Dietrich Domanski Alexandra Heath

+41 61 280 8353 +41 61 280 8514

[email protected] [email protected]

Financial investors and commodity markets1

Commodities have attracted considerable interest as a financial investment in recent

years. This article discusses the factors behind their growing appeal and assesses the
extent to which market characteristics, such as price volatility, have changed as a
result. The feature concludes that commodity markets have become more like financial
markets in terms of the motivations and strategies of participants, but that the physical
characteristics of commodity markets are still important.

JEL classification: G11, G15, Q41.

The sharp increase in commodity prices, especially for energy and base metals
since 2002, has gone hand in hand with growing derivatives market activity
(Graph 1). The number of contracts outstanding in exchange-traded commodity
derivatives almost tripled from 2002 to 2005. Over-the-counter (OTC) trading of
commodity derivatives also grew rapidly. According to BIS statistics, the
notional value of OTC commodity derivatives contracts outstanding reached
$6.4 trillion in mid-2006, about 14 times the value in 1998 (BIS (2006)). At the
same time, the share of commodities in overall OTC derivatives trading grew
from 0.5% to 1.7%.
Along with the rapid increase in commodity derivatives trading, the
presence of financial investors in commodity markets has grown rapidly over
the past few years. While commodity market investment is still small relative to
overall managed funds, it is large relative to commodity production. In addition,
there are indications that the types of financial investors and the strategies they
employ have changed.
These developments raise the question of whether growing investor
presence has altered the character of markets that are of key importance for
the global economy. Understanding the nature of the changes in investor types
and strategies is an important step in this regard. The first part of this article
documents the increasing role of financial investors in commodity markets,
while the second presents some evidence about changes in the motivations of
market participants. The third section looks at the effect these changes may
have had on the dynamics of commodity prices. The feature concludes that

1
The views expressed in this article are those of the authors and do not necessarily reflect
those of the BIS. We are grateful to Anna Cobau and Emir Emiray for excellent research
assistance.

BIS Quarterly Review, March 2007 53

Electronic copy available at: https://ssrn.com/abstract=1600058

 Commodity prices and derivatives activity
Prices1 Derivatives activity
Energy Exchange-traded (rhs)²
340 500 30
Precious metals OTC (lhs)³
Base metals
Agriculture 280 400 24

220 300 18

160 200 12

100 100 6

40 0 0
Jan 98 Jan 00 Jan 02 Jan 04 Jan 06 Jun 98 Jun 00 Jun 02 Jun 04 Jun 06
1
Goldman Sachs Commodity Index (GSCI) sub-indices, monthly averages; 1998–2002 average = 100.
2
Number of contracts outstanding, in millions. 3 Notional amounts deflated by the GSCI, June 1998 =
100.

Sources: Datastream; BIS. Graph 1

while physical characteristics, such as inventory levels and the marginal cost of
production, remain important, commodity markets have become more like
financial markets in terms of the motivations and strategies of participants.

The presence of financial investors in commodity markets

Financial activity in commodity markets is large compared with the size of Financial activity is
large relative to
physical production and has grown much faster in recent years. For gold, physical markets
copper and aluminium, the volume of exchange-traded derivatives was around
30 times larger than physical production in 2005 – a significant increase in this
ratio from 2002 (Table 1). The much lower ratio for crude oil may understate
the relative size of financial activity, given that OTC markets are particularly
important for this commodity. Bank of England market contacts suggest that up
to 90% of swaps and options trading in oil is done over the counter, reflecting
the need for tailored contracts and a lack of organised derivatives markets for
certain types of crude oil (Campbell et al (2006)).
Traditionally, specialised financial traders in commodity markets focused Traditional
arbitrage ...
on exploiting arbitrage opportunities (Kolb (1997)). Typically, such
opportunities arise as the consequence of commercial investors seeking to
hedge their production or consumption in futures markets. These arbitrage
trades, usually conducted by specialised commodity traders, typically involve
taking long or short positions in forward markets for specific commodities and
offsetting positions in spot markets. In doing so, financial investors provide
liquidity in commodity derivatives markets.
Normally in financial markets, opportunities for (risk-free) arbitrage exist ... limited by
constraints on short
when the futures price deviates from the relevant spot price plus the cost of
selling
carry, eg the cost of financing a position in the spot market. However, the
scope for arbitrage in commodity markets may be limited by constraints on
short selling. In particular, the stock of commodities available for lending is

