Professional Documents
Culture Documents
M. Asali (PhD)
__________________________________________________________________________________
Focusing mainly on inventories held by the demand sector, connection of the structure of oil prices
to commercial inventory emanates from the fact that an upward sloping futures curve is consistent
with an expected future spot price that rewards inventory holders for the cost of carrying
inventories, including marginal warehousing costs, insurance, and the interest foregone on the
capital invested in the inventories. This link between the futures price and the expected future spot
price is known as “cost of carry” arbitrage. The theory of storage [see Kaldor, (1939), Brennan
(1956)] can be stated in terms of basis, the difference between the contemporaneous spot in period
t, St and the futures price for delivery at date T, Ft, T. It views the basis as consisting of the cost of
carry: interest foregone to borrow to buy the commodity, plus the marginal storage cost minus a
“convenient yield”. Economic rationale for this statement is provided by the assertion that the
convenience yield, which is the basis, adjusted for interest charge and storage costs, falls at a
decreasing rate as aggregate inventory rises.
Another view of commodity futures is the theory of “Normal Backwardation”, which compare
futures prices to expected future prices. This theory, [due to Keynes and Hicks (1939) later expended
by Fama and French (1988)] views the futures market as a risk transfer mechanism whereby risk
averse long position holders (investors) earn a risk premium for bearing future spot risk that
commodity producers want to hedge. In modern version of the theory of inventory, [Deaton and
Laroque, (1992), Gorton et al.(2007)] the behavior of spot price, which display high volatility, high
positive skewness and significant kutrtosis change fundamentally when inventories are present. In
this modern version of the theory, both the convenience yield and the risk premium emerge
endogenously as functions of inventory. In this context the futures market provides the inventory
holders with an opportunity to hedge bankruptcy costs. The level of inventories matters for the risk
premium because future spot price volatility is negatively related to the level of inventories. That is
when inventories are low; the variance of the future spot price is higher due to an increased
likelihood of a stock out resulting in the risk-averse investors demanding a higher risk premium. The
main predictions of the theory according to Gorton et al. (2007) are as follows:
Here we will investigate the first two predictions of the theory in a brief empirical analysis. The third
topic is studied in another report and will not be taken here.
2. A brief review of the time series: oil prices and commercial crude stocks
In this section time series of oil prices and OECD commercial crude oil inventory are reviewed, very
briefly, before undertaking the statistical analysis. The data set used in our quantitative analysis
comprised of monthly observations of commercial oil stocks and oil prices, both spot and futures, for
the time period 2002 M02 -2010 M08. The data set is provided by the Data service Department
(DSD) of OPEC. The first graph below denotes historical movements of monthly spot and futures (of
up to three months maturity) prices of the crude oil benchmark WTI, and OECD commercial crude.
This is followed by two graphs that illustrate Hodric-Prescott filtering of commercial inventory of
crude oil and monthly average spot WTI respectively. A couple of observations could be made by
merely by visual inspection of the series. At the upper panel, both series exhibit ascending
movement while fluctuating around some imaginary trends. It can be seen that inventory level had
had more cyclical movements most likely due to the seasonal feature of oil consumption and
storage. We can also see that prior to oil price hike in mid 2008 there has been a dip in OECD crude
stock to recover later and that despite falling and then rising of the level of stock6s since second half
of 2009, oil prices tended to stay in an acceptable range with a quite low ascending slope. In
following graphs the two series
WTI Monthly Spot and Futures Prices and OECD Commercial Crude Stocks
(2002 M01 to 2010 M08)
160
120
80
1.350 40
1.300
0
1.250
1.200
1.150
1.100
02 03 04 05 06 07 08 09 10
W TI Spot
W TI Futures 01
W TI Futures 02
W TI Futures 03
OECD Commercial Crude Stocks
1,300
1,250
80
1,200
40
1,150
0
1,100
-40
-80
02 03 04 05 06 07 08 09 10
120
60 80
40
40
20
0 0
-20
-40
-60
02 03 04 05 06 07 08 09 10
are filtered to decompose the time series to their long term trends and cycle components of the
series. The filtering technique (the Hodrick-Prescott Filter) is based on smoothing the series by
minimizing variance around a moving average value of the time series of concerned. Importance of
these long term trends for our analysis is that for example in case of commercial oil inventory, the
long term trend could be considered as the normal or expected level of storage. This concept plays
an important role in explaining the deviation of actual stock level from its normal trend and its effect
on oil prices. Inspecting the graphs we observe that while the long term trend curve of OECD
commercial stocks still seems to be set to continue its upward trend at the end of our time period of
concerned, oil prices trend exhibits no such tendency. These could be interpreted as the oil market
expectation of oil prices staying more or less at the current range while oil inventory still could keep
growing, albeit very slowly.
