University of Southern California Dissertations and Theses

The price of oil and OPEC behavior: A utility maximization model

(USC Thesis Other)

The price of oil and OPEC behavior: A utility maximization model

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content TH E PRICE O F OIL A N D O P E C BEHAVIO R A UTILITY MAXIMIZATION M O D E L by M oham m ed K. Adeinat A Dissertation Presented to the FA C U LTY O F TH E G R A D U A T E S C H O O L UNIVERSITY O F S O U TH E R N CALIFORNIA In Partial Fulfillm ent of the Requirements for the Degree D O C T O R O F PHILO SO PHY (Economics) December 1985 UMI Number: DP23332 All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. Dissertation Publishing UMI DP23332 Published by ProQuest LLC (2014). Copyright in the Dissertation held by the Author. Microform Edition © ProQuest LLC. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106-1346 UNIVERSITY OF SOUTHERN CAUFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK . LOS ANGELES, CALIFORNIA 90089 j~> ^ I p ) ELc 'ss M , This dissertation, written by Mohammed K. Adeinat". under the direction of /z ? .§ ......... Dissertation Committee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of re quirements for the degree of DO CTO R OF PH ILO SO PHY D e f y o f Graduate Studies , October 17, 1985 D a te ......................... DISSERTATION COMMITTEE ■ * < 3 1 , . . _ ._ _ r • • • • • • • • • • » • m W t t u • • '» • « • • • • »7 «T « «» 11» i ^ n m - r M i? i ii • Chairperson T O M Y P A R E N TS ii A C K N O W LE D G M E N T S This acknowledgment is coming from deep in m y heart to those individuals w ho helped and encouraged m e during this research. I want to thank m y teachers and m y friends; without th e ir con tinuous help and support this dissertation could not have been com pleted. First and foremost, I am deeply indebted to the chairman of m y dissertation committee, Dr. Jeffery B. Nugent, whose continuous support and guidance were strong and unceasing. His close super vision and assistance was generous and invaluable in correcting the many e a rlie r drafts. I am also grateful for the other two members of m y committee, Dr. Niedercorn and Dr. McEachern, for th e ir encouragement, cooperation, and kindness. Fin ally, I want to thank m y wife who shared with m e the hard times of being a student. TABLE O F C O N TE N TS DEDICATION........................................................................................... i i A C K N O W L E D G M E N T S .................................................................................. i i i LIST O F TABLES..................................................................................... vi LIST O F FIGURES vi i i Chapter I. INTRODUCTION ........................................................................... 1 I I . A HISTORICAL B A C K G R O U N D ...................................................... 7 2.1 Period from 1960 to 1973............................................ 8 2.2 The 1973-1974 Oil Price Q u ad ru p lin g................. 12 2.3 The 1974-1978 P e rio d ................................................. 16 2.4 The 1979-1980 Price Doubling ................................. 17 2.5 Aftermath of the 1979-1980 Price Doubling . . 22 I I I . A CRITICAL L O O K A T TH E LITERATURE O N OIL PRICES A N D O P E C BEHAVIOR . . ......................................... 24 3.1 The P ro fit Maximization Models ............................. 24 3.2 The Non-profit Maximization or Target Revenue Model .............................................................. 33 3.3 The U tility Maximization Model ............................. 39 IV. SAVINGS FUNCTION O F S O M E O P E C COUNTRIES............................... 44 4.1 The Permanent Income Hypothesis ......................... 45 4.2 A C ritical Review of the Permanent Income Hypothesis and of the Technique for its T e s t in g .......................................................... ... 47 4.3 Model I ........................................................................ 57 4.4 Model I I ..................... ................................................... 64 4.5 Model I I I . . ....... ................................................. 82 iv V. TH E UTILITY MAXIMIZATION M O D E L . ............................. 96 5.1 The General U tility Maximization Model . . . . 98 5.2 Derivation of the Supply Curve for a Cobb-Douglas U tility Function ............................. 101 5.3 Absorptive Capacity and the Supply Curve . . . 104 5.4 Budget D e fic it with a Lower Limit on Government Expenditures and the "Supply Curve" ............................................................. 