Sie sind auf Seite 1von 6

Socro-Econ. Plan .Sci Vol. 19, No. I, PP. 35-40, 1985 GU384121/85 $3.00 + .

oO
Printed in the U.S.A. 0 1985 Pergamon Press Ltd.

MODELING OIL REFINERY OPERATIONS

NOEL D. URIt
Division of Antitrust, Bureau of Economics, Federal Trade Commission,
Washington, DC 20580, U.S.A.

(Received 29 May 1984)

Abstract-This paper develops a model to analyze the relative responsiveness of refined petroleum
product output to changes in the relative prices of these products. A scheme is devised and implemented
that shares out total production into motor gasoline, distillate fuel oil, kerojet fuel, and other refined
products. A multinomial logit specification is used whereby the share of each of these products is a
function of relative prices, a seasonal factor, and the relative amount of crude output to total product
output. Finally, the structural stability of the estimated relationships are tested for and the null hypothesis
of stability is not rejected.

INTRODUCIION products, which in turn are determined by the inter-


Modeling the activities of petroleum refineries with action of the demand for the products and costs (i.e.
regard to altering the mix of refined petroleum product marginal costs). Note that one cannot properly speak
output has proven to be a most elusive proposition. of supply in the neoclassical microeconomic sense,
The structural and data problems inherent in such since supply exists only in that nonexistent theoretical
analysis are difficult to resolve. It is imperative, world of perfect competition.
however, that a framework be developed to enable Consider the operation of a petroleum refinery
such modeling efforts because of the overall impor- from a simplified perspective. A refinery produces
tance of petroleum in a balanced energy network. output, as noted above, in the form of motor gasoline,
For example, refined petroleum product consumption distillate fuel oil (i.e. home heating oil), residual fuel
accounted for 42.6% of the total energy consumption oil, kerosene, kerojet fuel, etc. The typical refinery
in 1983 in the United States. can use a number of different types of crude oil. For
Refinery operation is characterized by a phenom- example, a typical refinery might be able to handle
enon of being able to alter the relative amounts of both light sweet and light sour$ crude that comes
output of refined petroleum products (e.g. motor from, say, West Texas and Saudi Arabia [3].
gasoline, distillate fuel oil, residual fuel oil, kerosene, It should be observed that the design of any specific
propane, and kerojet fuel) given a barrel of crude oil refinery will depend upon, among other things, the
input, within technical limitations. There exists the specific characteristics of the crude oil to be processed
possibility of output substitution in the refining of and the output mix desired, within limits. There are
crude petroleum. In 1983, for example, the share of different grades of crude oil and, as a result, there
total output attributable to each of the various refined will be differences in the yield obtained from them.
petroleum products continuously changed on a A light crude, for example, will yield a high percentage
month-to-month basis [I]. Thus, in PADD Region of useful products by distillation, whereas a heavy
III (this is defined below), for example, motor gasoline crude will produce significant amounts of crude-oil
accounted for between 40.2 and 45.5% of total 1983 material that must be sold as heavy fuel oil or else
refined petroleum product output. processed further. As a result, there is some flexibility
Large output substitution elasticities of refined in switching a refinery to alternative crude streams,
petroleum products are basic to the question of although this is limited [4].
defining the extent of the market for petroleum In the refining process, some of the available
products. That is, if refinery output shifted significantly crudes must be refined because of long-term minimum
between, say, kerosene and distillate fuel in response volume commitments or because of requirements for
to changing relative prices, this would suggest that specialty products. These crudes are considered fixed,
the market definition would be much broader than and yield constant motor gasoline and distillate fuel
if no such shift was observed. See Stigler and Sherwin oil volumes. From the remaining crudes, and from
[2] for more on this. those crudes which are available in volumes greater
than their minimum volume commitment, are se-
THE ECONOMIC ARGUMENT lected those which can supply the required refined
Given the unregulated nature of the market for products most economically. These are nominally
petroleum products in the United States, the quantity referred to as the incremental crudes. The refinery
supplied is a function of the prices of the various problem becomes one of determining the minimum
incremental cost of, say, distillate fuel oil as a function
of incremental distillate fuel oil production, while

t The views expressed are those of the author and do not


necessarily represent the policies of the Federal Trade Com-
mission or the views of other Federal Trade Commission $Sour crude contains significant amounts of sulfur and
staff members. refineries must be capable of moving the sulfur.