54 BIS Quarterly Review, March 2007

Electronic copy available at: https://ssrn.com/abstract=1600058

 generally small for energy and base metals. This limitation allows the futures
price to fall below the spot price – a situation known as backwardation (Duffie
(1989)).
Passive investment The current upturn in commodity prices has been accompanied by greater
strategies ...
variety in the types of financial investors and investment strategies in
commodity markets (Holmes (2006)). One rapidly growing area is passively
managed investment and portfolio products, which is consistent with investors
now viewing commodities as an attractive separate asset class. By mid-2006,
around $85 billion of funds were tracking the Goldman Sachs Commodity Index
(GSCI) and the Dow Jones/AIG Index, two important commodity indices
(Holmes (2006)).
... can provide
Passively managed investments often pursue a fully collateralised long-
diversification only futures strategy. This can be attractive to institutional investors with a
benefits
longer-term investment horizon, such as pension funds, for several reasons
(Beenen (2005)). First, this strategy allows diversification into commodities at a
relatively low cost. Historically, commodity prices have had a relatively low
correlation with prices in other asset classes and a high correlation with

Indicators of financial and physical activity in selected commodity markets in 2005

Financial activity World production2 Ratio3
Futures Options
1
Volume % chg Volume1 % chg
since since
2002 2002 2002 2005 2002 2005

Crude oil 93.0 34.4 14.8 27.2 67.0 73.6 3.2 3.9
Of which: NYMEX 59.7 30.6 14.7 28.5
ICE 30.4 41.5 0.0 –69.7
Gold 34.5 16.8 2.9 49.7 2.6 2.5 21.8 32.0
Of which: TOCOM 18.0 –12.4 0.3 .
COMEX 15.9 76.2 2.9 48.3
Aluminium 33.3 25.2 4.1 368.3 26.1 23.0 22.7 27.3

Of which: LME 30.4 36.3 4.1 368.3

SME 2.1 –9.0 . .
Copper 35.5 41.1 2.2 140.0 15.3 16.5 30.5 36.1

Of which: LME 19.2 16.0 2.1 134.5

SME 12.4 113.1 . .

Note: NYMEX = New York Mercantile Exchange; ICE = IntercontinentalExchange, United Kingdom; TOCOM = Tokyo Commodity
Exchange; LME = London Metal Exchange; SME = Shanghai Metal Exchange.
1
Number of contracts, in millions. 2 Oil: millions of barrels per day; gold: millions of kilograms; aluminium and copper: millions of
tonnes. 3 Defined as financial activity in the two largest contracts converted to units of physical production, divided by production.

Sources: Commodity Research Bureau, The CRB Commodity Yearbook; Energy Information Agency, Annual Energy Review; GFMS;
US Geological Survey. Table 1

BIS Quarterly Review, March 2007 55

Electronic copy available at: https://ssrn.com/abstract=1600058

 Crude oil prices and roll returns
3 30
Backwardation
2 20

1 10

0 0

–1 –10
Contango
–2 –20

–3 Spot price minus three-month futures price (lhs)¹ –30

Roll return (rhs)²
–4 –40
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
1
In US dollars per barrel. 2 Annual returns from rolling over consecutive three-month futures at maturity
in excess of spot price returns.

Sources: Bloomberg; BIS calculations. Graph 2

inflation (Gorton and Rouwenhorst (2004)). 2 Second, these authors also

provide evidence that, historically, the return on a diversified basket of long
commodity futures has been comparable with the return on other asset classes
with similar risk features, such as equities.
Several authors have emphasised the importance of the so-called roll Positive roll returns
have been
return from a long position in commodity futures as a component of total
important …
returns (Erb and Harvey (2005), Feldman and Till (2006)). Indeed, roll returns
are an important explanation for why the average return on commodity futures
has exceeded the average return from holding spot commodities (Gorton and
Rouwenhorst (2004)). Investors earn a positive roll return if they can roll over a
futures contract that is close to expiry into a new contract at a lower price. This
occurs when the spot price (to which the price of the original futures contract
converges over time) is higher than the price of the new futures contract, ie in a
backwardated market.
Roll returns can be considerable. For example, in the crude oil market, the ... but depend on
the persistence of
roll yield from purchasing three-month futures was about 14% per annum over
backwardation
2003–04 (Graph 2). However, roll returns became negative when the price of
the futures contract rose above the spot price, ie the market moved into
contango, in 2005. Essentially, the profitability of strategies aimed at
generating positive roll returns depends on the persistence of the factors that
cause markets to backwardate, including low levels of commodity stocks
available for short selling and positive returns received by owners from holding
the physical commodity (the so-called convenience yield).
Growing presence
The presence of investors with a shorter-term focus, such as hedge funds, of investors with
has increased considerably during the past three years. The number of hedge shorter-term focus

2
It is important to note that these calculations are all in US dollars and therefore the correlation
between commodity prices and exchange rate movements is not a consideration. To the
extent that commodity prices are in US dollars and other assets in the portfolio under
consideration are not, currency hedging may be important for obtaining diversification
benefits.