107 V I. SU M M A R Y, CONCLUSIONS, A N D SUG GESTIO NS F O R FU TU R E R E S E A R C H ............................................................................ 113 6.1 Summary and C o nclusio ns................................... .... 113 6.2 Main F in d in g s ........................................... 115 6.3 Suggestions for Future Research ........................ 117 M ATHEM ATICAL APPENDIX ................................................................................ 119 REFERENCES.................................................................................... 133 v LIST O F TABLES The Price Trend of Mideast Light Crude-34 of the Period 1960-1982 ............................................................. World Production of Oil during the 3-month Embargo Period from October to December 1973 in Thousands of Barrels per D a y ........................................ World Production of Crude Oil 1971-1981 in Thousand of Barrels per Day ............................ Published Estimates of World Oil E lasticities Government Revenue and Expenditure for O P E C Countries for the Period 1970/71-1979/80 in M illions of National Currency Units . . . . . Regression Results: Consumption Function of the Form, Ct=Kl Yt +K2Yt +et» where the Values of g Are Predetermined ana the Values of g Minimize the Residual Sum-square . ..................... Regression Results under the Assumption that K 2 =0 , where the Values of 8 Are Considered as Predetermined Given from Table IV .2 . . , Regression Results under the Assumption that l<2=0, where the Values of 8 Minimize the Residual Sum-squares ......................................... , Regression Results: Consumption Function of the Form C t =K0+K 1Y t +K2Y ^ et ................................................. Maximum Likelihood Estimates of the Consumption Function C.'*=Ki Y?*+KoY ■ wH ptp thp Valnpc nf rum iiuii 1 i t 2 t* wnere tne and g Are Estimated within the Model where the Values of Maximum Likelihood Estimates of the Consumption Function, C jt=Kl Y Tt+K2Y?t* where Kj Is Allowed to Vary Over Time . . . . . IV .8 Maxim um Likelihood Estimates of the Consumption Function C^t s s KjY^t +K 2 Yjt , where Ps5 =Psg=Ps7 =Psg . . 76 IV .9 M axim um Likelihood Estimates of the Consumption Function C it=Kl Yi t +K2Yit» with < 2 = 0 .............................. 77 IV .10 Maxim um Likelihood Estimates of the Consumption Function C ^ K jY L + I^ yL ., where a is Changing Over Time where e Is Assumed Constant and Equal to 0.022 78 IV .11 M axim um Likelihood Estimates of the Consumption Function C^=<]^u+< 2 Y ]t, where < 2 Is Assumed Equal to Zero, ana e Is Assumed Constant and Equal to 0.022 79 IV .12 M axim um Likelihood Estimates of the Consumption Function C ^ K iY u + I^ y L -, where < 2 Is Assumed Equal to Zero ana a Is Changing Over T im e.................. 80 IV .13 Maxim um Likelihood Estimates of the Consumption Function C ^ = KjY^t + KjY: t where < 2 = 0 and both a and 0 Are Changing Over T im e ..................... 81 IV .14 Maxim um Likelihood Results where the Permanent Income Model Is Defined as Y^c^l n^-Pf n^ t+ and K 2 Is Assum ed to Be Zero 90 IV .15 M axim um Likelihood Results where the Permanent Income Model Is Defined as Yt=allntpi ,t-l+a2lnRt+6t ................ * • • * 9I IV .16 Maxim um Likelihood Results where the Permanent Income Model Is Defined as Yt =a,lnpf ^_j+a2 lnRt + 6t and the Investment Function Is Defined as I=<{y£, with <2= 0 ................................................................ 92 IV .17 Maxim um Likelihood Results where Yt =ajlnt Pf t _j+ lnRt + < 5 t 2 and lnqt =0ilnP+02Y£, Where <2 = 0 . '.... 93 IV .18 M axim um Likelihood Results under the Assumption that <2=0, and that Permanent Income is Defined as Yt = alwt+6t ................................... 9^ IV .19 M axim um Likelihood Results of the Investment Function (I=<fYt ) under the Assumption that <2=0, and that Permanent Income Is Defined as Yt=alwt+<st • • 95 vi i LIST O F FIGURES Fi gure 111-1 The gain from collusion (the cartel case) . . . . . 26 111.2 The backward-bending supply curve (target revenue model) 36 111.3 Dem and and supply and the determination of the price of o il ......................................................................... 3 8 111.4 Saving and investment equilibrium in the presence of the absorptive capacity constraint ......................... 42 IV .l Consumption function under expenditure rig id ity . . 72 V .l The backward-bending supply curve (when wealth is considered as a normal good) ..................... 102 V.2 The demand curve for both consumption and investment goods for a Stone-Geary u t ility f u n c t io n .................... 