan alloy o gold and silver ound in Ephesus inmodern-day Turkey. In many cultures, gold coinsare given to children to mark special holidays,gold rings are exchanged between newlywedsand gits o gold have conerred the highest levelso esteem, aection and appreciation throughouthuman history.

In spite o its long history and special status,

gold is one o the hardest commodities or mostinvestors to evaluate. And or value investorssuch as Warren Buett, the task is nearly impos-sible. In 1998, at a speech at Harvard University,Warren Buett is quoted as saying: It gets dugout o the ground in Arica or someplace. Thenwe melt it down, dig another hole, bury it againand pay people to stand around guarding it.It has no utility. Anyone watching rom Marswould be scratching their head.

At this point in time, we eel like Martians;

to quote the late economist Peter Bernstein,Nothing is as useless and useul all at the sametime.14 Nevertheless, gold has rallied 56% romits recent low in November 2008 and 333%since its trough in August 1999 to its currentprice o about $1100/toz. There are our plau-sible explanations or the increase: 1) concernsabout the continued depreciation o the dollar;2) ears o high infation due to massive liquidityas well as governments alleged desire to createinfation as a means o reducing the debt burden;3) expected increases in demand rom investorsto capture potential incremental returns and,importantly, as an insurance policy against asignicant double-dip recession, in which goldbecomes the currency o last resort; and 4) anexpected rise in demand rom central banks(especially emerging market central banks) inan eort to diversiy their reserves away romdollar-denominated assets.

Our rst task is to ormulate some metrics or

gold so that our asset allocation recommenda-tions can be grounded in a sound ramework.We begin by examining the extent to which the


23/32


Another explanation or the rally in gold hasbeen infation expectations. As discussed above,the combination o a high expected debt/GDPratio o over 70%, a Federal Reserve balance

sheet o $2.2 trillion as o year-end 2009 and anear-zero interest rate policy has prompted somemarket observers to state that high infation isinevitable. We do not share that view. We expecta muted level o infation o about 2-3% ratherthan the levels o infation seen in the 1970s thatdrove gold prices up to about $2400 in 2009nominal dollars.

Irrespective o the outlook or infation, itis important to recognize the limited hedgingbenets o gold. Both the monthly and quarterly

correlation o gold with infation are negligibleat 0.04 and 0.09 respectively. It is only i we lookat annual data that we see a positive correlationo 0.30 but even then, the correlation fuctuatessharply on a rolling 10-year basis. As shown inExhibit 28, gold keeps pace with infation overthe very long run, but not over short to interme-diate periods. So even i infation does exceed2-3%, it is not obvious to us that gold will be thebest hedge to own.

So we turn to the supply and demand actors.

Lets review some o the basic supply anddemand data about gold. Most importantly,unlike oil and other consumable commodities,the stock o gold above ground does notdecrease, so when we talk about supply, we needto talk about total stock, total annual produc-tion and total scrap gold comprised o items(mostly jewelry) that can be recycled to meetnew demand. To quote Peter Bernstein, Unlikeany other element on earth, almost all the goldever mined is still around, much o it now inmuseums bedecking statues o the ancient godssome on the pages o illustrated manuscripts,some in gleaming bars buried in the darkcellars o central banks, a lot o it on ngers, earsand teeth. There is a residue that rests quietly inshipwrecks at the bottom o the sea.15

As shown in Exhibit 29, the current physicalstock o gold is about 163,000 tons, with thevast majority held in the orm o jewelry. Thisstock increases with annual production. So

rom 2007 to 2008, new production rom minestotaled 2,450 tons and was added to the prioryear-end stock o 160,550 tons. Interestingly,

Exhibit 28: An Imperect, Inconsistent Hedge

Exhibit 29: Physical Stock o Gold and Net Increase

RealGold

Prices

(1871=

1.0

0)

1871 1886 1901 1916 1931 1946 1961 1976 1991 2006

6

5

4

3

2

1

0

0%

20%

40%

60%

80%

100%

Net Increase from

Production

+2450 tons

(~1.5% of total stock)

(~160,550 Tons) (~163,000 Tons)

Beginning Gold Stock (2007) Ending Gold Stock (2008)

11.9%

3.8%

16.2%

16.5%

51.7%

12.0%

3.9%

16.4%

16.0%

51.6%

OtherIndustrial Fabrication

Private Investment

Official Holdings

Jewelry

Data through November 2009

Source: World Gold Council, Investment Strategy Group

Historically, gold has been an eective hedge against rising prices,

but only over the very long term. The metal oten ails to keep up

with inlation as these multi-decade periods indicate.



