03 - FRM 15. Robert McDonald, Derivatives Markets, 2nd Edition (Ch. 6 Commodity Forwards and Futures)
03 - FRM 15. Robert McDonald, Derivatives Markets, 2nd Edition (Ch. 6 Commodity Forwards and Futures)
03 - FRM 15. Robert McDonald, Derivatives Markets, 2nd Edition (Ch. 6 Commodity Forwards and Futures)
observed that all happy families are all alike; each unhappy family is unhappy in
its own way. An analogous idea in financial markets might be: Financial forwards are
all alike; each commodity forward, however, has some unique economic characteristic
that must be understood in order to appreciate forward pricing in that market. In this
chapter we will see how commodity forwards and futures differ from, and are similar
to, financial forwards and futures.
In our discussion of forward pricing for financial assets we relied heavily on the
fact that for financial assets, the price of the asset today is the present value of the asset
at time T, less the value of dividends to be received between now and time T. We will
explore the extent to which this relationship also is true for commodities.
6.1 INTRODUCTION TO
COMMODITY FORWARDS
Chapter 5 introduced the formula for a forward price on a financial asset:
FO. T
Soe(r-8)T
(6.1)
where So is the spot price of the asset, r is the continuously compounded interest rate,
and i.s the continuous dividend yield on the asset. The difference between the forward'
price and spot price reflects the cost and benefits of delaying payment for, and receipt of,
the asset. In Chapter 5 we treated forward and futures prices as the same; we continue
to ignore the pricing differences in this chapter.
On any given day, for many commodities there are futures contracts available that
expire in a number of different months. The set of prices for different expiration dates for
a given commodity is called the forward curve or the forward strip for that date. Table
6.1 displays futures prices with up to 6 months to maturity for several commodities.
Let's consider these prices and try to interpret them using equation (6.1). To provide a
reference interest rate, 3-month LIBOR on May 5, 2004, was 1.22%, or about 0.3% for
3 months. From May to July, the forward price of corn rose from 314.25 to 319.75. This
is a 2-month increase of 319.75/314.25 - 1 = 1.75%, an annual rate of approximately
11 %, far in excess of the 1.22% annual interest rate. In the context of the formula
for pricing financial forwards, equation (6.1), we would need to have a continuous
dividend yield, of -9.19% in order to explain this rise in the forward price over time.
169
170
May
314.25
1034.50
June
July
319.75
39.57
393.80
1020.00
127.15
39.36
394.30
959.00
122.32
38.79
394.80
845.50
116.57
38.13
109.64
37.56
105.49
37.04
August
September
393.40
131.25
316.75
October
November
786.50
395.90
= 319.75
171
If the forward curve is downward sloping, as with gasoline, we say the market is in
backwardation. Forward curves can have portions in backwardation and portions in
contango, as does that for crude oil.
It would take an entire book to cover commodities in depth. Our goal here is to
understand the logic of forward pricing for commodities and where it differs from the
logic of financial forward pricing. What is the forward curve telling us about the market
for the commodity?
(6.2)
ST - Fo,T
Forward contract payoff
FOT
= ST
Bond payoff
where ST is the time T price of the commodity. This investment strategy creates a
synthetic commodity, in that it has the same value as a unit of the commodity at time
T. Note that, from equation (6.2), the cost of the synthetic commodity is the prepaid,
price, e- rT FO,T.
Valuing a synthetic commodity is easy if we can see the forward price. Suppose,
however, that we do not know the forward price. Computing the time 0 value of a unit
of the commodity received at time T is a standard problem: You discount the expected
commodity price to determine its value today. Let EO(ST) denote the expected
price as of time 0, and let denote the appropriate discount rate for a
T cash flow
of ST. Then the present value is
(6.3)
The important point is that expressions (6.2) and (6.3) represent the same value. Both
reflect what you would pay today to receive one unit of the commodity at time T.
Equating the two expressions, we have
e -rT
= E 0 (ST ) e
(6.4)
172
NONSTORABllITY: ELECTRICITY
(6.5)
=EO(ST
Equation (6.5) demonstrates the link between the expected commodity price,
and the forward price. As with financial forwards (see Chapter 5), the forward pnce IS a
biased estimate of the expected spot price, EO(ST), with the bias due to the risk premium
on the commodity, - r. 1
Equation 6.4 deserves emphasis: The time- T forward price discounted at the riskfree rate back to time 0 is the present value of a unit of commodity received at time T.
This calculation is useful when performing NPV calculations involving commodities for
which forward prices are available. Thus, for example, an industrial producer who buys
oil can calculate the present value of future oil costs by discounting oil forward prices at
the risk-free rate. The present value of future oil costs is not dependent upon whether or
not the producer hedges. We will see an example of this calculation later in the chapter.
.
If a commodity cannot be physically stored, the no-arbitrage pricing principles
discussed in Section 5.2 cannot be used to obtain a forward price. Without storage,
to
this
equation (6.5) determines the forward price. However, it is
formula, which requires forecasting the expected future spot pnce and estlmatmg
Moreover, even when physically possible, storage may be costly. Given the difficulties
of pricing commodity forwards, our goal will be to .interpret forward prices and to
understand the economics of different commodity markets.
In the rest of the chapter, we will further explore similarities and differences
between forward prices for commodities and financial assets. Some of the most important
differences have to do with storage: whether the commodity can be stored and, if so,
how costly it is to store. The next section provides an example of forward prices when
a commodity cannot be stored.
I Historical commodity and futures data, necessary to estimate expected commodity returns, are rel(1980) examine quarterly futures returns from 1950 to
atively hard to obtain. Bodie and
1976. while Gorton and Rouwenhorst (2004) examine monthly futures returns from 1959 to 2004.
Both studies construct portfolios of synthetic commodities-T-bills plus commodity futures-and find
that these portfolios earn the same average return as stocks, are on average negatively correlated with
stocks, and are positively correlated with inflation. These findings imply that a portfolio of stocks
and synthetic commodities would have the same expected return and less risk than a diversified stock
portfolio alone.
173
0000
Price
Time
Price
Time
Price
$35.68
0600
$40.03
1200
$61.46
Price
1800
$57.81
$62.18
0100
$31.59
0700
$49.64
1300
$61.47
1900
0200
$29.85
0800
$53.48
1400
$61.74
2000
$60.12
0300
$28.37
0900
$57.15
1500
$62.71
2100
$54.25
0400
$28.75
1000
$59.04
1600
$62.68
2200
$52.89
0500
$33.57
1100
$61.45
1700
$60.28
2300
$45.56
Source: Bloomberg.
that distinguish it not only from financial assets, but from other commodities as well.
What is special about electricity?
First, electricity is difficult to store, hence it must be consumed when it is produced
or else it is wasted. 2 Second, at any point in time the maximum supply of electricity
is fixed. You can produce less but not more. Third, demand for electricity varies
substantially by season, by day of week, and by time of day.