One point is worth mentioning at the outset that although in this analysis OECD commercial crude
stock is used, because of availability and reliability of the data on this variable with different
frequencies; however, this should not cause any loss of generality in our analysis due to existence of
a strong correlation between OECD commercial stocks and the world total petroleum inventory, as
can be seen in the graph below.
6.400
Total World Stock (mb)
6.200
6.000
5.800
5.600
5.400
1,100 1,150 1,200 1,250 1,300 1,350
A convenient way to investigate the relation between oil futures curves and crude inventory is to
look at the returns of the futures prices in the time period of concern. Monthly futures return of oil
Ft 1,T Ft ,T
prices is defined as: where t is the month t of the futures contract and T is the
Ft ,T
expiration date of the contract. The expression above proposed for the futures return measures the
returns from the end of month t on the nearest contract whose expiration date T is the end of the
month t+1.Table below contains simple summary statistics for crude oil futures returns for time
period January 2002 till August 2010. Average basis is calculated as the difference between the one
month futures and of F1-F3 benchmark WTI table below provides a summary statistics on the
returns of futures prices of for te time period WTI matured in two months.
Summary of Descriptive Statistics of the Crude Futures Returns (2002 M01 to 2010 M08)
Observations 103
(% per annuam)
Mean
11.1
Median 31.8
Std. Dev. 31.5
Skewness -1.458863 (monthly return distribution)
Kurtosis 6.072590 (monthly return distribution)
Average Futures Basis -8.49
The forth row indicates that the average return on oil futures has been positive (11.1% per annum),
which implies a historical risk premium to the long side of crude oil futures position. At the same
time the last row of the table shows that the sample average (percentage) basis has been negative.
The basis is calculated as 100*{[(F1-F3)/F3]*12} %. This implies that on average, during the time
period 2002 M01 to 2010 M08 futures prices of crude have exceeded contemporaneous spot prices.
Otherwise stated, on average oil futures markets have been in “contango” situation. As the basis is a
close proxy for the “convenience yield” of holding oil inventory, a sample average basis of -8.49%
implies that in the time period under study, the reward to holding oil inventory has been high
enough to compensate the cost of storage. At the same time futures prices returns which is
considered as a proxy for the so called “risk premium” of the expected price risk is estimated 11.1%
that shows from the investor’s point of view the futures prices have been set at a discount to the
expected future spot prices rewarding the long side of the futures positions for providing price
insurance. Observe that the annualized mean of the standard deviation of oil futures prices in this
period has been estimated to be 31.8% which is by far larger than average volatility index of other
non energy commodities including metals and agriculture commodities (Gorton et al. 2007). This
high average volatility requires high risk premium.
The following graphs clearly show that there has been a negative relation between the crude oil
basis (difference between the prompt prices and futures prices) and commercial oil stocks. These
findings are in accordance with the theory’s predictions as stated above.
Contango structure of crude oil futures prices
OECD Commerecial Crude stok and WTI Basis:
(2002 M01- 2010 M08) (2002 M01-2010 M08)
4 4
2
2 0
-2
0
WTI F1-F3 Basis
-4
1.350 -6
-2 -8
1.300
1.250
-4
1.200
-6 1.150
1.100
-8 02 03 04 05 06 07 08 09 10
1,100 1,150 1,200 1,250 1,300 1,350
OECD Commercial Crude Storage
OECD Commercial Crude Stock WTI Basis F1-F3
As can be seen from the graph on the right, generally speaking, behavior of the basis during the time
period 2002 M01- 2010 M08 has been almost mirror image to the OECD commercial crude oil stocks.
An upward movement by crude stock is matched by a downward trend by the basis. We have found
a long term co-integration relation between the two variables as follows:
WTI .Basis 0.056 * OECD.Crude.Stock 0.037 * t
( S .D.).............(0.007)................................(.011)
As the unit of the OECD commercial crude stock in our equation is million barrel of crude, these
results mean that an increase of one million barrel crude to OECD stocks would have caused a 5.6%
decrease in WTI basis. We also find that there has been two ways (Granger) causality running
between the two variables meaning that adding observations on OECD commercial crude stock, for
example, would have increased forecasting capability of our model regarding future changes in WTI.
