opec's pricing policy and the international transmission of oil price effects

OPEC’s pricing policy and the international transmission of oil price effects

Jaime Marquez and Peter Pauly

We endogenize OPEC’s pricing policy recognizing that oil price changes affect the real income of oil importers, and that changes in the real income of oil importers affect oil price changes. We determine real income, international trade flows, and pn’ces in a three-region (DCs, OPEC, non-OPEC LDCs) econometric world model. Applying optimal control theory, we derive optimal oil pricing strategies. We find that not allowing for income feedback effects results in an upward bias in the total price elasticity of oil demand and in the optimal oil price path, neither of which is in OPEC’s best interests.

Keywords: Optimal oil prices; income feedback effects; OPEC

Our purpose is to study OPEC’s pricing policy recogn- nizing that real income of oil importers is affected by oil price changes and that oil price changes are affected by real income of oil importers. In doing this, we relax one of the most common assumptions in the analyses of oil price effects, namely that oil price changes can be considered as exogenously given ‘shocks’ (Marion and Svensson,’ Bruno and Sachs? Findlay and Rodriguez? Schmid4). Price taking behaviour is an assumption that may be valid in the short run, or for countries having a very small share of the oil market. However, for countries like the USA, Japan, and West Germany, which have a large share of the world oil market, the assumption of oil price exogeneity is not tenable given that an increase in oil prices affects their real income, and thus their oil consumption, with a feedback effect on OPEC’s revenues and oil prices.

Jaime Marquez is with the Board of Governors of the Federal Reserve System, international Finance Division, Washington, DC 20551, USA. Peter Pauly is with the Department of Economics, University of Pennsylvania, Philadelphia, PA 19104, USA.

This paper represents the views of the authors and should not be interpreted as reflecting the views of the Board of Governors of the Federal Reserve System or other members of its staff.

Final manuscript received 3 March 1984.

As a corollary of removing the assumption of oil price exogeneity, we also relax one of the most common and important assumptions made in the analyses of oil price determination, namely that the demand for oil faced by OPEC is either a stable function or one that shifts at a known (exogenous) growth rate (Hotelling: Pindyck! Salant: Cremer-Weitzman’). This assump tion implies that oil price changes have no effect on either the level of real income or its growth rate, ie real income, and the rest of the economy are considered exogenous, which is in contrast to the ‘supply shock’ literature and to 10 years of empirical findings.

What the above discussion suggests is that, up to now, the effects of oil price increases and the determination of oil prices have been analysed in a dichotomous manner since what is assumed exogenous in one analysis is assumed endogenous in the other and vice versa. We remove this dichotomy by treating oil prices and economic activity as jointly determined. This joint determination of oil prices and real income implies that, in addition to the problem of allocating oil production through time, OPEC faces a tradeoff between exploiting the direct price inelasticity of oil demand on the one hand and avoiding the feedback effects of oil prices on real income on the other. In particular, if the price path is too low, then OPEC doesnot erploit the price inelasticity of oil demand and thus incurs revenue losses. However, if the price path is too high, then real income of oil

0140-9BB3/B4/040267-09 303.00 0 1984 Butterworth & Co (Publishers) Ltd 267

OPEC’s pricing policy and tranmission of oil price effects: J. Marquez and P. Pauly

importers is adversely affected and this feeds back to OPEC in the form of lower oil imports and oil revenue losses. Recognition of these feedback effects is important because it implies that the price elasticity of oil demand is not a parameter, but as Stiglitz’ has pointed out, an endogenous variable depending on oil prices and on the effects of oil prices on real income of oil importing economies.*

We start the analysis of the joint determination of oil prices and world economic activity by developing a three region world model to study the international transmission of oil price changes to real income, prices and international trade assuming that oil price changes are exogenous. Later we endogenize oil price changes by postulating a welfare function including OPEC’s objectives, and applying optimal control to derive formally a relation linking macroeconomic variables to oil prices. We implement our approach empirically and determine the importance of relaxing the assumption of (oil) price taking behaviour for both the optimal oil price path and OPEC’s revenue.