To illustrate the effects of nonstorability, Table 6.2 displays I-day ahead hourly
prices for 1 megawatt-hour of electricity in New York City. The I-day ahead forward
price is $28.37 at 3 A.M., and $62.71 at 3 P.M. Since you have learned about arbitrage,
you are possibly thinking that you would like to buy electricity at the 3 A.M. price and
sell it at the 3 P.M. price. However, there is no way to do so. Because electricity cannot
be stored, its price is set by demand and supply at a point in time. There is also no way to
buy winter electricity and sell it in the summer, so there are seasonal variations as well
as intraday variations. Because of peak-load plants that operate only when prices are
high, power suppliers are able to temporarily increase the supply of electricity. However,
expectations about supply are already reflected in the forward price.
Given these characteristics of electricity, what does the electricity forward price
represent? The prices in Table 6.2 are best interpreted using equation (6.5). The large
price swings over the day primarily reflect changes in the expected spot price, which in
tum reflects changes in demand over the day.
Notice two things. First, the swings in Table 6.2 could not occur with financial
assets, which are stored. (It is so obvious that financial assets are stored that we usually
don't mention it.) As a
the 3 A.M. and 3 P.M. forward prices for a stock
2There are ways to store electricity. For example, it is possible to use excess electricity to pump water
uphill and then, at a later time, release it to generate electricity. Storage is uncommon, expensive, and
entails losses, however.
174
will be almost identical. If they were not, it would be possible to engage in arbitrage,
buying low at 3 A.M. and selling high at 3 P.M. Second, whereas the forward price for
a stock is largely redundant in the sense that it reflects information about the current
stock price, interest, and the dividend yield, the forward prices in Table 6.2 provide
not otherwise obtain, revealing information about the future price
information we
of the commodity. This illustrates the forward market providing price discovery, with
forward prices revealing information, not otherwise obtainable, about the future price of
the commodity.
Now suppose that the continuously compounded interest rate is 10%. What is the
forward price for a pencil to be delivered in 1 year? Before reading any further, you
should stop and decide what you think the answer is. (Really. Please stop and think
about it!)
One obvious possible answer to this question, drawing on our discussion of financial forwards, is that the forward price should be the future value of the pencil price:
eO. 1 x $0.20 = $0.2210. However, common
suggests that this cannot be the correct answer. You hlOW that the pencil price in one year will be $0.20. If you entered
into a forward agreement to buy a pencil for $0.221, you would feel foolish in a year
when the price was only $0.20.
Common sense also rules out the forward price being less than -$0.20. Consider
the forward seller. No one would agree to sell a pencil for a forward price of less than
$0.20, knowing that the price will be $0.20.
Thus, it seems as if both the buyer and seller perspective lead us to the conclusion
that the forward price must be $0.20.
asset.
One way to describe this difference between the pencil and the stock is to say that,
in equilibrium, stocks and other financial assets must be held by investors, or stored. This
is why the stock price appreciates on average; appreciation is necessary for investors to
willingly store the stock.
The pencil, by contrast, need not be stored. The equilibrium condition for pencils
requires that price equals marginal production cost. This distinction between a storage
3
and production equilibrium is a central concept in our discussion of commrnodities.
175
Short-sell pencil
+$0.20
-$0.20
Total
$0.20 -
FO,I
-$0.20
$0.221
$0.221 -
FO,I
176
Once again it is time to stop and think before proceeding. Examine Table 6.3
closely; there is a problem.
The arbitrage assumes that you can short-sell a pencil by borrowing it today and
returning it in a year. However, recall that pencils cost $0.20 today and will cost $0.20
in a year. Borrowing one pencil and returning one pencil in a year is an interest-free
loan of $0.20. No one will lend you the pencil without charging you an additionalfee.
If you are to short-sell, there must be someone who is both holding the asset and
willing to give up physical possession for the period of the short-sale. Unlike stock,
nobody holds pencils in a brokerage account. It is straightforward to borrow a financial
asset and return it later, in the interim paying dividends to the owner. However, if you
borrow an unused pencil and return an unused pencil at some later date, the owner of
the pencil loses interest for the duration of the pencil loan since the pencil price does not
change.
Thus, the apparent arbitrage in the above table has nothing at all to do with
forward contracts on pencils. If you find someone willing to lend you pencils for a year,
you should borrow as many as you can and invest the proceeds in T-bills. You will earn
the interest rate and pay nothing to borrow the money.
You might object that pencils do provide a flow of services-namely, making
marks on paper. However, this service flow requires having physical possession of the
pencil and it also uses up the pencil. A stock loaned to a short-seller continues to earn its
return; the pencil loaned to the short-seller earns no retum for the lender. Consequently,
the pencil borrower must make a payment to the lender to compensate the lender for lost
time value of money.
177
$0.20
+$0.20
-$0.20
$0.221
$0.20
Total
FO,1
Fa, I
FO,1
$0.20
-$0.20
+$0.20
Borrow @ 10%
+$0.20
-$0.221
Total
FO,1
$0.221
the forward price is $0.21. We would buy a pencil and sell it forward, and simultaneously
lend the pencil. To see that this strategy is profitable, examine Table 6.6.
Income from lending the pencil provides the missing piece: Any forward price
greater than $0.20 now results in arbitrage profits. Since we also have seen that any
forward price less than $0.20 results in arbitrage profits, we have pinned down the
forward price as $0.20.
Finally, what about equation (6.5), which we claimed holds for all commodities
and assets? To apply this equation to the pencil, recognize that the appropriate discount
rate, for a risk-free pencil is r, the risk-free rate. Hence, we have
FO,T
e(O,lO-O.lO)
= 0.20
178
179
Then from equation (6.6), the NPV of the commodity loan, without payments, is
Cash and carry arbitrage with pencil lending. When the
pencil is loaned, interest is earned and the no-arbitrage
price is $0.20.
-$0.20
Borrow @ 10%
FO,1
Total
Soe(g-a)T
So
(6.7)
$0.20
+$0.20
(6.8)
-$0.221
FO,1
If g <
the commodity loan has a negative NPY. However, suppose the lender demands
that the borrower return e(a-g)T units of the commodity for each unit borrowed. If one
unit is loaned, e(a-g)T units will be returned. This is like a continuous proportional
rate is the difference between the
lease payment of - g to the lender. Thus, the
commodity discount rate and the expected growth rate of the commodity price, or
0.021
+$0.20
NPV
$0.20
The pencil is obviously a special example, but this discussion establishes the important point that in order to understand arbitrage relationships for commodity forwards,
we have to think about the cost of borrowing and income from lending an asset. Borrowing and leasing costs also determine the pricing of financial forwards, but the cash
flow associated with borrowing and lending financial assets is the dividend yield, which
is readily observable. The commodity analogue to dividend income is lease income,
which may not be directly observable. We now discuss leasing more generally.