35
36 N. D. URI

maintaining motor gasoline production and general of the explanatory variables are associated with in-
refinery operation at a fixed level. The mechanics of creases (or decreases) in the value of the share. These
solving this type of problem take us beyond the scope requirements suggest that use of a cumulative function
of the current concern. (The interested reader is whose upper bound is one and whose lower bound
referred to Garvin et al. [5] for its mathematical is zero will provide a suitable transformation.
solution.) The element that needs to be highlighted, There are a couple of such transformations that
however, is that changing relative prices of refined would prove acceptable [8]. The one adopted here is
products that precipitate the desire to alter the output the logit specification because it has proven to be the
mix can de facto change the relative shares of total most mathematically tractable.
output from a refinery. The logit model specification is based on the
Given the technical flexibility that exists to change function
the relative output share of refined petroleum prod-
ucts, the question to be empirically examined is &I = l/( 1 + exp (ai + C bifl,rl,,)) (1)
whether refineries actually do alter their behavior in I
response to changing relative prices and, if they do,
what is the extent of this response. where S,, denotes the share of a given refined petro-
leum product produced in period t; x,, are the exog-
AN ANALYTICAL FRAMEWORK enous factors causing variations in output shares i in
There are only a limited number of ways of period t; and ai, b,, , . . . , bin are coefficients to be
addressing the question of refinery output response estimated.
to changing relative prices of the various refined The specification provided in eqn (1) can be esti-
products. One approach is to use a linear program- mated directly by means of nonlinear least squares
ming framework along the lines, say, of the Refinery or, after a suitable transformation of the dependent
Evaluation Modeling System (REMS) of the Depart- variables S,,, by means of classical least squares. Such
ment of Energy [6]. This structure assumes that estimation techniques, however, would not use all of
demands for the various types of refined petroleum the information efficiently. Namely, the sum of the
products are given (i.e. quantities demanded are shares is equal to one given that the available choices
exogenously determined). Subsequently, the quantities are mutually exclusive (i.e. one gets motor gasoline
demanded and an aggregate petroleum price deter- or distillate fuel oil, not both).
mined. This allows one to relate price and output In what follows only four refined petroleum prod-
and hence to determine the relative responsiveness ucts are going to be considered. Consequently, when
of output to variations in price. This approach does limiting the logit model to the four-choice case, one
not really give any insights into the question at hand, has
however, since demand is assumed to be perfectly
inelastic in such a setting (which has been repeatedly
log $ = a2, + C b,x,
shown not to be the case [7]). This would suggest 0 I 1
that the level of output is determined solely by
demand while price is determined solely by the
position of the cost curves. This is too unrealistic.
An alternative approach in analyzing refinery out-
put flexibility is to use an econometric approach. To
date, unfortunately, this does not appear to have log : = ad1 + C b,xj
0 I I
been tried. An appropriate structure will be suggested
below.
A major issue that needs resolution in considering where S, denotes the share of all refinery output
a way to analyze refined petroleum product output besides motor gasoline, distillate fuel oil, and kerojet
is how to insure that the components of refined fuel, S2 denotes the share of motor gasoline output,
petroleum product output are exactly equal to total S3 denotes the share of distillate fuel oil output, and
output. That is, if we are examining changes in, say, S, denotes the share of kerojet fuel output; a*,, a3,,
relative prices, and total output is predetermined, a4,, bll, . . , bdh are coefficients to be estimated,
how can we be certain that the aggregation across the and x, are the exogenous factors causing variation in
components precisely equals this total. To address the relative outputs of the refined products. The
this apparent anomaly, a scheme must be designed subscript t, designating observations in each period,
to share out total output based on some objective has been deleted to simplify the notation. That this
criteria. These shares as a proportion of total output is tantamount to the logit specification can easily be
will necessarily lie between zero and one. To guarantee seen by solving each of the equations as a simple
that aggregation holds, it is only necessary to require function of one specific share or by referring to
that these shares across all refined petroleum products McFadden [9].
sum to one. Each of these equations presumes that the logarithm
Given that each of the shares is constrained by the of the ratio of the share of one refined petroleum
interval (0, l), a linear model specification is not product to one minus the share of that product
acceptable, due to the possibility that values may lie relative to a similar ratio for another refined petroleum
outside of it. The obvious solution is to have a product is a linear function of the set of explanatory
transformation whereby for all possible values of the factors. These values are dependent on the values
explanatory variables, the share values will be in the associated with the remaining equations only to the
(0, 1) interval. One would also like the transformation extent that the system must be constrained so that
to maintain the property that increases in the value the sum of the individual shares is one.
Modeling oil refinery operations 37