A global model of oil price effects

The model we use is simple but comprehensive and is based on a theoretical model developed by Marquez.” We divide the world economy into three regions:

The developed economies, whose GDP is determined from the demand side, where the supply side is intro- duced by using a production function to determine prices and factor input demands. The OPEC countries, for whom we only analyse the recycling of oil export earnings to purchase imports of manufactures. The non-OPEC developing countries, whose GDP is determined from the supply side using a production function framework. Centrally planned economies have been excluded from the present model.

The model includes 11 prices:

For developed economies we include the export price of manufactures, wage rate, rental price of capital, price of coal, the price of investment goods, the consumption deflator, and the value added price deflator. For OPEC we only include the price of oil. For non-OPEC developing countries we have export prices of both manufactures and raw materials, as well as the rental price of capital.

*This can ba seen if we assume that the demand for oil, D, depends on real oil prices Po and real income of oil importers, Y, which is also a function of real oil prices: D = D[P,, YtPo)). The price elasticity of D is equal to:

(Po/DI (aDlaP = {aDlaP, + [aolav(.)l tdY~-)kf~Ol}Pom~ The term (aD/aY)fdY/dP,) represents the income feedback effect of oil price changes on oil demand, and given that dY/ dP, < 0 it follows that the existence of this income effect raises the price elasticity relevant for oil pricing decisions.

The international trade flows included in the model are:

0 Oil, sold by OPEC to developed economies and non- OPEC developing economies. The latter also export oil to developed economies.

0 Manufactures, exported by the DCs to both OPEC and LDCs.

l Manufactures and raw material exported by the non-OPEC LDCs to DCs.

In a compact form, the model has 52 equations, 16 of which are behavioural (see Marquez” for a detailed description of the model). The parameters of the model are estimated using data for 1960-79, although some of the relationships are estimated using data only up to 1977. The estimation method we use is OLS; the advan- tage of alternative parameter estimators such as 2SLS, 3SLS, and FIML need not hold for small samples such as ours (Marian~‘~). The main behavioural equations are shown, in general form, in Table 1.

Developed economies Equation (1) represents consumption, c’, depending on a distributed lag of nominal value added eYd deflated by the consumption price index P”,,, ; the short-run mpc is 0.47 and the long-run income elasticity is one. Investment, Id in Equation (2), is a function of a distributed lag of real income as well as the long-term nominal interest rate rd; the short-run mpi is 0.13 and the long- run elasticity with respect to income is 1.18; the long- run elasticity with respect to the interest rate is -0.20.

Table 1. Main behavioural relations in global econometric model.

Developed economies

cd = f CP$Yd/pcdo”,

Id = f( Yd, rd)

Mf = f(P,/pdm, Ydl

A%J$ = f(A?;P,, A%P/, A%P,, _, , U_, I

GYd = f {L, K, E(0, CM}

Od = f(Po, PC, P/, Pk. GYd)

#+Gd-GS

n;i$&- X~W,/p,, GYd)

M$ = (P;/P,) c& + n;,

OPEC

M% = f[Po&JM$ +M;)l

NowDPEC developing countries

GY’ = f(K’, 0’)

II = f(Pi/po, Y’. R’/p$

M’ = f(/’ R’Bd I m nl m I

x, = f(P$/p&, Yd)

0’s flP$-o, GY’)

#&0’__0sJ*j&

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(81

(91

(10)

(11)

(12)

(13)

(14)

(15)

(161

ENERGY ECONOMICS October 1984

OPEC’S pricing policy and transmission of oil price effects: J. Marquez and P. Pauly

Imports of raw materials from non-OPEC developing countries, Mf in Equation (3), are a function of present and past values of the price of raw materials, P,, relative to the export price of manuactures of DCs, P$, and real income of DCs, Yd; the long-run relative price elasticity is - 0.69 and the long-run income elasticity is 0.63.

The inflation rate in the export price of manufactures, A%& in Equation (4), depends on the rate of change of wages, A%P,, with a coefficient of 0.3266; the inflation rate of oil prices, A%P,, lagged one period with a coefficient of 0.09; the inflation rate of the price of raw materials with a coefficient of 0.34; and a measure of capacity of utilization, U, obtained as the difference between potential output and actual output.