So = 0
(6.9)
Now the commodity loan is a fair deal for the lender. The commodity lender must be
compensated by the borrower for the opportunity cost associated with lending. When
the future pencil price was certain to be $0.20, the opportunity cost was the risk-free
interest rate, 10%.
Note that if ST were the price of a nondividend-paying stock, its expected rate
of appreciation would equal its expected return, so g =
and no payment would be
for the stock loan to be a fair dea1. 4 Commodities, however, are produced; as
with the pencil, their expected price appreciation need not equal
(6.10)
Tables 6.7 and 6.8 verify that this formula is the no-arbitrage price by performing the
cash-and-carry and reverse cash-and-carry arbitrages. In both tables we tail the position
in order to offset the lease income.
The striking thing about Tables 6.7 and 6.8 is that on the surface they are exactly
like Tables 5.6 and 5.7, which depict arbitrage transactions for a dividend-paying stock.
In an important sense, however, the two sets of tables are quite different. With the stock,
is an observable characteristic of the stock, reflecting payment
the dividend yield,
received by the owner of the stock whether or not the stock is loaned.
4As we saw in Chapter 5, for a nondividend-paying stock, the present value of the future stock price is
the current stock price.
180
CARRY MARKETS
In some markets, consistent and reliable quotes for the spot price are not available,
or are not comparable to forward prices. In such cases, the near-term forward price can
be used as a proxy for the spot price, S.
By definition, contango-an upward-sloping forward curve-occurs when the
lease rate is less than the risk-free rate. Backwardation-a downward-sloping forward
curve-occurs when the lease rate exceeds the risk-free rate.
Fo,T - ST
-Soe- a/T
Borrow @ r
+Soe- a/T
6.6
+ST
5a e(r-li/lT.
o
+Soe-a/T
-Soe-a/T
ST - Fo,T
-ST
+Soe(r-a/lT
Total
With pencils, by contrast, the lease rate, = - g, is income earned only if the
pencil is loaned. In fact,. notice in Tables 6.7 and 6.8 that the
never stores
the commodity! Thus, equation (6.10) holds whether or not the commodIty can be, or
is, stored.
One of the implications of Tables 6.7 and 6.8 is that the lease
has to be
consistent with the forward price. Thus, when we observe the forward pnce, we can
infer what the lease rate would have to be if a lease market existed. Specifically, if the
forward price is Fo,T, the annualized lease rate is
1
= r - y:ln(Fo,T/S)
(6.11)
If instead we use an effective annual interest rate, r, the effective annual lease rate is
_
(1+r)
-1
(FO,T/S)ljT
MARKETS
Sometimes it makes sense for a commodity to be stored, at least temporarily. Storage is
market.
also called carry, and a commodity that is stored is said to be in a
One reason for storage is seasonal variation in either supply or demand, which
causes a mismatch between the time at which a commodity is produced and the time at
which it is consumed. With some agricultural products, for example, supply is seasonal
(there is a harvest season) but demand is constant over the year. In this case, storage
permits consumption to occur throughout the year.
With natural gas, by contrast, there is high demand in the winter and low demand
the summer, but relatively constant production over the year. This pattern of use and
production suggests that there will be times when natural gas is stored.
Total
181
(6.12)
Storage is not always feasible (for example, fresh strawberries are perishable) and when
technically feasible, storage is almost always costly. When storage is feasible, how
do storage costs affect forward pricing? Put yourself in the position of a commodity
merchant who owns one unit of the commodity and ask whether you would be willing
to store this unit until time T. You face the choice of selling it today, receiving So, or
selling it at time T. If you elect to sell at time T, you can sell forward (to guarantee the
price you will receive), and you will receive FO,T. This is a cash-and-carry.
The cash-and-carry logic with storage costs suggests that you will store only if the
present value ofselling at time T is at least as great as that ofselling today. Denote the
T).
future value of storage costs for one unit of the commodity from time 0 to T as
Indifference between selling today and at time T requires
= e- rT JFo,T Revenue from selling today
This relationship in tum implies that if storage is to occur, the forward price is at least
Fo,T
Soe rT +
T)
(6.13)
In the special case where storage costs are paid continuously and are proportional to the
value of the commodity, storage cost is like a continuous negative dividend of and we
can write the forward price as
Fo,T =
(6.14)
182
CARRY MARKETS
Example 6.1
+ ($0.05
x 1.01)
+ ($0.05
= ($0.05/.01) x
= $0.1515
[(l + 0.01)3 -
1]
+ 0.1515 =
corn, you can sell the excess. However, if you hold too little and run out of corn, you
must stop producing, idling workers and machines. Your physical inventory of corn in
this case has value-it provides insurance that you can keep producing in case there is
a disruption in the supply of corn.
In this situation, corn holdings provide an extra nonmonetary return that is sometimes referred to as the convenience yield.S You will be willing to store corn with a lower
rate of return than if you did not earn the convenience yield. What are the implications
of the convenience yield for the forward price?
Suppose that someone approached you to borrow a commodity from which you
derived a convenience yield. You would think as follows: "If I lend the commodity, I
am bearing interest cost, saving storage cost, and losing the value I
from having
a physical inventory. I was willing to bear the interest cost already; thus, I will pay a
commodity borrower storage cost less the convenience yield."
Suppose the continuously compounded convenience yield is c, proportional to the
value of the commodity. The commodity lender saves - c by not physically storing the
compensating the lender
commodity; hence, the commodity borrower pays = c for convenience yield less storage cost. Using an argument identical to that in Table 6.8,
conclude that the forward price must be no less than
2.7273
Fo,T
Soe(r-8)T
may be no wayfor you to eam the convenience yield when pelforming a cash-and-carry.
Those
do not earn the convenience yield will not own the commodity.
Thus,for all average investor, the cash-and-carry has the cash flows 6
Suppose that there is a carry market for a commodity, so that its forward price is given
by equation (6.13). What is the lease rate in this case?
Again put yourself in the shoes of the commodity lender. If you lend the commodity, you are saved from having to pay storage cost. Thus, the lease rate should equal the
negative of the storage cost. In other words, the lender will pay the borrower! In effect,
the commodity borrower is providing "virtual storage" for the commodity lender, who
receives back the commodity at a point in the future. The lender making a payment to
the borrower generates a negative dividend.
5The term convenience yield is defined differently by different authors. Convenience yield generally
means a return to physical ownership of the commodity. In practice it is sometimes used to mean the
lease rate. In this book, the lease rate of a commodity can be inferred from the forward price using
equation (6.1 I).
The discussion of commodities to this point has ignored business reasons for holding
commodities. For example, suppose you are a food manufacturer for whom corn is an
essential input. You will hold an inventory of corn. If you end up holding too much
183
Fo,T -
ST
+ ST
Fo,T
6In this expression, we assume we tail the holding of the commodity by buying
and selling off units of the commodity over time to pay storage costs.
if there is to be
e)T
units at time 0,
184
In summary, from the perspective of an arbitrageur, the price range within which
there is no arbitrage is
Soe(r+/.-clT
FO.T
(6.15)
The convenience yield produces a no-arbitrage region rather than a no-arbitrage price.