24/32

24 Goldman Sachs

january 2010

additional supply to meet the demand or goldcame rom recycled gold. In 2008, the last yearor which ull-year data is available, scrap goldamounted to 1,050 tons o old rings, chains and

bracelets. The 2009 scrap tonnage is probablyhigher. A December 2009 article in The WallStreet Journaldiscussed how gold parties, inwhich newly ormed businesses convene groupso neighbors who want to sell their gold or cash,have become the new Tupperware parties insuburban America.16 A pattern o increasedrecycling o gold jewelry also occurred in thelate 1970s and early 1980s. On Jan. 12, 1980,The New York Times reported similarselling where people would line up or hours

to sell their gold and silver jewelry. Just over aweek later, on Jan. 21, gold prices peakedat $850/toz.

As we look at the supply and demand picture,the two most important actors and biggestuncertainties in the near term are investordemand and central bank activity. The biggestand most notable change on the demand sidewas the increase in investment demand startingin the second hal o 2008. As shown in Exhibit30, investment demand increased rom 17% in

2007 to an estimated 35% in 2009 while jewelrydemand, which tends to be more price-elastic(especially in India, the largest consumer o goldjewelry), dropped by 17%.

As we discussed above, the demand or goldETFs, gold bullion, gold coins and gold stocksis reaching euphoric proportions and sucheuphoria can continue or a long time. Fears opolicy mistakes and sovereign credit risk, as wellas doubts about the ability o governments tonavigate the current environment, may last ora while. But by the same token, an improvingglobal economy, better data out o the UnitedStates with respect to growth, employment andhousing, and a bipartisan eort to address thelong-term budget decit in the US might allaythose ears as early as mid-2010. In such a case,the investment demand or gold might disappearquite suddenly, with no natural buyer o thatvolume o gold at current prices. As we men-tioned earlier, physical gold-backed ETFs now

hold about 1,495 tons o gold, which is largerthan the gold reserve holdings o China and oSwitzerland.

Exhibit 30: Gold Investment Demand

Exhibit 31: Annual Ofcial Sector Net Sales

Exhibit 32: Central Bank Gold Reserves: Developed

Markets Versus Emerging Markets

$

/TroyOunce

Investment Demand Share (LHS)

Average Gold Price (RHS)

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

09e01938577

1980: 45%

0

200

400

600

800

1000

1200

Thousand

Tons

$

/TroyOunce

-400

-200

0

200

400

600

800

1000

1200

1400

1600

09e98877665

Official Sector Net Sales (LHS)

Average Gold Price (RHS)

0

100

200

300

400

500

600

700

800

900

1000

Thousand

Tons

29.4

4.0

25.2

4.4

0

5

10

15

20

25

30

35

09e0807060504030201009998

Developed Markets and Institutions

Emerging Markets


Source: CPM Gold Yearbook 2009, World Gold Council, Bloomberg, Investment Strategy Group


Source: CPM Gold Yearbook 2009, World Gold Council, Bloomberg, Investment Strategy Group




25/32


and lasting impact on gold prices.One actor that probably put some upward

pressure on gold prices over the last several yearshas been the de-hedging activities o produc-ers. Since 2000, producers have reduced theirhedges o uture production by purchasing backgold utures. As one can see in Exhibit 33, theremoval o these hedges had contributed to asmuch as 450 tons o additional demand peryear in some years, compared to net supply o asimilar magnitude in years like 1999. From thedata shown in Exhibit 34, it seems that most othe de-hedging is over. To the extent that this

de-hedging provided some upward pressure onprices, its removal rom the market is no longera positive or gold prices.