56 BIS Quarterly Review, March 2007

Electronic copy available at: https://ssrn.com/abstract=1600058

 funds active in energy markets has reportedly tripled to more than 500 since
the end of 2004, with an estimated $60 billion in assets under management
(Fusaro and Vasey (2006)). The $6 billion loss on natural gas derivatives that
the hedge fund Amaranth reportedly incurred in September 2006 is a further
indication of the size of positions that hedge funds take in commodity markets.
Partly as a result of increased demand from financial investors following
shorter-term strategies, the number of exchange-traded funds (ETF) for
commodities has increased since the first ETF for gold was opened in 2003. A
related area of growth is the development of instruments that facilitate the
implementation of more complex strategies, including cross-market arbitrage or
taking positions on volatility. A specific example is the rapid expansion in
structured commodity notes (McNee (2006)).
CFTC data ... An important source of quantitative information on trading activities in
commodity markets is the Commodity Futures Trading Commission (CFTC),
which publishes weekly data on the open positions in US futures markets of
commercial and non-commercial traders (Graph 3). The non-commercial trader

Total open interest and shares of non-commercial holdings

Futures and options combined; six-month moving averages

Total commodities Crude oil

Open interest (lhs)1
16,000 24 2,400 Long (rhs)2, 3 16
Short (rhs)2, 4

12,000 18 1,800 12

8,000 12 1,200 8

4,000 6 600 4

0 0 0 0
95 97 99 01 03 05 95 97 99 01 03 05

Gold Copper

6,000 60 120 60

4,500 45 90 45

3,000 30 60 30

1,500 15 30 15

0 0 0 0
95 97 99 01 03 05 95 97 99 01 03 05
1
Number of contracts, in thousands. 2 Ratio to total open interest, in per cent. 3
Non-commercial long
positions. 4 Non-commercial short positions.

Source: CFTC. Graph 3

BIS Quarterly Review, March 2007 57

Electronic copy available at: https://ssrn.com/abstract=1600058

 Activity of managed money traders in selected commodity markets
Market Number of MMTs holding MMT open interest as % of total
positions1 open interest2
19943 2003–44 19943 2003–44

Crude oil Average 40 80 Long 6.4 14.0

Maximum 48 100 Short 2.2 6.9
Natural gas Average 33 66 Long 2.3 11.9
Maximum 44 81 Short 7.0 15.4
1 2 3
Daily averages and maximums. In futures and options markets. April–September
4
1994. August 2003–August 2004.

Sources: CFTC (1996); Haigh et al (2005). Table 2

group includes participants who are not primarily using the market for hedging,
and encompasses a variety of subgroups. In 2003–04, the non-commercial
trading category for both natural gas and oil was dominated by managed
money traders (MMTs) (Haigh et al (2005)). This group includes specialised
investors such as commodity pool operators and funds advised or operated by
commodity trading advisers. Hence, it is likely to capture most financial
investors who are operating in centralised commodity markets.
The importance of MMTs seems to have grown significantly since 1994. ... confirm growing
importance of
Data available for the crude oil and natural gas markets show that the average
financial investors
number of MMTs trading has roughly doubled and their share of total open
interest in each of these markets has increased sharply (Table 2). In addition,
assets under management by commodity trading advisers are significant and
rose from about $20 billion in 2002 to about $75 billion by end-2005 (IMF
(2006)).
The share of non-commercial traders in aggregate has gone up from Share of financial
traders varies
about 17% in the second half of the 1990s to about 25% in the past three
across markets
years. This increase is mainly attributable to an upward trend in the share of
long positions held by non-commercial investors. While this broad pattern holds
across markets, the share of non-commercial positions varies considerably.
Since spring 2006, the share of open interest attributed to non-commercial
traders has fallen by almost 3 percentage points. This is consistent with a
withdrawal of investors during the period of falling commodity prices since May
last year, but also with an increase in the hedging activity of commercial
producers (JPMorgan Chase (2007)).
As regards OTC commodity derivatives markets, the available evidence Limited information
on OTC markets
also supports the notion of a rapidly growing presence of financial investors.