The observed lease rate will depend upon both storage costs and convenience. Also,
as in Section 5.3, bid-ask spreads and trading costs will further expand the no-arbitrage
region in equation (6.15).
As another illustration of convenience yield, consider again the pencil example
Section 6.4. In reality, everyone stores a few pencils in order to be sure to have one
available. You can think of this benefit from storage as the convenience yield of a pencil.
However, because the supply of pencils is perfectly elastic, the price of pencils is fixed
at $0.20. Convenience yield in this case does not affect the forward price, but it does
explain the decision to store pencils.
The difficulty with the convenience yield in practice is that convenience is hard to
observe. The concept of the convenience yield serves two purposes. First, it explains
patterns in storage-for example, why a commercial user might store a commodity
when the average investor will not. Second, it provides an additional parameter to better
explain the forward curve. You might object that we can invoke the convenience yield to
explain any forward curve, and therefore the concept of the convenience yield is vacuous.
While convenience yield can be tautological, it is a meaningful economic concept and
it would be just as arbitrary to assume that there is never convenience. Moreover, the
upper bound in equation (6.15) depends on storage costs but not the convenience yield.
Thus, the convenience yield only explains anomalously low forward prices, and only
when there is storage.
We will now examine particular commodities to illustrate the concepts from the
previous sections.
6.7 GOLD
FUTURES
Gold is durable, relatively inexpensive to store (compared to its value), widely held,
and actively produced through gold mining. Because of transportation costs and purity
concerns, gold often trades in certificate form, as a claim to physical gold at a specific
location. There are exchange-traded gold futures, specifications for which are in Figure
6.1. 7
Figure 6.2 is a newspaper listing for the NYMEX gold futures contract. Figure 6.3
graphs the futures prices for all available gold futures contracts-the forward curvefor the first Wednesday in June, from 2001 to 2004. (Newspaper listings for most
futures contracts do not show the full set of available expiration dates, so Figure 6.3
is usually denominated in troy ounces (480 grains), which are approximately 9.7% heavier than
the more familiar avoirdupois ounce (437.5 grains). Twelve troy ounces make I troy pound, which
weighs approximately 0.37 kg.
Underlying
Where traded
Size
Months
Trading ends
Delivery
HIGH
July
AU9
o,t
AU9
406.00
40850
4lLOO
420.00
CHG
LIFETIME
HIGH
OPEN
INT
40L90 -3JO
399.70 402.10 -3.70
40L20 403.40 -3.70
402.00
-3.70
405.00 406.40 -3.70
4lLOO 4lL80 -3.70
420.00 415.90 -3.70
31,225; open int 262,052.
40010 38050
433.00 324.70
432.00 332.00
290.00
435.00 33L50
379.00
298.40
4
139,287
U310
62,036
LOW SETTLE
GoId CCMX)100
407.10
408.00
409.90
40350
4lLOO
2,290
6,387
06-jun-2001
05-jun-2002
04-]un-2003
02-]un-2004
460
440
420
400
380
360
340
320
300
280
260
7 Gold
185
GOLD FUTURES
__
10
__
20
30
40
Months to Maturity
50
60
186
GOLD FUTURES
is constructed using more expiration dates than are in Figure 6.2.) What is interesting
about the gold forward curve is how relatively uninteresting it is, with the forward price
steadily increasing with time to maturity.
From our previous discussion, the forward price implies a lease rate. Short-sales
and loans of gold are common in the gold market, and gold borrowers in fact have to pay
the lease rate. On the lending side, large gold holders (including some central banks) put
gold on deposit with brokers, in order that it may be loaned to short-sellers. The gold
lenders earn the lease rate.
The lease rate for gold, silver, and other commodities is computed in practice using
equation (6.12) and is reported routinely by financial reporting services. Table 6.9 shows
the 6-month and I-year lease rates for the four gold forward curves depicted in Figure
6.3, computed using equation (6.12);
Example 6.2 Here are the details of computing the 6-month lease rate for June
6, 2001. Gold futures prices are in Table 6.9. The June and September
futures prices on this date were 96.09 and 96.13. Thus, 3-month LIBOR from June
to September was (100 - 96.09)/400 x 91/90 = 0.988%, and from September to
December was (100 - 94.56)/400 x 91/90 = 0.978%. The June to December interest
rate was therefore (1.00988) x (1.00978) - 1 = 1.9763%, or 1.0197362 annualized.
Using equation (6.12), the annualized 6-month lease rate is therefore
1.0197632
)
6-month lease rate = ( (269/265.7)(1/0.5) - 1 = 1.456%
187
Gold Investments
If you wish to hold gold as part of an investment portfolio, you can do so by holding
physical gold or synthetic gold-i.e., holding T-bills and going long gold futures. Which
should you do? If you hold physical gold without lending it, and if the lease rate is
positive, you forgo the lease rate. You also bear storage costs. With synthetic gold,
on the other hand, you have a counterparty who may fail to pay so there is credit risk.
Ignoring credit risk, however, synthetic gold is generally the preferable way to obtain
gold price exposure.
Table 6.9 shows that the 6-month annualized gold lease rate is 1.46% in June
2001. Thus, by
physical gold instead of synthetic gold, an investor would lose
this 1.46% return. In June 2003 and 2004, however, the lease rate was about -0.10%. If
storage costs are about 0.10%, an investor would be indifferent between holding physical
and synthetic gold. The futures market on those dates was compensating investors for
storing physical gold.
Some nonfinancial holders of gold will obtain a convenience yield from gold.
Consider an electronics manufacturer who uses gold in producing components. Suppose
that running out of gold would halt production. It would be natural in this case to hold
a buffer stock of gold in order to avoid a stock-out of gold, i.e., running out of gold.
For this manufacturer, there is a return to holding gold-namely, a lower probability of
stocking out and halting production. Stocking out would have a real financial cost, and
the manufacturer is willing to pay a price-the lease rate-to avoid that cost.
PV gold production
[Fo,li -
x(t;)] e-r(O,li)li
(6.16)
;=1
June 6, 2001
265.7
269.0
271.7
1.46%
1.90%
June 5, 2002
321.2
323.9
326.9
0.44%
0.88%
June 4,2003
362.6
364.9
366.4
-0.14%
0.09%
June 2,2004
391.6
395.2
400.2
-0.10%
0.07%
This equation assumes that the gold mine is certain to operate the entire time and that
the quantity of production is known. Only price is uncertain. (We will see in Chapter 17
8The cost of I ounce of physical gold is So. However, from equation (6.10), the cost of I ounce of gold
bought as a prepaid forward is Soe- oIT Synthetic gold is proportionally cheaper by the lease rate,
188
Expiration Year
For,va.rdPrice ($)
1
2
313.81
295.53
328.25
291.13
3
4
5
6
343.36
286.80
359.17
282.53
375.70
278.32
392.99
274.18
how the possibility of mine closings due to low prices affects valuation.) Note that in
equation (6.16), by computing the present value of the forward price, we compute the
prepaid forward price.