Potential output, Y*, is estimated as a trend of actual output but we split the period of estimation into two subperiods: 1960-72 and 1973-79. The estimated growth rate for potential output for the period prior to 1973 is 4.7% and for the period after 1972 is 2.9%.

The demand for oil (bbl), od in Equation (6), is derived as a conditional demand function from a three level CES production function, Equation (5), whose arguments are labour, L, capital, K, oil, 0, and coal, C. In this way we account for both capital-energy and interfuel substitution possibilities. In addition, an aggregate measure of energy consistent with the structure of production can be derived. The conditional demand for oil depends on the prices of oil, coal, labour, and the rental price of capital as well as on gross output (value added plus oil imports) GYd. We estimate a linearized version of this demand function with a distributed lag (four periods) for prices and income, allowing for homogeneity of degree zero in prices. In estimating this relationship, we use switching regression with a split of the sample in 1972. The long- run oil price elasticity declines from -0.27 to -0.57. The long-run coal price elasticity increases from 0.29 to 0.91. The estimated income elasticity declines from 1.70 to 1.34. The hypothesis of homogeneity of zero degree in prices cannot be rejected.

Total imports of oil (bbl), M$ in Equation (7), are equal to the difference between the demand for oil (bbl) and the exogenously given supply of oil (bbl), OS. Imports of oil from OPEC, 2: in Equation (8), are equal to total imports of oil minus imports of oil from non-OPEC LDCs, xi, which in turn depend on the price of oil relative to the price of coal as well as on real income of DCs. To link oil imports in barrels to oil imports in real value we use the identity between the value of oil imports given by the product of total oil imports in real terms, Mt, and the oil price index, P,, and the value of oil imports given by the product of the price of oil ($/bbl), P$, and total imports of oil in barrels, $z. We then solve from this identity for oil imports in real terms as in Equation (9).

OPEC We emphasize here the absorption capacity of OPEC. For this we explain their imports of manufactures, ML in Equation (IO), as a function of a distributed lag


of oil revenues deflated by the export price of manufactures of DCs. The absorption elasticity in the first year is 0.31, 0.36 in the second year and 0.04 after four years. The long-run absorption elasticity is 0.98.

Non-OPEC developing countries To determine gross output (value added plus oil imports in real terms), GY’ in Equation (1 I), we use a two level nested CES production function with capital, K’, and oil consumption, O’, as arguments. The parameters of this function are estimated in two steps by using the first order conditions for cost minimization and the production function itself.

The short-run elasticity of substitution between oil and capital is 0.05 and the long-run elasticity is 0.73.

Following Coen,‘3 we determine capital formation, I’ in Equation (12), as a function of the rental price of capital, Pi, relative to the price of oil, and on gross output. In addition, we allow the speed of adjustment of capital formation to depend on the availability of foreign exchange resources in real terms, RyPd,. We find an inelastic response of investment with respect to changes in both income and relative prices. However, we find investment to be quite responsive to changes in foreign exchange reserves (an elasticity in excess of one).

Imports of manufactures, Mf, in Equation (13), are derived using the fact that the capital stock is an aggregate between the domestic capital stock and the foreign capital stock with a non-zero elasticity of substitution. Following Marquez;‘4 we obtain imports of manufactures as a function of net investment, If,, and foreign exchange reserves. Our results point to an elasticity of imports of manufactures with respect to investment of 0.8 and an elasticity with respect to (real) foreign exchange reserves in excess of one.

Exports of manufactures to DCs, Xf, in Equation (14), depend on the export price of manufactures of DCs relative to the export price of manufactures of LDCs, pd,/Pf, , and the real GDP of DCs. Oil consumption, 0’ in Equation (15), is modelled as a conditional demand function derived from the production function (Equation 11) and thus depends on relative factor prices and gross output GY’. Using the identity between world consumption (bbl) and world oil production (bbl), we derive imports of oil from OPEC, @, (bbl) as the difference between LDCs oil demand and (exogenously given) oil supply, OS*’ plus exports of oil of LDCs to DCs. Finally we link oil imports in barrels to oil imports in real value following the same steps as in Equation (9) for DCs.