One is now in a position to estimate the parameters (3) Technical characteristics. As discussed above,
~21, ~31, ~41, b2,, . . . , b4,,. If the set of explanatory not all crude oils yield the same output mix, nor the
factors (i.e. xj) in each equation are not identical, same volume of total output of refined products. As
efficiency can be gained by using the iterative seem- crude input increases, given that there is a quality
ingly unrelated technique of Zellner [lo]. This is just change in the crude and hence more or less refined
a generalized least squares estimation procedure used products produced, relative output shares will change.
to reflect the correlation among the error terms This affect is reflected by including a variable defined
associated with each equation in the multinomial as the ratio of crude input to total refined product
model. The iterative Zellner technique is used here. output. These basic data were taken from the Petro-
leum Supply Monthly.
DATA
Other variables were considered in preliminary
Before proceeding to a presentation of the estima- analyses and proved to be statistically insignificant.
tion results, a discussion of the data is in order. The In particular, a weather variable (measured by heating
data employed in the estimation are monthly obser- degree days), the relative price of imported refined
vations over the period 1981-1983. The year 1981 petroleum products, and electric utility generation
was used as the line of delineation since that is when (residual fuel oil is a significant boiler fuel) all failed
decontrol became effective, ostensibly permitting the to demonstrate any impact on the output shares of
market mechanism to work, at least as far as refined the products.
petroleum products are concerned. What one is left with now is a mechanism for
Monthly refined petroleum product data were ob- analyzing refined petroleum product output shares
tained from the Department of Energy’s Petroleum based on relative prices, technical refinery character-
Supply Monthly [ 1 I]. Given the limited time period istics, and crude oil input (as a proxy for seasonal
for which data are available, practical econometric variation). Empirical estimates of the impact of these
consideration suggested that looking at a large number factors on changing shares is the subject of the
of individual refined petroleum products (data are following section.
collected on 18 separate items by the Department of
Energy) would present problems. Consequently, the EMPIRICAL RESULTS
investigation was limited to just four products: motor
Using the data and the estimating technique pre-
gasoline, distillate fuel oil (these two combined ac-
viously discussed, the empirical results were obtained.
count for about 65% of refinery output), kerosene-jet
Preliminary analyses suggested that the imposition of
fuel type (which accounts for about 6% of refinery
output) and all other refined petroleum products. symmetry constraints might be appropriate (that is,
The output data were collected on a Petroleum
b, = b,, on the relative price terms). To test this
hypothesis, a Quandt test is employed. The test
Administration for Defense District (PADD) level
consists of computing the determinants of the unre-
(see the Monthly Energy Review [ 1] for a character-
ization of the five PADD regions). This is the most stricted and restricted disturbance covariance matrices.
disaggregated basis publicly available for these data. The logarithm of this times minus the number of
Regionally disaggregated data were selected over na- observations used in the estimation has a chi-square
tional aggregates since there are some differences in distribution with degrees of freedom equal to the
refinery vintages by region and some differences in number of independent restrictions being imposed
the technical characteristics of the refineries in the [ 12l.t The test was performed for each of the regions
various regions. considered with the null hypothesis being that sym-
The set of exogenous variables used in the esti- metry holds. In all instances, the null hypothesis
mation consisted of the following: could not be rejected. That is, it is not possible to
statistically reject the contention that a relative in-
(1) Prices. The price of motor gasoline relative to crease in the price of refined petroleum product j will
the price of all other refined petroleum products (i.e. affect the logarithm of the relative product share i to
of motor gasoline, distillate fuel oil, and kerojet fuel), the same extent that a relative price increase in the
the price of distillate fuel oil relative to the price of price of refined petroleum product i will affect the
all other refined petroleum products, and the price logarithm of the relative product share j.
of kerojet fuel relative to the price of all unleaded The estimation results for each of the PADD
refined petroleum products. The price of other refined regions are given in Tables l-5. The standard errors
products is simply measured as a weighted average. of the estimates are in parentheses. All of the reported
All of the price data were taken from Platt ‘s Oil Price estimated coefficients are statistically significant at
Handbook [ 121. the 95% level or better. The Durbin-Watson statistics
(2) Seasonal variation. To capture the seasonal are either inconclusive or suggest the absence of serial
variations in the output shares, crude oil input into correlation.
the refining process is used. This variable really is a In interpreting the results, note that they relate to
proxy for changing refined product requirements (i.e. the change in the share of refined petroleum product
more motor gasoline in the summer and more distil- output of either motor gasoline, distillate fuel oil, or
late fuel oil in the winter) over the year. The data
are taken from the Petroleum Supply Monthly. Note
that this method of reflecting seasonal variation was
t This result is valid only for maximum likelihood esti-
adopted in difference to using dummy variables since mates. Kmenta and Gilbert [ 141, however, have shown that
the latter method would use up eleven degrees of iteration of the Zellner estimation procedure until conver-
freedom. This was judged unacceptable given the gence (as is done here) results in maximum likelihood
limited number of observations to begin with. estimates.
38 N. D. URI