Example 6.3 Suppose we hav.e a mining project that will produce 1 ounce of gold
every year for 6 years. The cost of this project is $1,100 today, the marginal cost per
ounce at the time of extraction is $100, and the continuously compounded interest rate
is 6%.
We observe the gold forward prices in the second column of Table 6.10, with
implied prepaid forward prices in the third column. Using equation (6.16), we can use
these prices to perform the necessary present value calculations.
HIGH
$1100 = $119.56
(6.17)
i=l
LIFETIME
-3.00
-3.00
-2.50
-3.00
-2.50
-2.75
-2.75
34100 1.29.75
341.50 232.50
342.00 239.00
243.50
342.00 246.50
299.00 260.00
288.50 235.00
+1,4BB.
162,000
307,442
52,447
16,60B
13,717
10,707
Net present
189
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
__
0.5
1.0
__
__
1.5
Year
2.0
2.5
3.0
190
of com rises to reward storage, and it falls at each harvest. Let's see how this graph was
constructed.
The com price is $2.50 initially, the continuously compounded interest rate is 6%,
and storage cost is 1.5%/month. The forward price after n months (where n < 12) is
F
01l
= $2.50 x
e(O.005 +O.015)x1I
Thus, the 12-month forward price is $2.50eo.o6+o.ls = $3.18. After 1 year, the process
starts over.
Farmers will plant in anticipation of receiving the harvest price, which means that
it is the harvest price that reflects the cost of producing com. The price during the rest of
the year equals the harvest price plus storage. In general we would expect those storing
corn to plan to deplete inventory as harvest approaches and to replenish inventory from
the new harvest.
This is a simplified version of reality. Perhaps most important, the supply of com
varies from year to year. When there is a large crop, producers will expect corn to be
'stored not just over the current year, but into the next year as well. If there is a large
harvest, therefore, we might see the forward curve rise continuously until year 2. To
better understand the possible behavior of com, let's look at real com prices.
Table 6.11 shows the June forward curves for com over a lO-year period. Some
clear patterns are evident. First, notice that from December to March to May (columns
3-5), the futures price rises every year. We would expect there to be storage of com
during this period, with the futures price compensating for storage. A low current price
suggests a large supply. Thus, when the near-July price is low, we might also expect
storage across the coming harvest. Particularly in the years with the lowest July prices
(1999-2002), there is a pronounced rise in price from July to December. When the
price is unusually high (1996 and 2004), there is a drop in price from July to December.
Behavior is mixed in the other years. 9 We can also examine the July-December price
relationship in the following year. In 6 of the 10 years, the distant-December price
(column 8) is below the distant July price (column 6). The exceptions occur in years
with relatively low current prices (1998-2001). These patterns are generally consistent
with storage of corn between harvests, and storage across harvests only occasionally.
Finally, compare prices for the near-July contract (the first column) with those for
the distant-December contract (the last column). Near-term prices are quite variable,
ranging from 216.75 to 435.00 cents per bushel. In December of the following year,
however, prices range only from 239 to 286.50. In fact, in 7 of the 10 years, the price
is between 251 and 268. The lower variability of distant prices is not surprising: It is
difficult to forecast a harvest more than a year into the future. Thus, the forward price is
reflecting the market's expectation of a normal harvest 1 year hence.
If we assume that storage costs are approximately $0.03/month/bushel, the forward
price in Table 6.11 never violates the no-arbitrage condition
FO,T+s
< FO,Te
rs
265.50
272.25
277.75
282.75
285.25
286.00
269.50
253.50
05-Jun-1996
435.00
373.25
340.75
346.75
350.50
348.00
298.00
286.50
04-Jun-1997
271.25
256.50
254.75
261.00
265.50
269.00
257.00
255.00
245.75
253.75
259.00
264.00
261.00
268.00
244.50
248.25
248.00
251.50
07-Jun-1995
03-Jun-1998
238.00
242.25
191
NATURAL GAS
02-Jun-1999
216.75
222.00
230.75
07-Jun-2000
219.75
228.50
234.50
239.25
242.50
248.50
254.50
259.00
06-Jun-2001
198.50
206.00
217.25
228.25
234.50
241.75
245.00
251.25
05-Jun-2002
210.25
217.25
226.75
234.50
237.25
240.75
235.00
239.00
04-Jun-2003
237.50
236.25
237.75
244.00
248.00
250.00
242.25
242.50
02-Jun-2004
321.75
319.75
319.25
322.50
325.50 324.50
296.50
279.00
+ s)
(6.18)
9It is possible to have low current storage and a large expected harvest, which would cause the December
price to be lower than the July price, or high current storage and a poor expected harvest, which would
cause the July price to be below the December price.
192
NATURAL GAS
OPEN
HIGH
Naturai Gas
AU9
Sept
Oct
Nov
5.815
5.865
5.910
6.560
Feb
6.705
Mar
June
July
Aug
Oct
Nov
JI06
Ap07
Oct
5.980
5.890
5.900
5.960
5.990
6.150
6350
5.520
5.109
5.111
5.090
5.960
5.990
6360
6.660
6.835
6.800
6.650
6.010
5.900
5.910
5.960
5.960
5.990
6.160
6350
5.520
UFETIME
HIGH
OPEN
INT
$
5.837
5.797
5.835
5.898
5.8n
6.584
6.730
6.705
6.560
5.870
5.890
5.935
5.950
5.990
6.150
6350
6.722
6.592
5.987
5.890
5.927
5.937
5.952
6302
5.189
5.109
5.109
5.111
5.Ill
5.090
5.090
vol
61.746;
.019
.Oll
.015
.016
.016
.016
.017
.018
.Oll
.017
.017
.017
.008
.008
6.780
6.800
6.940
7.110
7.230
6.970
6.200
6.020
6.030
6.070
6.080
6.080
6.240
6.400
5.520
5.ll0
5.111
5.090
int 380,824. +3,187.
5.039
3.110
3.100
3.100
3170
3.520
3.400
3.500
3.530
3.560
3.230
3.790
3.960
3.580
4.711
4.000
4.891
60,098
40,028
15,510
17,636
1l,388
06-Jun-2001
05-Jun-2002
04-Jun-2003
===
7.5
,\
7
6.5
02-Jun-2004
....
'\
8,340
5.5
8,036
6,627
6,nO
3,074
3,187
767
376
I
i
Underlying
Where traded
Size
Months
Trading ends
Delivery
Figure 6.8 displays 3-year (2001) and 6-year (2002-2004) strips of gas futures
prices for the first Wednesday in June from 1997 to 2000. Seasonality is evident, with
high winter prices and low summer prices. The 2003 and 2004 strip shows seasonal
cycles combined with a downward trend in prices, suggesting that the market considered
prices in that year as anomalously high. For the other years, the average price for each
coming year is about the same.