But or now, we can readily see that investorear is likely to lead to continued investmentdemand, which will in turn drive pricessignicantly higher. Some analysts have targets

o $1350 based on anticipated continuedinvestor fows. Some have speculated that pricescan reach the infation-adjusted levels seen in

January 1980, which in todays dollars is equiva-lent to $2400. Others have said that i we revertto the pre-1971 gold standard without a dollardevaluation, gold would have to be priced atover $6,400, given the size o the US Treasurysgold reserves and the stock o US dollars asmeasured by M1. These price projections remindus o oil targets o $200 to $400 two years ago.

Finally, we should examine the impact ocentral bank activity on gold demand. Overthe last two years, there have been severalannouncements about a handul o central banksadding physical gold to their reserves. At the topo the list is China, which announced that it hadincreased its reserves by about 450 tons overthe last ve years. India bought 200 tons romthe International Monetary Fund in November2009. On a smaller scale, Russia added 70tons to its reserves; the Philippines, 22 tons; Sri

Lanka, 10 tons; and Kazakhstan, Mexico andBelarus each added 2.5 tons in 2008 and 2009.

While the emerging market countries havebeen adding to their gold reserves recently, theoverall ocial sector, which includes centralbanks and multinational organizations, hasbeen a net seller o gold over the last decade.As shown in Exhibit 31, however, the pace osales has been declining and is estimated to befat to negative at the end o 2009 when naldata is released.

A great deal o attention is ocused onwhether central banks o emerging markets willcontinue to make meaningul additions to theirreserves. While their actions receive a lot oattention, it is important to keep the size o mosto these purchases in perspective. As Exhibit 32illustrates, the overall amounts in emergingmarket central banks are not signicant relativeto those o the developed market central banks.So unless there is a signicant shit away rom

the dollar into gold as a better store o value orthe larger central banks, we do not anticipatecentral bank activity to have an overwhelming

Exhibit 33: Net Supply/Demand rom Producer

Hedging Activities

Tons

-600

-400

-200

0

200

400

600

09e07050301999795939189

Net Demand from

De-Hedging

Net Supply from

Hedging

445350

190



Tons

0

500

1000

1500

2000

2500

3000

3500

09e07050301999795939189

Outstanding book

forecast at only 300t at

year end

Exhibit 34: Outstanding Producer Hedge Positions




26/32

26 Goldman Sachs

january 2010

equity exposure than commodity exposure gold companies being a notable exception.

Third, or those who eel a general concernabout the state o the world and eel that at

currencies across the globe have lost and willcontinue to lose their value, then physical gold,with all its attendant insurance and storage costs,might seem suitable. In a global depression, goldmay well turn out to be the best insurance policy but that is not a tested hypothesis. In act,during the recent nancial crisis and deeprecession, the dollar rallied 24.3% as a saehaven while gold actually declined over 30%rom its peak in March 2008 through its troughin November 2008. As economist Nouriel

Roubini pointed out in his Dec. 11, 2009 piece,The New Bubble in the Barbarous Relic That isGold, in a depression or near-depression, onewould be better o stockpiling canned ood andother commodities like oil that areuseul or riding out Armageddon.17

There is no doubt that gold has specialqualities. It is chemically inert so it maintainsits radiance. It has unusual density so smallamounts can unction as money or largedenominations. It is very sot and malleable,

enabling it to be used or gilding and or illus-trated manuscripts. And most importantly, it isimperishable. So one can understand the valuebeing attributed to gold beyond its immediateuses. As Peter Bernstein said, Gold refects theuniversal quest or eternal lie the ultimatecertainty and escape rom risk.18 And at timeso crisis, gold might provide an excellent escaperom risk. But the key question is whether ourclients diversied portolios need gold as anescape rom the risk that currently remains.We do not believe so.