112 Chapter I INTRODUCTION The objective of this dissertation is to a lle v ia te economists' frustration arising from the fa ilu re of existing models to explain the behavior of individual OPEC-type countries. More than two decades have passed since O P E C was established. While m uch has been w ritten , especially since the 1973-1974 price quadrupling, the result has been widespread disagreement on the major issues involved. Is O P E C a cartel or not? H ow can one model O P E C behavior? The theory of the exhaustible resources has long been considered the main the oretical approach to the modeling of O P E C behavior. Recently, how ever, som e economists have started questioning the a p p lic a b ility of this theory to O P E C behavior, leading at least som e of them to judge that O P E C behavior is inconsistent with neoclassical theory in gen eral (Teece, 1982). Notably, even those w ho were prominent advocates of this theory have started questioning the v a lid ity of such a theory (Pindyck, 1982). With the evidence against i t mounting rapidly, the theory of exhaustible resources seems to have h it a dead end. Spe c ific a lly , w e are le ft with the conclusion that O P E C does not maxi mize p ro fits , but without m uch of a clue as to how to explain its behavior. Most of the existing lite ra tu re applying the theory of exhaustible resources to O P E C behavior seems to amount to m uch ado about nothing. The economics of o il became a hot subject a fte r the 1973-1974 price quadrupling, which, by 1974-1975, had led most of the industrial countries into a severe recession. Many economists started looking for a theory which might both explain the price quadrupling and serve as an analytical tool for studying the o il industry, for predicting future supplies and prices of o il, and for minimizing the effects of any future disruption O n oil-im porting countries. At that time, i t was commonly believed that the 1973-1974 price quadrupling was a result of the cartel power of O P E C in the efforts to maximize discounted values of oil p ro fits . Models of this type became the accepted dogm a over time*, however, i t has gradually become clear that during the 1970s at least, O P E C did not behave as a cartel in the trad ition al sense. I t also became clear that O P E C countries were not maximizing the discounted values of th e ir o il p ro fits. In order to find a way out of the existing dead end, a fresh look at O P E C and the experience of its member countries would seem in order. The fir s t step must be to review the historical background of O P E C countries to see how th e ir prices of oil have been determined and to identify the factors influencing their decisions on how m uch crude oil to produce. The next step should be to suggest a model which is consistent with both the historical record of O P E C countries and the factors iden tified as having influenced O P E C behavior in the recent past. A u t ilit y maximization model subject to absorptive 2 capacity and budget constraints would seem consistent with the above c rite ria . The suggested theory is motivated by the facts that several O P E C countries have m ade i t obvious th at they determine th eir production of crude oil with reference to th e ir budgetary needs and numerous public statements have been made by o ffic ia ls of som e O P E C countries admitting this (Teece, 1982). Because of the importance of absorp tive capacity and budgetary constraints, i f an O P E C country's total revenue is more than its requirements of total expenditures, i t is lik e ly to cut back its production of crude o il. O n the other hand, i f its required total expenditure is more than its total revenue, i t is lik e ly to expand its production of crude o il. Where an O P E C country's governmental budgetary needs (to ta l expenditures) are determined by utility-m axim izing behavior, given o il prices, its production of crude oil is determined in such a way as to maximize the present value of its present and future government consumption and investment. A s a result of the dramatic increases in the price of o il which have been witnessed during the last decade, government revenues for O P E C countries have generally increased so rapidly as to generate savings in amounts which are well beyond th e ir economy's absorptive capacity. At the sam e time, and with a few notable excep tions, O P E C investment in foreign assets has been surprisingly small (Teece, 1982). A m ong the factors which have made O P E C countries reluctant to invest in foreign assets are the following: 3 1. Foreign investments may yield returns which are low when compared with those realizable either by keeping the oil in the ground in a situation in which the price of o il is expected to rise sig n ifican tly in the future or by investing in local real estate, buildings, and services. 