Gas storage is costly and demand for gas is highest in the winter. The steady rise
of the forward curve during the fall months suggests that storage occurs just before the
heaviest demand. Table 6.12 shows prices for October through December. The monthly
increase in gas prices over these months ranges from $0.13 to $0.23. Assuming that the
interest rate is about 0.15% per month and that you use equation (6.13), storage cost in
November 2004, would satisfy
6.947 = 6.75geo.OOI5
'..
4.5
I
\
!.
.....
......:
1\
4
3.5
10
193
20
30 40 50 60 70
Months to Maturity
80
06-Jun-2001
2.173
2.305
2.435
05-Jun-2002
3.352
3.617
3.850
04-Jun-2003
6.428
6.528
6.658
02-Jun-2004
6.581
6.759
6.947
Source: Futures data from Datastream.
implying an estimated storage cost of = $0.178 in November 2004. You will find
different imputed marginal storage costs in each year, but this is to be expected if marginal
storage costs vary with the quantity stored.
Because of the expense in transporting gas internationally, the seasonal behavior
of the forward curve can vary in different parts of the world. In tropical areas where gas
is used for cooking and electricity generation, the forward curve is relatively flat'because
demand is relatively flat. In the Southern hemisphere, where seasons are reversed from
the Northern hemisphere, the forward curve will peak in June and July rather than
December and January.
194
Recent developments in energy markets could alter the behavior of the natural gas
forward curve in the United States. Power producers have made greater use of gas-fired
peak-load electricity plants. These plants have increased summer demand for natural
gas and may permanently alter seasonality.
6.10 OIL
Both oil and natural gas produce energy and are extracted from wells, but the different
physical characteristics and uses of oil lead to a very different forward curve than that
for gas. Oil is easier to transport than gas. Transportation of oil takes time, but oil has
a global market. Oil is also easier to store than gas. Thus, seasonals in the price of
crude oil are relatively unimportant. Specifications for the NYMEX light oil contract
are shown in Figure 6.9. Figure 6.10 shows a newspaper listing for oil futures. The
NYMEX forward curve on four dates is plotted in Figure 6.11.
Seven-year strip of
NYMEX crude oil
futures prices, $/barrel,
for four different dates.
06 lun 2001
05-]un-2002
..... 04-]un-2003
02-]un-2004
===
40
\
38
36
34
-- --
32
22
20
Underlying
Specifications for the
NYMEX light, sweet
crude oil contract.
Where traded
Size
Months
Trading ends
Delivery
UFETlCvlE
OPEN
HIGH
INT
bbls.; per bbl.
AU9
4230
4L90
Oct
flov
Ja05
Feb
Apr
June
Sept
0,06
0,07
0,08
0,09
oelO
vol
41.63
4L41
40.B5
4038
39.85
3889
38.05
36.65
36.00
195
COMMODITY SPREADS
4230
4L05
40.45
40.10
39.50
39.05
38.20
37.45
36.65
36.05
34.15
32.75
32.75
3L75
3L20
3L20
30.85
30.95
204,514, vol
40.51
4031
39.80
3952
39.00
39.20
3865
38.20
37.85
36.60
36.65
35.90
34.15
32.75
3L70
40.B6 -0.78
- LOO
39.88 -LOO
39.45 -0.93
39.04 -0.B9
38.55 -0.86
-0.83
37.74
3737 -0.78
36.68 -0.73
35.96
35.44 -0.61
3358 -058
32.19 -0.55
-0.55
30.48
30.63 30.18 -055
24B,39B;
Int
40.70
4030
39.50
39.15
38.65
38.50
3755
36.85
36.30
20.84
20.82
B.75
24.75
16.35
B.85
B.05
B6,034
68,764
35,763
60,672
B,149
13,656
10,589
22.40
24.00
17.00
19.10
19.50
19.75
32.75
32.00
3L20
3LOO 27.15
-1,567.
8,357
47,774
34,365
10,916
15,928
..
=
....=.=.=
.====...
__
10
20
30
40
50
60
Months to Maturity
70
80
On the four dates in the figure, near-term oil prices range from $25 to $40, while'
the 7-year forward price in each case is between $22 and $30. The long-run forward
price is less volatile than the short-run forward price, which makes economic sense. In
the short run, an increase in demand will cause a price increase since supply is fixed. A
supply shock (such as production restrictions by the Organization ofPetroleum Exporting
Countries [OPEC]) will cause the price to increase. In the long run, however, both supply
and demand have time to adjust to price changes with the result that price movements
are
The forward curve suggests that market participants in June 2004 did not
expect the price to remain at $40lbarrel.
196
HEDGING STRATEGIES
hence, the difference in price between crude oil and equivalent amounts of heating oil
and gasoline is called the crack spread.
Oil can be processed in different ways, producing different mixes of outputs. The
spread terminology identifies the number of gallons of oil as input, and the number of
gallons of gasoline and heating oil as outputs. Traders will speak of "5-3-2," "3-2-1,"
and "2-1-1" crack spreads. The 5-3-2 spread, for example, reflects the profit from taking
5 gallons of oil as input, and producing 3 gallons of gasoline and 2 gallons of heating oil.
A petroleum refiner producing gasoline and heating oil could use a futures crack spread
to lock in both the cost of oil and output prices. This strategy would entail going long
oil futures and short the appropriate quantities of gasoline and heating oil futures. Of
course there are other inputs to production and it is possible to produce other outputs,
such as jet fuel, so the crack spread is not a perfect hedge.
Example 6.4 Suppose we consider buying oil in July and selling gasoline and heating oil in August. On June 2, 2004, the July futures price for oil was $39.96/barrel,
or $0.95l4/gallon (there are 42 gallons per barrel). The August futures prices for unleaded gasoline and heating oil were $1.2427/gallon and $1.0171/gallon. The 3-2-1
crack spread tells us the gross margin we can lock in by buying 3 gallons of oil and
producing 2 gallons of gasoline and 1 of heating oil. Using these prices, the spread is
(2 x $1.2427)
+ $1.0171 -
(3 x $0.9514) = $0.6482
There are crack spread options trading on NYMEX. Two of these options pay
based on the difference between the price of heating oil and crude oil, and the price of
gasoline and heating oil, both in a 1: 1 ratio.
197
location underlying the principal natural gas futures contract (see again Figure 6.7). In
some cases, one commodity may be used to hedge another. As an example of this we
discuss the use of crude oil to hedge jet fuel. Finally, weather derivatives provide another
example of an instrument that can be used to cross-hedge. We discuss degree-day index
contracts as an example of such derivatives.
Basis Risk
Exchange-traded commodity futures contracts call for delivery of the underlying commodity at specific locations and specific dates. The actual commodity to be bought or
sold may reside at a different location and the desired delivery date may
match that of
the futures contract. Additionally, the grade of the deliverable under the futures contract
may not match the grade that is being delivered.