Participants in OTC trading on the ICE

OTC participants’ trading (as % of total
commissions) 2003 2004 2005

Commercial companies 64.1 56.5 48.8

Banks and financial institutions 31.3 22.4 20.5
Hedge funds, locals and proprietary
trading shops 4.6 21.1 30.7

Source: ICE (2006). Table 3

58 BIS Quarterly Review, March 2007

Electronic copy available at: https://ssrn.com/abstract=1600058

 IntercontinentalExchange (ICE) reports that hedge funds, locals and proprietary
trading shops accounted for almost one third of trading commissions paid on
OTC transactions conducted through ICE in 2005, compared to less than 5% in
2003 (Table 3). However, this increase might in part reflect the higher
propensity of institutional investors, in particular hedge funds, to use electronic
trading platforms (Davidson (2006)). It may therefore overstate the increase in
financial investor participation in commodity markets as a whole.

An empirical examination of investor activity

Empirical approach To obtain a general sense of the changes in the motivations underlying
investment activity, we next estimate the relationship between the activity of
financial investors and possible motivating determinants. The results of this
simple, illustrative exercise are broadly consistent with the view that the
motivations for investing in commodity markets have changed along with the
growing presence of financial investors. Given data limitations, this exercise is
constrained to using CFTC data on non-commercial open interest in US
exchange-traded commodity markets. The dependent variable is defined as the
share of net long open interest of non-commercial traders in four somewhat
heterogeneous commodity markets that have experienced particularly large
price movements since 2002: crude oil, natural gas, gold and copper. 3
Explanatory To capture the effect of expected returns on the share of non-commercial
variables
traders, we include the percentage changes in spot commodity prices and a
variable capturing the size of the roll return over the previous 12 months. 4 The
standard deviation of monthly percentage changes in three-month futures
prices is included to capture any response there may be to volatility in returns.
A priori, the effect of such volatility on the position-taking of financial investors
is ambiguous. On the one hand, rising volatility may discourage position-taking
because it lowers risk-adjusted returns, all else equal, particularly for strategies
such as carry trades. On the other hand, volatility is likely to attract more
activity if traders are actively taking exposure to it. Another shorter-run return
consideration may be the opportunity cost of investing in commodities. To
account for this, a world short-term interest rate has also been included. The
longer-term demand for commodities arising from their diversification
properties is proxied in two ways: by the correlation between percentage
changes in commodity prices and a measure of world equity prices over the

3
Net long positions of non-commercial traders are frequently used as a variable to capture
financial investor activity in commodity markets; see eg IMF (2006) and Micu (2005). By
defining the dependent variable as a share, factors that increase net long positions for
commercial and non-commercial traders have been controlled for. However, the dependent
variable cannot distinguish an increase in non-commercial net long open interest arising from
factors that have increased financial activity across all financial markets from an increase
arising from a portfolio shift towards commodity markets as a whole, or portfolio shifts
between individual commodity markets. These issues serve as qualifications to the
interpretation of the estimates.

4
This variable is defined as the difference between the spot price and the three-month futures
price, normalised by the spot price, averaged over the previous 12 months. To the extent that
roll returns encourage investor activity, the estimated coefficient on this variable should be
positive. All explanatory variables are included with a lag of one month.

BIS Quarterly Review, March 2007 59

Electronic copy available at: https://ssrn.com/abstract=1600058

previous five years; and by inflation expectations, defined as the difference
between nominal and real bonds.
Two broad observations can be made by comparing the results of Shorter-term factors
seem to have
estimating this model for the period 1998–2001 with those for the period
become more
2002–06 (Table 4). First, shorter-term factors reflecting return considerations important ...
appear to have become, on balance, more important over time. Past increases
in spot prices have a significant positive effect on the share of non-commercial
net long positions across both periods, as expected. Higher roll returns have a
more positive effect on the share of non-commercial net long positions in the
second period than in the first in the natural gas and oil markets, which have
been backwardated for considerable periods since 1998, as well as in the
copper market, although the estimated coefficient is not significant. 5 The
volatility of futures returns has a negative effect across markets in the second
period, which is particularly significant in the copper market. This pattern is
consistent with a growing importance of leveraged investors speculating on
short-term price trends, as this group is particularly sensitive to short-term
price fluctuations.