For the purposes of the next part, we represent the econometric model in Table 1 in a compact form as:

Yt = fIyt* y,(L), Xt. s-1. e. et1 (17)

where

yt = vector of endogenous variables -f = vector of functions L = lag operator zt = vector of exogenous variables

269

OPEC’s pricing policy and transmission of oil price effects: J. Marquez and P. Pauly

c = vector of parameters e = vector of residuals, 5 - N(0, S?,)

Simultaneous determination of oil prices and real income: an optimal control approach

In order to endogenize oil prices we minimize the expected cost of not achieving OPEC’s objectives subject to the behavioural constraints and identities embodied in the econometric model in Table 1:

mh E(W) = E 1

5 @r - ar)‘K(yr - a,> t=1 I

subject to

for oil prices’ growth, {x”}~. Using Equation (17)’ (the econometric model), we determine the effect that oil price changes have on real income of oil importers. These income effects, together with the substitution effects, are transmitted to OPEC in the form of both movements along and shifts of the demand schedule. These shifts in the oil demand schedule are now, using Equation (18), taken into account by OPEC (with a lag of one period) to obtain a second oil price path {xl}, which, by Equation (17)‘, is used to recompute the effects of oil prices on economic activity. This second round of oil price effects is, in turn, used to update the previous oil price path and so on. This iterative process between oil prices, real income, and oil demand continues until

Notice that our representation of the econometric model here differs from that of Equation (17) in that it has been expressed in its state-space representation, and exogenous variables have been reclassified as either instruments, st, or non-controllables, cr. K is a positive semidefmite matrix with weights along the main diagonal; a, is a vector of desired values for targets, such as OPEC’s oil revenues, and instruments, such as oil price changes. The choice of a quadratic welfare function is for mathe- matical convenience.

abs {t$“)t - {z”-’ ItI m (I,

where s is the sth iteration.

Experimental design Description of th_e welfare function The particular welfare function we use is:

(

1990 4

minE(W)=E C C wit(Y&-air)* r=1982 i=l

The solution of the first order conditions of the above minimization problem involves applying Pontryagin’s minimum principle and, due to space limitations, the required steps are not shown here (see Marquez”). Following Chow,r6 the solution to this problem takes the form of a linear feedback rule linking real income, and other exogenous variables included in yt _r , to oil prices and other exogenous variables included in xr:

1990 4

+ 1 C kjr(Xjr - rjr)2 r=1982 j=l I

where the targets variables are:

Yl = the growth rate for real oil revenues of OPEC, A%R”

(18)

Y2 = the growth rate for real income of developed economies, A% Yd

where C, is a time varying matrix with elements depending on the parameters of the model, the desired values for instruments and targets, and the elements of the K matrix. The advantages of using a feedback rule to obtain the values for the instruments are that their values are updated (with a lag of one period) as infor- mation about the future values of the endogenous variables becomes known, and current policies determine, to some extent, future policies. This is an important consideration in the case of oil pricing policies since a change in oil prices today will have a lasting influence and this must be taken into account when determining the price of oil. Having derived an expres- sion for optimal oil prices, we combine it with the econometric model to obtain an ‘enlarged’ econometric model where real income, an element of Yt, and (optimal) oil prices, an element of St, are simultaneously determined:

Ys = the inflation rate of the export price of manufactures of developed economies, A%Pd,

Y4 = the growth rate for non-OPEC developing countries, A% Y’

The policy instruments we use are:

Xl = the growth rate for oil prices, A%P, X2 = long-term interest rates of developed economies,

rd X3 = real government expenditures in DCs, Cd X4 = nominal net capital transfers to non-OPEC

developing countries, CF

4(r=f(Yr)Yr-l,Xr,LLr-l,~r,e,er) (17)’

~r=Gr@,KK, %t)Yr-1 (18)

To illustrate this joint determination of oil prices and economic activity, suppose we are given an initial path

We recognize here that OPEC members are countries in the early stages of industrialization using oil revenues, which represent more than 90% of their export earnings and a significant share of their GDP, to finance the foreign component of their capital stock in order to accelerate their development process (Aperjis,” Marquez”). Thus we assume that OPEC sets the price of oil in order to secure a flow of oil export earnings capable of sustaining their development programmes. We also include in the welfare function growth rates for developed and non-OPEC developing economies to study both the feasibility of a high growth path