Table I. PADD 1 refinery share equation estimates”)

Explanatory Variables c2)


~2(3) D. W.c4)
Equation Constant Pmg’Po pd’po pkj’po Crude/Total Output Crude

log ( s2 ) -6.9229 1.5933 (5) (5) 3.8799 0.0167 0.6722 1.57


- (0.9177) (0.7551) (1.2328) (0.0014)
Sl
log ( s3 ) 0.1719 (5) 0.5111 0.3310 -0.7564 0.0171 0.7154 1.75
(0.0308) (0.2155) (0.1518) (0.3138) (0.0028)
G

log ( s4 ) -2.4013 (5) 0.3310 1.4729 -0.0404 (5) 0.7078 1.79


(0.7120) 0.1518) (0.6449) (o.ooao)
G

(1) Standard errors of estimates in parentheses.


(2) Pk. : price of kerojet, Pw : price of motor gasoline, P : price of other nfined petroleum products,
crdde : crude oil input to refinery, and total output = ?otal refinery petroleum output.
(3) Coefficient of determination.
(4) Durbin-Watson statistic.
(5) Not statistically significant.

kerojet fuel relative to all other refined petroleum kerosene from other uses (e.g. kerosene blended with
products. It is consistently the case that month-to- heavier gas oil to produce distillate fuel oil). Another
month variations in the relative prices of the own way is for refiners to technically adjust the refining
fuels are significant in explaining relative variations process by changing the boiling point ranges on the
in the relative amounts produced of the various distillation towers to increase or decrease the per-
refined petroleum products. Of equal or more signif- centage of crude oil extracted as kerosene (at the
icance is the relative shift between refined products expense of other light refined products such as either
in response to changing relative fuel prices. Thus, for motor gasoline or distillate fuel oil). Finally, it is
example, in region PADD 3, an increase in the possible to change crude oil types to those with a
relative price of kerojet fuel will result in a reduction higher percentage of kerosene or for which the kero-
in the relative share of distillate fuel oil output. sene is kerojet quality.
In this latter regard, just what is the mechanism There are two other inferences to be drawn from
through which this substitution can take place? Take these results. First, as the ratio of crude input to total
kerojet fuel as an illustration. Kerojet production can output increases, there is a general indication that
be manipulated in the refining process via a number the relative share of motor gasoline increases at the
of alternatives. One way is to divert kerojet quality expenses of distillate fuel oil and kerojet fuel. This is

Table 2. PADD 2 refinery share equation estimates”’

Explanrrtory Variables (2)


g2(3) D. W.(‘I)
Equation constant Pmg’Po pd’po pkj’po Crude/Total Output Crude

log ( S2 ) -6.1965 0.9005 -0.4051 1.1442 3.7031 0.0104 0.6524 1.62


-
(2.4393) (0.4589) (0.2023) (0.5811) (1.921) (0.0044)
Sl

log ( “3 1 0.3676 -0.4051 0.5119 (5) -0.4153 0.0129 0.6937 1.99


(0.0526) (0.2033) (0.1420) (0.1700) (0.0063)
Sl

log ( S4 ) -2.3297 1.1442 (5) 1.7197 -0.0397 (5) 0.6845 2.20


(0.7770) (0.5811) (0.4994) (0.0075)
<

(1) Standard errors of estimates in parentheses.


(2) Pk. = price of kerojet, Pw = price of motor gasoline, P : price of other refined petrolem products,
CFa de = crude oil input to refinery, and total output : ?otal refinery petroletan output.

(3) Coefficient of determination.