We recommend that clients who are consid-ering an investment in gold begin by clariyingtheir objectives. First, is investing in gold anexpression o a bearish view on the dollar?

In such a case, while we are not bearish thedollar and, in act, have a avorable view o it,we would recommend diversiying some o onesequity exposure away rom US assets. Thatseems like a more eective and less risky hedgeagainst dollar depreciation.

Second, or those who are looking at goldpurely as an investment opportunity with upsidepotential, then we recommend two strategies:either invest in gold with some downside pro-tection through structured notes, or consider

gold-related equity exposure through public andprivate equity where excellent operators canbetter manage the downside. While the upsidepotential o owning gold bullion or gold ETFsis tantalizing, the downside is also signicant.Ater peaking on Jan. 21, 1980 at $850, pricesdropped by 17% on Jan. 22. By March 18, inless than two months, prices had dropped a totalo 43%.

Our concern with direct gold exposure atcurrent prices is that at the earliest signs o a

broad reassessment o the global economicoutlook, especially that o the US, prices candrop quite precipitously. A structured note canprovide some upside but limit the downside.Similarly, investing in gold mining companiesprovides some upside but limits the downsiderom signicant gold price drops, because aneective mine operator actively manages thegold exposure. As Barrick Golds ounder andchairman wrote to shareholders in the 2008Barrick Annual Review: While a $1000 invest-ment in Barrick at our ounding in 1983 wouldtoday be worth $39,000, that same investment ingold bullion would today be worth only around$2,000. We should note that in the InvestmentStrategy Group, we have no view on individualcompanies; we just use Barrick as an exampleo a company that can manage the upside anddownside o gold in the long run more eectivelythan a typical investor owning the commoditydirectly. But remember, the exposure o com-

modity-related stocks to commodities varies; asnoted earlier, when we decomposed the returnso commodity-related equities, most have more


27/32


to produce subsistence or man, implying thatpopulation growth would exceed agriculturaland livestock commodities and amine wasinevitable.19 In the mid-1800s, concern about

shortages in whale oil the prime lubricant andilluminant in the US at the time drove pricesup 143%. (Within a ew years thereater, whaleoil was replaced by petroleum.) In 1968, PaulEhrlich wrote that India couldnt possibly eedtwo hundred million more people by 1980and hundreds o millions o people will starve todeath.20 As a result o Nobel Laureate NormanBorlaugs work that improved yields in rice andwheat, the Green Revolution spread across Asiaand Arica, improving the lives o billions o

people.In an interesting book, The Doomsday Myth:

10,000 Years of Economic Crises,proessorsCharles Maurice and Charles W. Smithsonsurveyed 10 historical resource shortages andshowed how market orces alleviated the crisesthrough innovation and technology: New materi-als are discovered, substitutions are made, yields be they in agricultural commodities, humanlabor, or oil wells are all improved and crisesare alleviated.21 As our clients read about those

asset management rms that now suggestinvestors ownphysicalcommodities such aslivestock, corn and soybeans they may want toremind themselves o all the ood subsidies thatdierent governments around the world provideto sustain their armers, and determine howthey can make a reasonable net return ater thecosts o managing the physical assets includingstorage, transportation, insurance and manage-ment ees.

Conclusion

Commodities are clearly essential to our lives.We need agricultural commodities to eat. We

need energy commodities to heat our homes andto drive our cars. We need industrial commoditiesto build our bridges and our oces. Andwe need precious commodities or jewelry,electronics and solar panels. But being essentialto our lives does not make them essential in ourdiversied portolios.

Upon extensive analysis o the risk and returncharacteristics o commodities, we nd that along-only exposure to utures-based commod-ity indexes in a well-diversied portolio is not

necessary. A typical moderate-risk portolio withglobal public equities already has some exposurethrough commodity-related companies, sincecommodity-related equities represent 19.5%o developed equity benchmarks and 29.3%o emerging market equity benchmarks. Werecommend some incremental exposure throughcommodity-related hedge unds including macro-t

25035641 24986279 Goldman Sachs Reseach Commodities Gold Oil

Documents