Regression results1
Dependent variable: non-commercial long minus short positions, as a share of total open interest
Return2 Roll3 Volatility4 Interest5 Correlation6 Inflation7 Adjusted
Expected sign R2
+ + – – – +
1998–2001

Crude oil 0.04 –0.45** 3.30** 2.88** –0.01 –2.12 0.67

Natural gas 0.11** –0.19 1.15 –2.47* 0.53** 11.17** 0.60
Gold 1.09** 18.97* –1.06 –3.17 –0.58** 5.19 0.39
Copper –0.03 –26.30** 4.10 –4.86 –2.19** 24.24** 0.59
2002–06

Crude oil 0.11** 1.35** –1.61* 4.50** 0.30** 3.01* 0.42

Natural gas 0.02* 0.15* –0.26 1.44* 0.06 0.92 0.15
Gold 0.53* –23.10* –1.75 –11.77* 0.22 8.03* 0.41
Copper 0.24 1.14 –9.56** –36.51** –0.63 1.50 0.81

Note: * indicates significance at the 10% level, ** at the 5% level; bold red indicates expected sign and significance; light red indicates
expected sign and non-significance; bold black indicates incorrect sign and significance; light black indicates incorrect sign and non-
significance.
1
The seemingly unrelated regression methodology was used to estimate these results on monthly data in order to allow for
contemporaneous correlation in the errors across equations. All variables are lagged once. Other lag structures were tested, but the
effectiveness of this strategy was limited by the relatively short sample period. 2 Monthly percentage change in the spot
price. 3 Twelve-month moving average of the spot price minus the three-month forward price, divided by the spot price. 4 Twenty-
month rolling standard deviation of the monthly percentage change of the three-month futures price. 5 Average of three-month
interest rates of Canada, Germany, Japan, Sweden, the United Kingdom and the United States. 6 Correlation between the
percentage changes in the spot price and in the Morgan Stanley world equity price index over a rolling period of five years. 7 The
difference between nominal and real US 10-year bonds.

Sources: Bloomberg; CFTC; Datastream; Goldman Sachs Research; national data; BIS calculations. Table 4

5
The crude oil futures curve has been backwardated around half the time since 1998. Over this
period, the natural gas market has been backwardated only 15% of the time, while copper has
been backwardated 34% of the time. The futures curve for gold has almost always been in
contango due to the large level of above-ground inventories. Since 1975, the gold market has
backwardated only four times (in August 1976, May 1983, March 1986 and January 1993).

60 BIS Quarterly Review, March 2007

Electronic copy available at: https://ssrn.com/abstract=1600058

... although The coefficient on the interest rate is more significant across markets in
differences across
the second period, although with different signs. This supports the view that the
markets appear
considerable size and character of financial investor activity differ considerably across
markets. The negative sign for the gold and the copper markets, where the
shares of non-commercial positions are three to four times larger than the
share for crude oil markets (Graph 3), might indicate that the interest rate
variable reflects opportunity costs of financial investors with a shorter time
horizon. In energy markets, the positive coefficient might capture a trend
increase in net long positions resulting from passive tracking of commodity
indices, which tend to place a high weight on energy commodities. For
example, oil had an average weight of 27% in the second subperiod in the
GSCI index. However, no separate role could be found in the regression for the
GSCI commodity weights.
Diversification The second observation is that the share of non-commercial net long
benefits less
positions appears to have been less influenced by perceived diversification
significant
benefits than in the past. In the earlier subperiod, before prices started to
accelerate, there is a negative relationship between investor activity and the
correlation between returns on commodities and world equities in most cases.
In the second subperiod, this relationship is either statistically insignificant, or
has a perverse sign. One possible alternative explanation for this outcome is
that short-term strategies have been more important than before and dominate
the variation in the data. Another possibility is that the correlation variable does
not capture the full range of assets which have been relevant for the
assessment of diversification benefits in the recent period, although including
the correlation between commodity returns and other asset classes such as
high-yield credit does not change the result. Commodity investment might also
have been motivated by long-term historical correlations that are not apparent
in the relatively short span of the second subperiod. The relationship between
the share of non-commercial long positions and expected inflation is generally
positive, although not always significant, consistent with commodities being
purchased as a hedge against future inflation.

Financial investors and market dynamics

Questions Changes in the scale and character of involvement of financial investors in
commodity derivatives markets may have affected the price dynamics of these
markets. The first question in this regard is whether the exploitation of
perceived profit opportunities by financial investors has fundamentally changed
the relationship between prices and the physical characteristics of commodity
markets. The second issue is whether the broadening of the investor base has
led to significant market deepening and hence affected features such as short-
term price fluctuations.