The (Granger causality test results are reported below:
OECD Commercial Crude Stock does not Granger Cause WTI Basis F1-F3 99 4.89832 0.0033
WTI Basis F1-F3 does not Granger Cause OECD Commercial Crude Stock 2.82428 0.0430
On the basis of these results one cannot reject the (Granger) causality of the two variables on each
other on 5% level; however, it appears that in the time period under study, probability of this
causality effect has been higher from crude stock to WTI Basis compared to the other way round.
We noticed that according to the modern theory of commodity storage we should expect a negative
relation between the crude futures returns (as a proxy for risk premium of oil prices) and oil
inventory. The reasoning line for this proposition was that when inventories are high the buffer
function of inventories is increased. In this circumstances the risk of a stock-out decrease which
lowers the volatility of the future spot price. Decreasing price risk would decrease the risk premium.
The scatter plot of WTI futures return and OECD crude stock below is in favor of this prediction.
Although in this case the inverse slop of the regression line between the two variables is not as sharp
as is was in case of the basis and the commercial stocks, however, it denotes that in the
Average WTI Futures Returns Vs. OECD Commercial Crude Stocks WTI Monthly Futures Returns and OECD Commercial Crude Stocks:
(2002 M01-2010 M08) (2002 M01-2010 M08)
0,2 .2
.1
0,1 .0
-.1
WTI Futures Return
0,0
1.350 -.2
-.3
-0,1 1.300
-.4
1.250
-0,2
1.200
-0,3 1.150
1.100
-0,4 02 03 04 05 06 07 08 09 10
1,100 1,150 1,200 1,250 1,300 1,350
OECD Commercial Crude Stock
WTI Returns dlog(F2)
OECD Commercial Crude Stock
time period of concerned (2002 M01-2010 M08), accumulation of crude oil stocks has been reducing
volatility of oil prices and the risk premium of oil prices. We examined other oil stocks relation with
the risk premium as well and the general pattern did not change. For example the scatter plot
between WTI futures return and oil in float even yields slightly steeper downward regression line.
One reason for this could be the rising importance of in float and oil on water stocks in the futures
markets trade in recent years.
0,1
WTI Futures Returns dlog(F2)
0,0
-0,1
-0,2
-0,3
-0,4
40 80 120 160 200 240
OIL in Float
Apart from interactions between each of the variables “basis” and “futures prices return” with
commercial oil stocks, relations between the two financial variables, i.e. “futures return” and “basis”
is also of interest. We have found a positive relation between these two variables as the scatter plot
of WTI basis against WTI futures returns shows in the graph below:
Backwardation
0,1
WTI Futures Returns
0,0
-0,1 Contango
-0,2
-0,3
-0,4
-.15 -.10 -.05 .00 .05 .10
These observations are of interest, because sometimes positive futures basis, also referred to as
“backwardation”, is viewed as a requirement for the existence of a positive risk premium to a long
position in commodity futures markets. Indeed from the scatter plot above it can be seen that in
most cases positive returns have been coincidental with the backwardation situation. As we find a
positive long term co-integration relation between the two variables, reported below, a simple
regression line is fitted to the time series of the variables to examine connection between the risk
premium and the basis. According to these results, on average, one unit (per cent) increase in the
basis could increase the risk premium of holding oil futures contracts as much as 0.50 units (per
cents). May be because a widening differences between the futures and spot prices have been seen
as an indication of increasing oil price risk and hence a higher risk premium. As mentioned earlier in
the time period 2002 M 01-2010 M 08, volatility of oil prices has been higher than most
commodities, justifying a higher risk premium.
It is worth noting that contango situation and positive risk premium are not mutually exclusive and
the scatter plot above indicates that there have been cases of contango situation with positive risk
premium. Observe that the futures basis compares futures prices to contemporaneous spot prices,
while the risk premium is the difference between futures prices and expected future spot prices. In
order to commodities, oil for example, to be stored, futures prices have to exceed contemporaneous
spot prices to compensate inventory holders for the cost of storage. Only when inventory is less than
expected level, spot prices may exceed the futures prices to curb demand and restore the expected
level of inventory. In this situation convenience yield, which is return to a unit of stored commodity
accrued to stock holder, is sufficiently high to increase demand for storage. It seems that in the time
period of concerned, oil futures prices have been set at a discount compared to average expected
future spot prices of oil, rewarding the long positions of the futures for providing price insurance in a
time of high oil price volatility.