270 ENERGY ECONOMICS October 1984

for the world economy and the distribution of income Real oil revenues of OPEC, R”, are targeted to grow gains between these blocks of countries. Finally, we at 1% during 1982 and 1983, a relatively small growth include the inflation rate of the export price of manu- rate by the standards of the 197Os, but relatively ambi- factures because its acceleration would dampen the tious in relation to the performance of oil revenues for volume of capital goods that OPEC and non-OPEC the 1982-83 period. We increase this growth rate to developing countries could afford and this would 4% in 1984, 6% in 1985, reaching 10% in 1989 and adversely affect their development possibilities. 1990. The target for the inflation rate of exports of

With respect to the instruments, the growth rate manufactures starts at 7% in 1982, steadily declines to for oil prices is included since it is the instrument that 5% in 1985, and remains at 5% for the remainder of the OPEC uses to ensure a steady flow of export earnings. horizon. We include government expenditures and interest Oil prices are targeted to have zero growth in rates of developed economies because they allow us 1982, a decline of 15% in 1983, and a 6% increase for to study what type of coordinated fiscal, and (only the 1984-90 period which implies a 1% increase in real indirectly) monetary policies can be implemented oil prices for this last period. Long-term nominal interest by developed economies in order to achieve a non- rates, rd, are targeted to decline from 14% in 1982 to inflationary growth path for the world economy. In 8% by 1986 and to remain at this level for the remainder addition, it allows us to examine the reaction of fiscal of the horizon. This path for nominal interest rates policy to changes in the price of oil. Finally, capital implies a 3% real interest rate for the period 1986-90. transfers are included to study the financial require- Real government expenditures of DCs, Gd, and nominal ments implied by sustained growth for the LDCs. net capital transfers to non-OPEC developing countries,

Parameters of the welfare function CF, are expressed as an index (1982 = 1 .O) in order to avoid the problems that arise between units of measure-

The desired values for the targets and instruments are ment and the magnitudes of the weights of the welfare shown in Table 2. To determine these desired values we function. For government expenditures we assume an postulate a transition of the world economy from a average annual growth rate of 0.9% per year to reflect ‘no-growth’ path to a ‘high growth’ path while reducing the concern over increasing government deficits of

inflation. We assume a target growth rate for developed current administrations. For nominal capital transfers

economies of 2.5% in 1982, which is steadily increased to LDCs we use an 8.5% growth rate starting in 1980,

to 5.8% in 1989 and 1990. Similarly, for developing which we obtain from the ‘high-transfer scenario’ of

countries we assume a target growth rate of 4% in 1982; the World Bank Report for 1982.

we increase this growth rate to 7% for the period 1987- The weights in the welfare function (the w’s and the

90 which is the target growth rate reported in the k’s) represent our preferences for the various targets.

United Nations Development Report (United Nations”). In particular, it is unlikely that all the target variables will reach their desired values given the non-linear nature of our model and the choice of ambitious targets.

Table 2. Base optimal control solution. Desired values for targets and instruments (1982-1990).

Tar-

A% Yd A%R” A% Y’ A%P$

1982 2.5 1.0 4.0 7.0 1983 3.4 1.0 4.8 6.5 1984 4.0 4.0 5.5 5.5 1985 4.3 6.0 6.0 5.0 1988 4.8 7.0 6.5 5.0 1987 5.2 8.0 7.0 5.0 1988 5.5 9.0 7.0 5.0 1989 5.8 10.0 7.0 5.0 1990 5.8 10.0 7.0 5.0

In other words, there are tradeoffs between our objectives. We give an equal preference ranking to each target and thus we assign weights equal to one. The tradeoffs between targets that may arise here are due to the behavioural constraints in our model and the weights given to the instruments. Admittedly, the choice of weights for instruments is less intuitive than the choice of weights for targets. In particular, the iterative algorithm used (closed loop) is sensitive to the choice of values for the weights for the instruments, especially low values (similar findings are reported by Klein and Suzo). To select the values for the instruments weights we take into account that policy changes are not likely to be large or frequent (Klein*‘). Thus we give a large weight