The relationship with physical commodity markets

Investor activity and Intuitively, one might expect large inflows of funds into commodity markets to
commodity prices
cause prices to rise sharply, possibly to higher levels than are justified by
economic fundamentals. The prima facie evidence seems to support this view,

BIS Quarterly Review, March 2007 61

Electronic copy available at: https://ssrn.com/abstract=1600058

as financial activity has broadly increased in parallel with prices during the past
four years. However, the results of empirical work on the impact of the growing
presence of financial investors on commodity prices are less clear-cut. Several
recent studies, which explore the relationship between investor activity and
commodity prices, indicate that price changes have led to changes in investor
interest rather than the other way around (Haigh et al (2005), IMF (2006)).
This section uses the physical characteristics of specific commodities as a Physical
characteristics as a
rough benchmark for assessing whether the increased presence of financial
benchmark
investors has altered price dynamics. Constraints on supply and storability
affect the prices of commodity derivatives. In the longer run, production can be
changed and the elasticity of commodity supply depends on the marginal costs
of production. In the short run, supply from production is relatively inelastic and
depends more on above-ground stocks. With the exception of gold, above-
ground commodity stocks are small relative to demand. For example, it is usual
for four to six weeks of demand to be held in inventories for base metals. For
gold, in contrast, stocks either available for production or for lease represent
close to 45 years’ worth of demand, depending on how this is measured
(O’Connell (2005)).
In efficient markets, the expected marginal costs of commodity production Marginal costs of
production have
should act as an anchor for longer-run futures prices. Consistent with this, the
been a strong
long ends of oil and copper futures curves have overall tended to fluctuate anchor for long-
much less than spot and short-dated futures prices (Graph 4). The tenors that dated futures
prices ...
are affected by this “anchoring” may vary, depending on the time needed to
adjust production. For instance, from 1998 to 2002, a period of ample spare
capacity, marginal costs were steady and production could be expanded at
relatively short notice. Indeed, futures prices at tenors from about one year
were quite closely aligned with estimates of marginal costs of production in

Prices and marginal costs1

Daily data, 1998–2006

Crude oil2 Copper3

1998–2002
2003–05
75 10,000
2006
MC 98–02
60 MC 03–05 8,000
MC 2006
45 6,000

30 4,000

15 2,000

0 0
0 3 12 15 24 27 0 3 15 27 63
Months
1
The bold lines have been constructed by averaging the prices at each tenor within each time period. The
thin lines represent the futures curves associated with the minimum and maximum spot prices within each
time period. 2 In US dollars per barrel. 3 In US dollars per tonne.

Sources: Bloomberg; Goldman Sachs Research; JPMorgan Chase; BIS calculations. Graph 4

62 BIS Quarterly Review, March 2007

Electronic copy available at: https://ssrn.com/abstract=1600058

 Inventories and the slope of the futures curve
Crude oil Copper
1995–2001
2002–06

Backwardation 2 200

Backwardation

Slope1
0 100

–2 0
Contango
Contango
–4 –100
250 275 300 325 350 375 0 250 500 750 1,000 1,250
Inventories, in thousands of barrels Inventories, in thousands of tonnes
1
Spot price minus three-month futures price; for oil, in US dollars per barrel; for copper, in US dollars per
tonne.

Sources: Bloomberg; London Metal Exchange; BIS calculations. Graph 5

both oil and copper markets over this period.

... but seem to have Since 2003, however, long-dated futures prices have increasingly
lost power since
diverged from estimates of current marginal costs. In 2006, prices for two-year
2003 …
oil futures were on average about 20% higher than the measure of marginal
costs shown in Graph 4. In the case of copper, the deviation was much larger.
Several factors related to economic fundamentals could cause such a
deviation. For example, a sharp increase in expected marginal costs owing to
buoyant demand growth and uncertainty about the costs of further expansion of
production in the face of capacity constraints may have been a factor in the oil
market. Moreover, the need to explore and develop new sources has probably
lengthened the time required to extend production.
... to a degree In addition, futures prices are likely to embody risk premia, not least
which is difficult to
because long-dated futures markets are typically relatively thinly traded.
reconcile with
fundamentals Reluctance by producers to forgo upside opportunities through hedging in an
environment of rising prices might have further reduced liquidity. In contrast,
there is some tentative evidence that the size of the risk premium in oil futures
markets is positively related to the share of net non-commercial long positions
in the oil market, controlling for other factors (Micu (2005)). Notwithstanding all
these factors, it still appears difficult to reconcile the increases in futures prices
until mid-2006 with economic fundamentals, especially in the case of copper.
Inventory-slope A second physical anchor is inventories, which link current and future
relationship has
supply and consequently connect the spot price and expected spot prices in
remained intact ...
the future (Gorton and Rouwenhorst (2004)). It is not clear that growing
investor activity can have a systematic direct effect on inventory decisions: the
convenience that producers derive from holding stock importantly depends on
factors related to real activity such as production smoothing. Indeed, the strong
historical relationship between the slope of the futures curve for non-gold