In this essay we examined, very briefly, relation between oil inventory and futures prices in the
framework of the storage theory using quantitative techniques. In line with the theory of storage,
one of our main points here has been establishing a negative (and non-linear) relationship between
oil inventory and futures basis of oil prices. The futures basis is a close proxy for the so called
“convenience yield” of holding (oil) inventory which is not directly observable. In addition to this
important parameter we also estimated the average “risk premium” of expected spot oil prices and
average annualized volatility of oil prices in the time period under study 2002 M01-2010 M08.
We argue that a general tendency to expect high futures prices compared to prompt prices, (mainly
due to a global growing demand for oil), on one hand and low opportunity cost of oil storage (low
interest rate, availability of funds for oil stock holders) on the other, has caused a relatively large
average futures basis “convenience yield” encouraging accumulation of oil inventory. At the same
time, since the level of commercial oil stocks, in the time period under study, has rarely exceeded
the expected or normal level of oil inventory and also volatility of oil prices has been fairly high,
therefore, on average the expected future spot prices has been set higher than futures prices by the
investors providing the necessary reward for the long futures positions for price insurance. So on
average we have had “contango” situation with positive “risk premium” of oil prices in most part of
the time period 2002 M01-2010 M08. It seems that so long as a prospect for a growing demand for
oil exists together with a low opportunity cost of oil storage, (reasons for high convenience yield),
and at the same time high volatility of oil prices persist, requiring a large risk premium, the current
contango situation could continue. Accumulations of the commercial oil stocks curb the prices and
futures basis and cause a narrowing contango situation, but unless the actual inventory level is not
considered in excess of the expected level of oil stocks in a significant manner, the current situation
could continue.
It is recommended that in order to monitor the oil markets developments more successfully, actual
level of oil inventory with respect to its expected or normal level together with volatility of oil prices,
average returns of the futures prices (risk premium) and the futures basis (convenience yield) are
measured and estimated continuously along with other fundamental variables of oil markets
analysis.
M. Asali
References
Brennan, M. (1958); “The Supply of Storage”, American Economic Review, 48: 50-72.
Deaton, A. G. Laroque (1992). “On the Behavior of Commodity Prices” Review of Economic Studies
59: 1-23.
Fama, Eugene and Kenneth French (1987), “Commodity Futures Prices: Some Evidence on Forecast
Power, Premiums and theory of Storage, “Journal of Business: 60: 55-73.
Gorton, B. Gary, F. Hayashi, K. G. Rouwenhorst, (2007) “The Fundamentals of Commodity Futures
Returns” http://www.nber.org/papers/w13249.pdf
Hichs, J. R. (1939) “Value and Capital” Cambridge, England, Oxford University Press.
Pindyck, S. R. (2004) “Volatility in Natural Gas and Oil Markets” The Journal of Energy and
Development, Vol.30, N0.1
Kaldor, N. (1939), “Speculation and Economic Stability, “Review of Economic Studies 7: 1-27.
Keynes, J.M. (1930). “A Treatise on Money” Vol.2, London Mac Millan.
WTI Monthly Spot and Futures Prices and OECD Commercial Crude Stocks
(2002 M01 to 2010 M08)
160
120
80
1.350 40
1.300
0
1.250
1.200
1.150
1.100
02 03 04 05 06 07 08 09 10
W TI Spot
W TI Futures 01
W TI Futures 02
W TI Futures 03
OECD Commercial Crude Stocks
1,300
1,250
80
1,200
40
1,150
0
1,100
-40
-80
02 03 04 05 06 07 08 09 10
120
60 80
40
40
20
0 0
-20
-40
-60
02 03 04 05 06 07 08 09 10
0
WTI F1-F3 Basis
-2
-4
-6
-8
1,100 1,150 1,200 1,250 1,300 1,350
Backwardation
0,1
WTI Futures Returns
0,0
-0,1 Contango
-0,2
-0,3
-0,4
-.15 -.10 -.05 .00 .05 .10