2. Foreign assets are risky, th e ir risk being compounded by the exchange risk resulting from v o la tile foreign exchange markets. 3. Foreign investments contribute l i t t l e to the aspirations on the part of O P E C countries to increase the degree of their p o litic a l and economic independence from developed countries. O P E C countries learned a lesson in this respect from the action taken during the hostage cris is in 1979 by U S President Carter in freezing a ll the Iranian assets in the USA. Given the lim ited interest in foreign investments and the lack of other sources of government finance, o il production in O P E C countries would Seem to have been determined in large part by absorptive capa c ity and budget constraints. W hen the price of o il went up sharply as i t did in 1973-1974 and again in 1979-1980, these increases in price generated such dramatic increases in o il revenues as to make total government revenues exceed budgetary requirements by such a large margin as to go well beyond their absorptive capacities. Therefore, in order to reduce the amount of saving to a level consis tent with the optimal amount of investment, o il production was reduced. To understand how O P E C countries determine th eir production of crude o il, considerable e ffo rt should be m ade in studying saving 4 | behavior. Since optimal savings and consumption behavior can be | derived from life -c y c le u t ilit y maximization,* a permanent income model was developed in order to achieve the above objective. The results obtained in estimating the permanent income hypoth esis model should help us in specifying our u t ilit y approach model. A s far as this investigator is aware, this is the fir s t study to test i j the application of the permanent income hypothesis to behavior of the j government sector. Tests are conducted under a number of d ifferen t I sets of assumptions and restrictions. Three permanent income models, I j each with d ifferent features, are specified. In some of them, some 1 of the restrictions which are commonly imposed by economists are ! relaxed. Also, the models allow us to capture interactions among changes in the price of o il, the quantity of o il produced, and the amount of reserves and how they affect government consumption, government saving, and government investment. Next is presented the u t ilit y maximization model, in which each j O P E C country at any given time is assumed to maximize its u t ility i j defined in terms of current consumption expenditure, its investment expenditure, and its future expenditure, subject to its budget con- j straint. The supply of o il function is derived fir s t for the general case, and then for three special cases, namely, (1) the case of the Cobb-Douglas u t ilit y function, (2) the case of the Cobb-Douglas *The linkage between a permanent income model and a u t ility j maximization model has been done in the lite ra tu re , specifically in j Labor Economics (see Macurdy, 1981 and Hickman, 1974). u t ilit y function combined with an absorptive capacity constraint, and (3) the case of a Stone-Geary u t ility function with lower lim its (subsistence levels) on both government current expenditure and government investment expenditure. The remainder of this dissertation is organized as follows: Chapter I I gives some historical background on O P E C behavior and the price of o il. Chapter I I I provides a c ritic a l review of existing theories and models of O P E C behavior and the world o il market. Chap ter IV presents a permanent income model in which both saving and consumption functions are estimated and tested under d ifferen t assumptions. Chapter V presents the utility-m axim ization model. Chapter VI provides a summary of findings and implications along with suggestions for future extensions of the model and related research. 6 Chapter I I A HISTORICAL BACKGROUND In the fir s t half of this century and the la tte r part of the last century, American and European petroleum ccompanies obtained con cessions for o il exploration and exploitation in the Middle East and other areas thought to be promising for oil discovery. At that time these companies had re la tiv e ly complete control over the petroleum industry, and to som e extent they interfered with the p o litic a l lif e of the host countries. After World War I I , these host countries began to seek independence and realized the d e s ira b ility of bringing about fundamental changes in th e ir relations with concessionaire companies. O ne of the e a rlie s t efforts of this