BIS Quarterly Review, March 2007 63

Electronic copy available at: https://ssrn.com/abstract=1600058

commodities and the level of physical inventories has remained intact
(Graph 5).
It is more likely that financial investors could indirectly affect inventory ... but indirect effect
on inventory
decisions through futures prices. To the extent that taking long positions in
decisions possible
futures markets increases futures prices, the value of holding inventories for
future delivery increases. The effect on the slope of the yield curve remains
open, depending on how spot prices respond to possible inventory decisions.

Market depth

The second question is whether the increase in the size and diversity of Financial investors
and market depth
financial investors has increased market depth. Greater market depth would
imply that transactions of a given size cause smaller fluctuations and, other
things equal, that short-term price volatility should decline. The prima facie
evidence on changes in commodity price volatility is mixed. Price volatility has
declined in the oil market, especially in the shorter maturities of futures
contracts where trading is particularly active (Graph 6). In contrast, it has
increased in the copper market. 6
Another approach is to look at the interaction of the trading behaviour of Interaction of
commercial and
commercial and non-commercial traders. Non-commercial traders will add to
non-commercial
market depth if they contribute to a two-sided market. This is the case if they traders …
act as counterparties to commercial traders’ hedging transactions or if they
take positions offsetting other financial investors.
The pattern of changes in the open positions of commercial and non- ... seems to have
reduced volatility
commercial traders supports the view that financial investors have, overall,

Volatility of commodity futures prices

Annualised standard deviation of daily changes in the log of prices

Crude oil Copper

2002–06
1998–2001 40 25

35 20

30 15

25 10

20 5

15 0
0 3 12 15 24 0 3 15 27 63
Months

Sources: Bloomberg; BIS calculations. Graph 6

6
This highlights one of the limitations of the econometric work done earlier, insofar as changes
in investor activity cause changes in variables, such as volatility, that we have included as
explanatory variables.

64 BIS Quarterly Review, March 2007

Electronic copy available at: https://ssrn.com/abstract=1600058

 Volatility and correlation
Crude oil1 Correlations2

2001 Copper
16 Natural gas 1.2
Crude oil
1998
12 0.6
2003 2000

Volatility
2002
8 0
2004
1999
2005
4 –0.6

Correlations
0 –1.2
–0.3 0 0.3 0.6 0.9 Jan 98 Jan 00 Jan 02 Jan 04 Jan 06
1
The x-axis is the correlation between the changes in non-commercial long positions and commercial
short positions: the y-axis is the standard deviation of the spot oil prices. 2 Correlation between the
changes in non-commercial long positions and non-commercial short positions; 12-month moving average.

Sources: Bloomberg; CFTC; BIS calculations. Graph 7

contributed to deeper markets. 7 First, a higher correlation between changes in

non-commercial long and commercial short positions has been associated with
lower volatility in oil markets (Graph 7, left-hand panel). However, the
correlation has not significantly increased since 2002, suggesting that a
growing presence of financial investors may have accommodated increased
hedging needs, but not fundamentally altered the character of the market.
Non-commercial Second, there is also evidence that non-commercial traders have, as a
traders increasingly
group, increasingly taken positions on both sides of commodity markets. Prior
active on both sides
of commodity to 2002, changes in long and short positions of non-commercial traders were
markets highly negatively correlated for copper, oil and natural gas: an increase in long
positions typically went hand in hand with a reduction of short positions and
vice versa. There is also some evidence that MMTs tended to act on the same
side of the market at similar times in the past (CFTC (1996)). In the past few
years, however, the correlation between changes in long and short positions of
non-commercial traders has increased and become positive (Graph 7, right-
hand panel). Evidence that non-commercial players are increasingly trading
between each other is also provided by the growing share of spread positions,
which arise when a trader takes long and short positions in the same
commodity at different tenors of the futures curve.
Growing similarities The emergence of trading among financial investors in commodity markets
with financial
on a substantial scale suggests that the determinants of market liquidity may
markets
become more similar to those in traditional financial markets. These
determinants include the amount of risk capital that financial investors allocate
to commodities trading and the heterogeneity of opinions of market
participants. One key risk in both regards is a high concentration of trading

7
In order to gauge the position-taking of the investor groups on both sides of the market, we
consider correlations of long and short positions separately (ie we do not calculate net long or
short positions).