This general problem of the futures or forward contract not representing exactly
what is being hedged is called basis risk. Basis risk is a generic problem with commodities because of storage and transportation costs and quality differences. Basis risk
can also arise with financial futures, as for example when a company hedges its own
borrowing cost with the Eurodollar contract.
Section 5.5 demonstrated how an individual stock could be hedged with an index
futures contract. We saw that if we regressed the individual stock return on the index
return, the resulting regression coefficient provided a hedge ratio that minimized the
variance of the hedged position.
In the same way, suppose we wish to hedge oil delivered on the East Coast with
the NYMEX oil contract, which calls for delivery of oil in Cushing, Oklahoma. The
variance-minimizing hedge ratio would be the regression coefficient obtained by regressing the East Coast price on the Cushing price. Problems with this regression are
that the relationship may not be stable over time or may be estimated imprecisely.
Another example of basis risk occurs when hedgers decide to hedge distant obligations with near-term futures. For example, an oil producer might have an obligation
to deliv.er 100,000 barrels per month at a fixed price for a year. The natural way to hedge '
this obligation would be to buy 100,000 barrels per month, locking in the price and
supply on a month-by-month basis. This is called a strip hedge. We engage in a strip
hedge when we hedge a stream of obligations by offsetting each individual obligation
with a futures contract matching the maturity and quantity of the obligation. For the
oil producer obligated to deliver every month at a fixed price, the hedge would entail
buying the appropriate quantity each month, in effect taking a long position in the strip.
An alternative to a strip hedge is a stack hedge. With a stack hedge, we enter into
futures contracts with a single maturity, with the number of contracts selected so that
changes in the pres.ent vallie of the future obligations are offset by changes in the value
of this "stack" of futures contracts. In the context of the oil producer with a monthly
delivery obligation, a stack hedge would entail going long 1.2 million barrels using the
near-term contract. (Actually, we would want to tail the position and short less than
1.2 million barrels, but we will ignore this.) When the near-term contract matures, we
198
HEDGING STRATEGIES
199
Jet fuel futures do not exist in the United States, but firms sometimes hedge jet fuel
with crude oil futures along with futures for related petroleum products. 1O In order to
perform this hedge, it is necessary to understand the relationship between crude oil and
jet fuel prices. If we own a quantity ofjet fuel and hedge by holding H crude oil futures
contracts, our mark-to-market profit depends on the change in the jet fuel price and the
change in the futures price:
(6.19)
where PI is the price of jet fuel and FI the crude oil futures price. We
estimate H
by regressing the change in the jet fuel price (denominated in cents per gallon) on the
change in the crude futures price (denominated in dollar per barrel). Doing so using
daily data for January 2000-June 2004 gives (standard errors are in parentheses)
PI -
PI_I
= 0.009
+ 2.037(F
I
(0.069)
(0.094)
FI _ I )
R2
= 0.287
(6.20)
The futures price used in this regression is the price of the current near-term contract.
The coefficient on the futures price change tells us that, on average, when the crude
futures price increases by $1, a gallon of jet fuel increases by $0.02. Suppose that as
part of a particular crack spread, 1 gallon of crude oil is used to produce I gallon of jet
fuel. Then, other things equal, since there are 42 gallons in a barrel, a $1 increase in the
price of a barrel of oil will generate a $1I42 = $0.0238 increase in the price of jet fuel.
This is approximately the regression coefficient. II
The R 2 in equation (6.19) is 0.287, which implies a correlation coefficient of about
0.50. The hedge would therefore have considerable residual risk.
Weather Derivatives
reestablish the stack hedge by going long contracts in the new near month. This process
of stacking futures contracts in the near-term contract and rolling over into the new
near-term contract is called a stack and roll. If the new near-term futures price is below
the expiring near-term price (i.e., there is backwardation), rolling is profitable.
Why would anyone use a stack hedge? There are at least two reasons. First, there is
often more trading volume and liquidity in near-term contracts. With many commodities,
bid-ask spreads widen with maturity. Thus, a stack hedge may have lower transaction
costs than a strip hedge. Second, the manager may wish to speculate on the shape of
the forward curve. You might decide that the forward curve looks unusually steep in the
early months. If you undertake a stack hedge and the forward curve then flattens, you
will have locked in all your oil at the relatively cheap near-term price, and implicitly
made gains from not having locked in the relatively high strip prices. However, if the
curve becomes steeper, it is possible to lose.
The box above recounts the story of MetallgesellschaftA. (MG), in which MG's
large losses on a hedged position might have been caused, atleast in part, by the use of
a stock hedge.
IOFor example, Southwest Airlines reportedly used a combination of crude oil and heating oil futures
to hedge jet fuel. See Melanie Trottman, "Southwest Airline's Big Fuel-Hedging Call Is Paying Off,"
Wall Street JOlll71al, January 16,2004, p. B4.
11 Recall that in Section 5.5 we estimated a hedge ratio for stocks using a regression based on percentage
changes. In that case, we had an economic reason (an asset pricing model) to believe that there was
a stable relationship based upon rates of return. With crude and jet fuel, crude is used to produce jet
fuel, so it makes sense that dollar changes in the price of crude would be related to dollar changes in
the price of jet fuel.
200
PROBLEMS
There are many other examples of weather risk: ski resorts are harmed by warm
winters, soft drink manufacturers are harmed by a cold spring, summer, or fall, and
makers of lawn sprinklers are harmed by wet summers. In all of these cases, firms could
hedge their risk using weather derivatives-contracts that make payments based upon
realized characteristics of weather-to cross-hedge their specific risk.
The payoffs for weather derivatives are based on weather-related measurements.
An example of a weather contract is the degree-day index futures contract traded at the
Chicago Mercantile Exchange. A heating degree-day is the maximum of zero and the
difference between the average daily temperature and 65 degrees Fahrenheit. A cooling
is the maximum of the difference between the average daily temperature and
65 degrees Fahrenheit, and zero. Sixty-five degrees is a moderate temperature. At higher
temperatures, air conditioners may be used, while at lower temperatures, heating may be
used. A monthly degree-day index is constructed by adding the daily degree-days over
the month. The futures contract then settles based on the cumulative heating or cooling
degree-days (the two are separate contracts) over the course of a month. The size of the
contract is $100 times the degree-day index. As of September 2004, degree-day index
contracts were available for over 20 cities in the United States, Europe, and Japan. There
are also puts and calls on these futures.
With city-specific degree-day index contracts, it is possible to create and hedge
payoffs based on average temperatures, or using options, based on ranges of average
temperatures. If Minneapolis is unusually cold but the rest of the country is normal,
the cooling degree-day contract for Minneapolis will make a large payment that will
compensate the holder for the increased consumption of energy. Notice that in this
scenario a natural gas price contract (for example) would not provide a sufficient hedge,
since unusual cold in Minneapolis alone would not have much effect on national energy
prices.