1982 1983 1984 1985 1986 1987 1988 1989 1990

Instruments (99999) to real government expenditures of DCs and to nominal net capital transfers (1500). Interest rates

A%P, d r Gd CF are assigned a weight of 20 and oil prices are assigned

0.0 14 1.0 1.0 a weight of 4.5. Notice that oil price changes have the

-15.0 11 1 .Ol 1.11 smallest weight, ie fluctuations in oil prices are the least

6.5 10 1.02 1.23 penalized and, in a sense, they are ‘encharged’ of 10 1.03 1.37 achieving the targets. 8 1.04 1.52 8 1.05 1.70 8 1.06 1.87 Empirical results 8 1.07 2.08 8 1.08 2.31 In this section we apply the optimal control algorithm

developed by Chow22*23 to the econometric model


ENERGY ECONOMICS October 1984 271

OPEC’S pricing policy and transmission of oil price effects: J. Marquez and P. Pauiy

for the period 1982-90 given the targets and weights described above. Figures 1 and 2 contain the mean fitted values for both instruments and targets.?

Starting with the instruments, we notice that devia- tions of interest rates from their desired values are, in general, very small Oil prices increase by 3.2% in 1982, decline by 8.8% in 1983, and increase in 1984 by 7%; for the 1985-90 period oil price changes remain in the 6% range, implying a growth rate in real oil prices of 1% for the 1984-90 period, where real oil price changes are computed as the growth rate in oil prices minus the inflation rate of manufacture exports of developed economies. Finally, real government expenditures and nominal capital transfers are very close to their desired values.

With respect to the targets, the inflation rate of manufacture exports decreases from 5.9% in 1982 to 3.9% in 1984, but rises again to 5.8% by 1990. The

tThe optimal control solutions converged in three iterations. We are grateful to Professor Chow for providing the software used in this paper.

272

6, , , , I , , ,

1962 1964 1966 1966 I990

initial decline in the inflation rate is due to the fall in oil prices in 1983 and to the reduction in economic activity of DCs in 1982. Growth in developed economies, after experiencing a decline in activity in 1982 of l%, shows a steady increase from 3.2% in 1983 to 4.5% in 1990. OPEC’s real oil revenues decline sharply in 1982 and 1983 (16% in each year) because of the recession in oil importing countries and the decline in oil prices. However, these revenues increase by 10.8% in 1984, 6.5% in 1985, and maintain an annual average growth rate of 5.8% for the period 1986-90. Non-OPEC developing countries show positive growth for the whole period starting with 3% in 1982, a small decline of the growth rate to 2.1% in 1984, but increasing to 3% in 1985, 3.6% in 1986, 5% in 1989, and 5.6% in 1990; the average annual growth rate is 3.5%. This growth rate is higher than the ‘high scenario’ target growth rate for developing economies (3.3%) of the 1982 report of the World Bank?4

Recognizing the limitations of our model, and the particular welfare function used, we find that if nominal

2.50 -

Figure 1. Mean fitted values for instruments: base optimal control solution.



6

t B - Bar optimal

I omtd solution

1982 1984 1986 1988 1990

1 1 I I I I I

1982 1984 1986 1986 1990 1982 1964 I986 1988 1990

6

I I I I 1 1 I I

Figure 2. Mean fitted values for targets: base optimal control solution.

oil prices grow at an average of 6% per year, nominal capital transfers to non-OPEC developing countries grow at 8.5% per year, long-term interest rates decline to 8.0%, and real government expenditures continue their (modest) expansionary path, then it is feasible to achieve relatively high growth rates for developed, OPEC, and non-OPEC developing countries.

The importance of income feedback effects and OPEC’s pricing policy Our main contention has been that allowance for the income feedback effects of oil price changes raises the absolute value of the price elasticity and therefore lowers the optimal price path consistent with OPEC’s best interests. What is not so clear, however, is whether these feedback effects are important enough to result in significantly different oil price paths.

One way to evaluate the importance of these feedback effects in the determination of an optimal oil price path is to contrast a path where oil pricesgrow at a rate equal to the interest rate, and thus no feedback

effects are allowed, with an optimal oil price path where feedback effects are allowed.