BIS Quarterly Review, March 2007 65

Electronic copy available at: https://ssrn.com/abstract=1600058

activity. The demise of Amaranth, which led to a sharp deterioration in liquidity
conditions in those tenors of the natural gas futures market where the firm held
extensive positions, provides a clear indication of these challenges.

Conclusion
The presence of financial investors in commodity markets has increased
considerably during the past four years or so. While it is difficult to be precise
about the exact magnitude and composition of inflows, there is much evidence
that the investor base, and with it the range of instruments and strategies
employed in commodity trading, has broadened substantially. It is not clear to
what extent these changes reflect structural shifts in investor behaviour or a
temporary boom supported by a “search for yield”. In any case, a full reversal
of the trend towards a greater role of financial investors appears unlikely
against the backdrop of greater investor sophistication and a broadening range
of commodity-related financial instruments.
Commodity markets have become more like financial markets in some
respects. Financial investors are increasingly active on both sides of trades,
creating a kind of financial trading sphere. Yet the characteristics of physical
markets, such as inventory levels and the marginal cost of production, are still
important. A lack of liquidity especially in the long tenors of commodity
derivatives markets and physical limits to short selling in the spot market may
at times significantly affect market dynamics. These effects require further
investigation.
While the increase in investor activity can be expected to bring benefits in
terms of market efficiency, the ongoing “financialisation” of commodity markets
raises issues similar to those in other financial markets. Among these is the
question of how to ensure robust market liquidity.

References
Bank for International Settlements (2006): Semiannual OTC derivatives
statistics, December.

Beenen, J (2005): “Commodities as a strategic investment for PGGM”, in

Michael Lewis (ed), An investor guide to commodities, Deutsche Bank.

Campbell, P, B-E Orskaug and R Williams (2006): “The forward market for oil”,
The Bank of England Quarterly Bulletin, Spring, pp 66–74.

Commodity Futures Trading Commission (1996): Report on a study of

managed money traders’ participation in futures and option-on-futures markets,
prepared by the Market Surveillance Section, Division of Economic Analysis,
March.

Davidson, C (2006): “Going electronic”, Risk, August, pp 44–6.

Duffie, D (1989): Future markets, Prentice-Hall, New Jersey.

Erb, C and C Harvey (2005): “The tactical and strategic value of commodity
futures”, NBER Working Papers, no 11222, May.

66 BIS Quarterly Review, March 2007

Electronic copy available at: https://ssrn.com/abstract=1600058

Feldman, B and H Till (2006): Separating the wheat from the chaff:
backwardation as the long-term driver of commodity futures performance;
evidence from soy, corn, and wheat futures from 1950 to 2004, EDHEC Risk
and Asset Management Research Centre, Nice.

Fusaro, P and G Vasey (2006): “Energy & environmental funds. Continuing to

offer superior opportunities?”, Commodities Now, September, pp 1–3.

Gorton, G and K Rouwenhorst (2004): “Facts and fantasies about commodity

futures”, NBER Working Papers, no 10595.

Haigh, M, J Hranaiova and J Oswald (2005): “Price dynamics, price discovery

and large futures trader interactions in the energy complex”, US Commodity
Futures Trading Commission Working Paper.

Holmes, D (2006): “A financial feast: a-la-carte commodity investing”, Alchemy,

issue 43, The London Bullion Market Association, pp 10–12.

International Monetary Fund (2006): Global Financial Stability Report,

September, Washington DC.

IntercontinentalExchange (2006): 10-K Filing, vol 10, March.

JPMorgan Chase (2007): Energy markets grow up part II: who trades energy
now and how much does it matter?, January.

Kolb, R (1997): Understanding futures markets, fifth edition, Blackwell

Publishers, Oxford.

McNee, A (2006): “Investors slake commodities thirst with structured products”,

The Banker, July, pp 40–2.

Micu, M (2005): “Declining risk premia in the crude oil futures market”, BIS
Quarterly Review, December, pp 50–1.

O’Connell, R (2005): “What sets the precious metals apart from other
commodities?”, gold:report, World Gold Council, December.

BIS Quarterly Review, March 2007 67

Electronic copy available at: https://ssrn.com/abstract=1600058