(4) Lurbin-Watson statistic.
(5) Not statistically significant.
Modeling oil refinery operations 39

Table 3. PADD 3 refinery share equation estimates”’

Eicplanatory Variables (*I


$(3) D. W.(4)
Equation constant Pmg'Po pd'po pkj'po Crude/Total Output Crude

log (2 ) -2.5924 0.8130 -1.5168 -1.5237 1.6474 3.0163 0.7862 1.55


(0.6233) (0.3847) (0.5847) (0.4640) (0.6235) (0.0059)
s1

log ( S3 ) 0.4086 -1.5168 0.7838 -0.3944 -0.9271 0.0476 0.8115 1.91


(0.0732) (0.5847) (0.3734) (0.1510) (0.4298) (0.0059)
G

log ( S4 ) -2.9126 -1.5237 -0.3944 1.9371 -0.6713 (5) 0.7527 1.93


- (0.9373) (0.4640) (0.1510) (0.6678) (0.0987)
Sl

(1) Standard errors of estimates in parentheses.


(2) Pk. z price of kernjet,Pm = price of motor gasoline,P : price of other refined petroleumproducts,
crdde = crude oil input,to refinery,and total output : total refinerypetrolem output.
(3) Coefficientsof determination.
(4) Durbin-Watsonstatistic.
(5) Not statisticallysignificant.

consistent with the results of Copp [ 151. Second, both other refined petroleum share. This latter term is
the relative production of motor gasoline and the equal to one over one plus the sum of the exponentials
relative production of distillate fuel oil are sensitive on the right-hand-side of each of the eqns (2a)-(2c).
to seasonal variations. This is hardly a surprising Needless to say, such an expression is quite complex
result [ 161. and not subject to easy interpretation. Thus, share
A final useful result would be obtained if estimates elasticities are not readily available.
of the actual share elasticities were available. This
would provide information on the percentage changes TESTING FOR STABILITY
in the various refined petroleum product shares in A final concern in the analysis is with the structural
response to changes in the various explanatory factors. stability of the estimated relationships in the sense
Unfortunately, as specified, there are no ready mach- that the estimated coefficients on the explanatory
inations available to allow for deriving such estimates variables have remained constant over time. The test
in an easily interpretable form. Thus, for example, is carried out using the test statistic reported in Uri
to obtain a refined petroleum product share for motor [ 171, where the test statistic equals the difference
gasoline, one needs to take the exponential of the between the sum of squared residuals of the entire
right-hand-side of eqn (2b) and multiply it by the all sample less the cumulative sum of the squared resid-

Table 4. PADD 4 refinery share equation estimates”’

ExplanatoryVariables (2)
$(3) D. W.c4)
Equation Constant Plllg’PO 'd"o pkj'po Crude/TotalOutput Crude

log ( s2 ) -3.0426 0.9328 (5) -1.3212 1.8353 0.0428 0.6151 1.87


(0.4320) (0.51731 (0.9255) (0.0206)
<

log ("3 I 0.1278 (5) 0.3158 (5) -0.6352 -0.0397 0.5924 1.97
(0.0423) (0.1356) (0.2206) (0.0129)
%

log ( s4 ) -1.9122 -1.3212 (5) 0.5697 -0.3606 (5) 0.6544 1.96


(0.3812) (0.5173) (0.1951) (0.1784)
s1

(1) Standarderrom of estimatesin parentheses.


(2) Pk. = price of kerojet, Pw = price of motor gasoline,P = price of other refined petroleumproducts,
crdde : crude oil input to refinery,and total output : ?otal refinerypetrolem output.
(3) Coefficientof determination.
(4) Durbin-Watsonstatistic.
(5) Not statisticallysignificant.
40 N. D. URI

Table 5. PADD 5 refinery share equation estimates”)

ExplanatoryVariables 12f
~2(3) D. W.t4)
Equation Constant Pnlg'Po pd'po pkj'po Crude/TotalOutput Crude

log csz ) -2.5749 1.0646 (5) -0.9337 0.1525 0.0218 0.6982 1.82
(a.83841 (0.3702) (0.4105) (0.0716) f0.0117)
s1

log (s3 1 0.2701 (5) 0.3465 -0.1447 0.3466 0.0556 0.7599 2.14
(0.0548) (0.1528) (0.0625) (0.1528) (0.0052)
Sl

log (2 ) -2.7998 0.9337 -0.1447 0.9455 -0.2381 (5) 0.7505 2.05


(0.7646) (0.4105) (0.0625) (0.3023) 10.0761)
$1

(11 Standarderrom of estimatesin parentheses.