The assumption of oil rate is based on Stiglitz’s’ sp

rices growing at the interest result establishing that if the

price elasticity of demand is constant, then the path for oil prices for the monopolist and the perfect com- petitor are equal to each other and equal to the interest rate (Hotelling,z6 Solow2’). Since this result is derived without allowing for feedback effects, it seems appro- priate to use it here in deriving a price path without feedback effects.

Table 3 and Figure 3 contrast the oil price paths, the growth of real oil revenues for OPEC, and the growth of real income of DCs for the cases of ‘no-feedbacks allowed’ and of ‘feedbacks allowed’. We notice that for every year, except the first two, the no-feedback price path results in a growth path for OPEC’s revenues lower than the path for prices derived with optimal control, even though oil prices grow faster in the no- feedback case than in the optimal control case. The deterioration of OPEC’s revenue growth in the no-



Table 3. Optimal oil prices with and without feedback effects allowed.

No feedbadc effects allowed Feedbaok effects allowed

1982 i 983 1984 1985 1986 1987 i 988 I 989 1990

A%P, A%R* A% Yd A%P, A%R* A% Yd

14 -8.3 -1.26 3.2 -15.8 -1.0 11 -9.4 0.33 -8.8 -16.2 3.2 10 -1.5 1.6 7.0 lo.8 2.1 10 1.2 2.1 6.3 6.5 3.2 a -0.6 1.5 6.4 4.9 2.9 a 2.3 2.2 5.9 6.6 4.5 a 2.7 3.3 6.2 5.4 4.2 a 4.6 2.9 6.2 6.4 4.4 a 2.8 2.9 6.7 5.7 4.5

1963 -

1984 -

1965 -

1986 -

1987 -

1988 -

1989 -

1990 r,

-10 0 IO Gmfh rate of 011 prices

1982 -

1983 -

1984 -

1985 -

1986 -

1987 -

I988 -

1989 -

1990 t -2

Gtuwth mte of DCs

Figure 3. Optimal oil prices with and without feedback effects.

feedback case occurs because the revenue increase that arises out of short-term price inelasticity of oil demand is offset by the reduction in real income of oil importers which dampens oil consumption and shifts to the left the export demand schedule faced by OPEC. In other words, the income feedback effects of oil price increases raises the (absolute value of) the price elasticity above one and therefore price increases lead to revenue losses.$

Conclusions

Our purpose in this paper has been to study the international transmission of oil price effects and the determination of optimal oil price paths, not as separate problems but rather as one problem by recognizing that economic activity and oil prices are jointly deter-

*It is possible to argue that the observed revenue growth for the optimal control solutions is not due to income feedback effects, but rather due to some response of some other instrument such as government expenditures, which stimulates DCs’ real income end thus OPEC’s revenue growth. To make sure that an instrument response, other than oil prices, is not responsible for the rapid growth in OPEC’s revenue, we have given a rather high weight (99999) to government expenditures. We find that the average deviation of government expenditures from their target is, in absolute value, equal to 0.996, which can hardly be considered as a reason for the increase in OPEC’s revenue growth.

1983

1984

1985

1986

1987

I988

I989

1990 d

-2( 1 -10 0 IO 20 OPEC’s mvwwe gmwth

mined. We start our analysis by developing a non-linear stochastic dynamic model for a three region world economy that highlights the channels of the international transmission of oil price effects and the feedback effects of oil price changes on the demand for oil. Recognizing the limitations of our model, we apply optimal control to an estimated version of this model and we find that if oil prices grow at 6% per year, government expenditures grow at O.% per year, and capital transfers to developing countries grow at 9% per year, then it is possible to achieve a non-inflationary recovery of the world economy. However, our most important findings are that taking into account income feedback effects leads to a path for oil price growth below the oil price path when income effects are not taken into account, and that real oil revenues of OPEC grow at a faster rate when income effects are allowed in the determination of oil prices than in the case where these income effects are not allowed.

To conclude, the common thread of this paper is that the feedback effects of oil price changes should be taken into account in determining the price of oil. Not taking these feedback effects into account results in a significant upward bias of the total price elasticity of oil demand and in the oil price path, neither of which is in OPEC’s best interests.