CHAPTER SUMMARY
At a general level, commodity forward prices can be described by the same formula as
financial forward prices:
FO.T
(6.21)
For financial assets, is the dividend yield. For commodities, is the commodity lease
rate-the return that makes an investor willing to buy and then lend a commodity. Thus,
for the commodity owner who lends the commodity, it is like a dividend. From the
commodity borrower's perspective, it is the cost of borrowing the commodity. As with
financial forwards, commodity forward prices are biased predictors of the future spot
price when the commodity return contains a risk premium.
While the dividend yield for a financial asset can typically be observed directly,
the lease rate for a commodity can typically be estimated only by obseJ11ing the forward
price. The forward curve provides important information
the commodity.
Commodities are complex because every commodity market differs in the details.
Forward curves for different commodities reflect different properties of storability, stor-
201
age costs, production, and demand. Electricity, gold, corn, natural gas, and oil all have
distinct forward curves, reflecting the different characteristics of their physical markets.
These idiosyncracies will be reflected in the commodity lease rate. When there are
seasonalities in either the demand or supply of a commodity, the commodity will be
stored (assuming this is physically feasible), and the forward curve for the commodity
will reflect storage costs. Some holders of a commodity receive benefits from physical
ownership. This benefit is called the commodity's
yield. The convenience
yield creates different returns to ownership for different investors, and mayor may not
be reflected in the forward price. The convenience yield can lead to no-arbitrage regions
rather than a no-arbitrage price. It can also be costly to short-sell commodities with a
significant convenience yield.
FURTHER READING
We will see in later chapters that the concept of a lease rate-which is a generalization
of a dividend yield-helps to unify the pricing of swaps (Chapter 8), options (Chapter
10), and commodity-linked notes (Chapter 15). One particularly interesting application
of the lease rate arises in the discussion of real options in Chapter 17. We will see
there that if an extractable commodity (such as oil or gold) has a zero lease rate, it will
never be extracted. Thus, the lease rate is linked in an important way with production
decisions.
A useful resource for learning more about commodities is the Chicago Board of
Trade (1998). The Web sites of the various exchanges (e.g., NYMEX and the CBOT)
are also useful resources, with information about particular commodities and trading and
hedging strategies.
Siegel and Siegel (1990) provide a detailed discussion of many commodity futures. There are numerous papers on commodities. Bodie and Rosansky (1980) and
Gorton and Rouwenhorst (2004) examine the risk and return of commodities as an inBrennan (1991), Pindyck (1993b), and Pindyck (1994) examine the behavior'
of commodity prices. Schwartz (1997) compares the performance of different models
of commodity price behavior. Jarrow and Oldfield (1981) discuss the effect of storage
costs on pricing, and Routledge et al. (2000) present a theoretical model of commodity
forward curves.
Finally, Metallgesellschaft engendered a spirited debate. Papers written about that
episode include Culp and Miller (1995), Edwards and Canter (1995), and Mello and
Parsons (1995).
PROBLEMS
6.1. The spot price of a widget is $70.00 per unit. Forward prices for 3,6,9, and 12
months are $70.70, $71.41, $72.13, and $72.86. Assuming a 5% continuously
compounded annual risk-free rate, what are the annualized lease rates for each
maturity? Is this an example of contango or backwardation?
202
PROBLEMS
COMMODITY FORWARDS AND FUTURES
6.2. The current price of oil is $32.00 per barrel. Forward prices for 3, 6, 9, and 12
months are $31.37, $30.75, $30.14, and $29.54. Assuming a 2% continuously
compounded annual risk-free rate, what is the annualized lease rate for each maturity? Is this an example of contango or backwardation?
6.3. Given a continuously compounded risk-free rate of 3% annually, at what lease
rate will forward prices equal the current commodity price? (Recall the pencil
example in Section 6.4.) If the lease rate were 3.5%, would there be contango or
backwardation?
6.4. Suppose that pencils cost $0.20 today and the continuously compounded lease
rate for pencils is 5%. The continuously compounded interest rate is 10%. The
pencil price in 1 year is uncertain and pencils can be stored costlessly.
a. If you short-sell a pencil for 1 year, what payment do you have to make
to the pencil lender? Would it make sense for a financial investor to store
pencils in equilibrium?
b. Show that the equilibrium forward price is $0.2103.
c. Explain what ranges of forward prices are ruled out by arbitrage in the
four cases where pencils can and cannot be short-sold and can and cannot
be loaned.
6.5. Suppose the gold spot price is $300/oz., the I-year forward price is 310.686, and
the continuously compounded-risk-free rate is 5%.
a. What is the lease rate?
b. What is the return on a cash-and-carry in which gold is not loaned?
For the next three problems, assume that the continuously compounded interest rate is
6% and the storage cost of widgets is $0.03 quarterly (payable at the end of the quarter).
for widgets:
Here is the forward price
2004
Dec
3.000
6.6.
Mar
3.075
Jun
3.152
2005
Sep
2.750
2006
Dec
2.822
Mar
2.894
Jun
2.968
a. What are some possible explanations for the shape of this forward curve?
sible?
c. What annualized rate of return do you earn on a cash-and-carry entered
into in December 2004 and closed in September 2005? Is your answer
sensible?
6.7.
203
6.8.
a. Suppose the March 2005 forward price were $3.10. Describe two different transactions you could use to undertake arbitrage.
b. Suppose the September 2005 forward price fell to $2.70 and subsequent
forward prices fell in such a way that there is no arbitrage from September
2005 and going forward. Is there an arbitrage you could undertake using
forward contracts from June 2005 and earlier? Why or why not?
6.9. Consider Example 6.1. Suppose the February forward price had been $2.80. What
would the arbitrage be? Suppose it had been $2.65. What would the arbitrage be?
In each case, specify the transactions and resulting cash flows in both November
and February. What are you assuming about the convenience yield?
6.10. Using Table 6.10, what is your best guess about the current price of gold per
ounce?
6.11. Suppose you know nothing about widgets. You are going to approach a widget
merchant to borrow one in order to short-sell it. (That is, you will take physical
possession of the widget, sell it, and return a widget at time T.) Before you ring
the doorbell, you want to make a judgment about what you think is a reasonable
lease rate for the widget. Think about the following possible scenarios.
a. Suppose that widgets do not deteriorate over time, are costless to store,
and are always produced, although production quantity can be varied.
Demand is constant over time. Knowing nothing else, what lease rate
might you face?
b. Suppose everything is the same as in (a) except that demand for widgets
varies seasonally.
c. Suppose everything is the same as in (a) except that demand for widgets
varies seasonally and the rate of production cannot be adjusted. Consider
how seasonality and the horizon of your short-sale interact with the lease
rate.
d. Suppose everything is the same as in (a) except that demand is constant
over time and production is seasonal. Consider how production seasonality and the horizon of your short-sale interact with the lease rate.
e. Suppose that widgets cannot be stored. How does this affect your answers
to the previous questions?