(2) - price of motor gasoline,P : price of other refined petroleumproducts,
"J = price of kerojet, P
cr de = crude oil input t?&finery, and total output = ?otal refinerypetroleumoutput.
(3) Coefficientof determination.

(4) Durbin-Watsonstatistic.

151 Not statisticallysignificant.

uals over the non-overlapping segments divided by tested for and the null hypothesis of stability is not
the cumulative sum of squared residuals of the non- rejected.
overlapping segments. The null hypothesis that a
regression relationship is constant over time implies REFERENCES
that the value of the test statistic is distributed as F. 1. Department of Energy, Monthly Energ_v Rev., U.S.
The data upon which the equations were estimated Government Printing Office, Washington, DC., monthly.
2. G. Stigier and R. A. Sherwin, The extent of the market.
were divided into three equal length intervals. Since
Unpublished manuscript (1983).
they do not provide much information in and of 3. R. 0. Jones, Vertical integration, cartel coordination
themselves, the tabulated value of the test statistic and the petroleum industry. Unpublished Ph.D. disser-
for each equation in each region is not reported. tation, University of Maryland (1983).
None of the estimated relationships, however, dem- 4. D. Teece, Vertical integration in the U.S. oil industry.
onstrates any instability over the period 198 i-1983. In Verticaf Integration in the Oil Industry (Edited by
The implications of these results in estimating the E. J. Mitchell), The American Enterprise Institute,
response of the relative refined petroleum product Washington, D.C. (1976).
shares is clear. Events since i 98 1 have left virtually 5. W. W. Garvin, H. W. Crandall, J. B. John and R. A.
Spellman, Application of linear pro~mming in the oil
unchanged the responsiveness of the relative shares
industry. Mgmt. Sci. 3, 407-430 (I 957).
to changes in relative prices, crude input, etc. That 6. Department of Energy, Annual Energ_p Outlook, U.S.
is, the relative importance of the various explanatory Government Printing Office, Washington, DC. (1984).
factors has not changed. 7. N. D. Uri, The Demand ,jbr Energy and Conservation
One must be careful, however, to avoid inferring in fhe United States, JAI Press, Greenwich, Connecticut
that relative shares have not changed. The estimation ( 1983).
results clearly demonstrate that, say, the relative price 8. R. Pindyck and D. Rubinfeld, Econometric Models and
of motor gasoline to other refined petroleum products Methods. McGraw-Hill, New York (1976).
influences the relative share of motor gasoline pro- 9. D. McFadden, Multinomial models. In Frontiers in
Econometrics (Edited by P. Zarembka). Academic Press.
duced. The magnitude of the associated responses in
New York ( 1974).
the aggregate, however, have remained unaltered over 10. A. Zellner, An efficient method of estimating seemingly
the sample period. unrelated regression and tests for aggregation bias, J. of
Am. Stat. Ass. 57, 348-368 (1962).
CONCLUSION 11. Department of Energy, Petroieum Supply ~~~nt~iy, U.S.
Government PrintingOffice, Washington, D.C., monthly.
The objective of the foregoing analysis has been to 12. R. Pindyck and D. Rubinfeld, P(att’s Oil Price Hand-
develop a model to analyze the relative responsiveness book. McGraw-Hill, New York, annual.
of refined petroleum product output to changes in 13. S. Goldfeld and R. Quandt, Nonlinear Methods in
the relative prices of these products. A scheme is Economeirics. North Holland, Amsterdam (197 1).
devised and implemented that shares out total pro- 14. J. Kmenta and R. Gilbert, Small sample properties of
duction into motor gasoline, distillate fuel oil, kerojet alternative estimaters of seemingly unrelated regression,
J. Am. Stat. Ass. 63, 1180-1200 (1968).
fuel, and the remaining refined products. A multi-
15. E. A. Copp, Regulation Competition in Oil. Texas A
nomial logit specification is used whereby the share and M University Press, College Station (1976).
of each of these products is a function of relative 16. A. Bopp and G. M. Lady, Conservation and U.S.
prices, a seasonal factor, and the relative amount of petroleum consumption. Energy 3, 533-538 (1978).
crude input to total product output. Finally, the 17. N. D. Uri, The demand for energy in Korean industry.
structural stability of the estimated relationships is App. Energy 14, 295-3 16 (1983).

Das könnte Ihnen auch gefallen