274 ENERGY ECONOMICS October 1984


References Modelina of World Commodity Policy, Lexington

1

2

3

4

5

6

7

8

9

10

11 12

N. Marion and L. Svensson, Structural Differences and Macroeconomic Adjustments to Oil Price Increases in a Three-Country Model, National Bureau of Economic Research. Working Paner No 839, Cambridge, MA, 1982. ’ - - M. Bruno and J. Sachs, Macroeconomic Adjustments with Import Price Shocks: Real and Monetary Aspects, The Maurice Falk Institute for Research in Economics. Discussion Paoer 793. Jerusalem. 1979. ’ R. Findlay and C. A. Rodriguez, ‘Intermediate imports and macroeconomic policy under flexible exchange rates’, Canadian Journal of Economics, Vol 10, No 2, 1977, pp 208-217. M. Schmid, ‘A model of trade in money, goods, and factors’, Journal of International Economics; Vo16. No 4. 1976. DD 347-362. H. -Hotel&g, ~’ I* ‘The economics of exhaustible resources’, The Journal of Political Economy, Vol 39, No 2, 1931, pp 137-175. R. Pindyck, ‘Gains to producers from the carteli- zation of exhaustible resources’, Review of Econo- mics and Statistics, Vol 60, No 2, 1978, pp 238- ?<I LJI.

S. Salant, Imperfect Competition in The World Oil Market. Lexinaton Books. Lexington. MA. 1982. I. Cremer and-M. Weitzman, ‘OPEC ‘and the monopoly price of oil’, European Economic Review, Vo18, No 2, 1976, pp 155-164. J. Stiglitz, ‘Monopoly and the rate of extraction of exhaustible resources’, American Economic Review. Vo166. No 4. 1976. PU 655-661. J. Marquez, The Intemationai Transmission of Oil Price Effects and the Derivation of Outimal Oil Prices, PhD dissertation, University of Pennsylvania, 1983. Ibid. R. Mariano, ‘Commodity market modeling: metho- dological issues and control theory applications’, in F. G. Adams and J. Behrman, eds, Econometric

13

14

15 16

17

18

19

20

21

22 23

24

;z 27

Books, Lexmgton, MA, 1978. _ R. Coen. ‘The effect of cash flow on the speed of adjustment’, in G. Fromm, ed, Tax Incentives and Capital Spending, Brookings Institution, Washington, DC. 1971. J. Marquez, Foreign Exchange Constraints, Znvest- ment Demand, and Imports of Manufactures in LDCs, Center for Analysis of Developing Economies, Discussion Paper 82-47, University of Pennsylvania, 1982. Marquez, op tit, Ref 10. G. Chow. Analvsis and Control ofDvnamic Svstems. John Wiley, New York, 1975. _ _ _ ’ D. Aperjis, The Oil Market in the 1980’s: OPEC Oil Policy and Economic Development, Ballinger, Cambridge, MA, 1982. J. Marquez, ‘Industrial policy in Venezuela’, in F. G. Adams and L. R. Klein, eds, Industrial Policies for Growth and Competitiveness, Lexington Books, Lexington, MA, 1983. Trade and Development Report, 1982, United Nations, New York,-1 982. _ L. Klein and V. Su. ‘Recent economic fluctuations and stabilization ‘policies: an optimal control approach’, in L. R. Klein, M. Nerlove and S. Tsiang, eds, Quantitative Economics and Development, Academic Press, New York, 1980. L. Klein, ‘Economic policy formulation: Theory and implementation (applied econometrics in the public sector)‘, in Handbook of Econometrics, Vol 2, North Holland, Amsterdam, 1983. Chow, op tit, Ref 16. G. Chow, Econometric Analysis by Control Methods, John Wiley, New York, 198 1. World Bank, World Development Report, Oxford University Press, New York, 1982. Stiglitz, op tit, Ref 9. Hotelhng, op tit, Ref 5. R. Solow, ‘The economics of resources or the resources of economics’, American Economic Review, Vol54, No 2, 1974, pp 1- 14.


opec's pricing policy and the international transmission of oil price effects

Documents