sort occurred when the Arab League was formed in 1945. Although from its beginning the Arab League recognized the importance of creating an organization for o il exporting Arab countries, i t was unable to do so because of its fear of the concessionaire companies. A s a resu lt, the domination of the international petroleum market by the large multinational com panies continued, forcing the host countries to bend to th e ir terms. The demand for o il increased rapidly, attracting more and more com panies to invest in o il in various parts of the world. By the late 1950s, the supply of o il was increasing m uch more rapidly than the demand for o il. A s a resu lt, these companies found themselves with 7 large surpluses of crude o il, imposing downward pressure on o il prices. The possibility of reductions in actual or effective prices w as a major concern for o il producing countries, since th e ir o il tax and royalty payments were based on market prices. Market prices had been declining since 1957. The concern about actual and prospective reductions in oil prices was accentuated by the fact that the com panies could reduce crude o il prices without any pre-consultation with the host countries. In addition, the host countries realized that th e ir most valuable asset was being exploited by foreign o il companies. For these reasons and others, representatives from five countries--Iran, Iraq , Kuwait, Saudi Arabia, and Venezuela--had a meeting in September 1960 and agreed to create an organization of the petroleum exporting countries, "OPEC." Four months la te r, Qatar was admitted. Both Libya and Indonesia became members in 1962; the United Arab Emirates joined in 1967 and was followed by Algeria in 1969, Nigeria in 1971, and Ecuador, the tw elfth member, in 1976 (Ghabrial, 1977). During the sam e year Gabon was accepted as an associate member. The objective of O P E C at that time was to s ta b il ize the price of oil in the international market by unifying petro leum policies am ong the O P E C countries, regulating the supply of o il, and ensuring that o il companies consulted with the host countries prior to changes in o il prices (Jacoby, 1974). 2.1 The Period from 1960 to 1973 In the foreign non-Communist world, the consumption of oil increased from 2.4 m illion barrels per day in 1948 to 10 m illio n 8 barrels per day in 1973, representing an average compound growth rate of 11% per annum (G riffin , 1982). During the 1960s O P E C was involved in changing its tax system; i t succeeded in abandoning the old system where taxes were based on market prices, and replacing i t with a new system in which taxes were based on a tax reference price. This price was higher than the o ffic ia l market price and never declined even though market prices fe ll throughout most of the 1960s, fa llin g from $1.86/barrel in 1960 to $1.75/barrel in 1970 (See Table I I . 1). Also during this period, O P E C emphasized the legitim ate right of a ll its members to permanent sovereignty over th e ir natural resources. O P E C in its conference in Vienna in June 1968 issued a declaration which showed the ultimate goal of O P E C to be to help member countries gain effective control over o il company operations in th e ir coun trie s . This declaration was not taken seriously until the early 1970s when som e O P E C countries announced nationalization programs for transferring ownership in local concessions from the companies to the host government (G riffin , 1982). By the beginning of 1970, many O P E C countries believed that the posted price of o il should be raised. Indeed, when O P E C met in Caracas in December 1970, i t agreed on the establishment of a commit tee comprised of Iran , Iraq , and Saudi Arabia to negotiate with o il companies on behalf of member countries of the Gulf (Abu Dhabi, Iran, Iraq , Kuwait, Qatar, and Saudi Arabia) for purposes of achieving a ^ com m on set of objectives. These objectives were to raise the tax I rate on the net income of o il companies to a minimum of 55%; to 9 Table I I . 1 The Price Trend of Mideast Light Crude-34 over the Period 1960-1982. __________ Mideast Light Crude-34_________ Year Posted Price O fficial Price Spot Price 1960 $1.85 $1.86 $1.63 1961 1.80 1.80 1.57 1962 . 1.80 1.80 1.52 1963 1.80 1.80 1.50 1964 1.80 1.80 1.45 1965 1.80 1.66 1.42 1966 1.80 1.53 1.36 1967 1.80 1.50 1.33 1968 1.80 > 1.45 1.32 1969 1.80 1.40 1.27 1970 1.80 1.35 1.21 1971 1st half 2.18 1.75 1.64 2nd half 2.29 1.75 1.74 1972 1st half 2.44 1.90 1.77 2nd half 2.48 1.90 1.87 1973 1st Q 2.59 2.10 2.08 2nd Q 2.79 **2.25 2.35 3rd Q 3.03 **2.55 2.70 4th Q 4.76 **3.65 *4.10 1974 1st Q 11.65 **8.65 *13.00 2nd Q 11.65 **9.60 10.60 3rd Q 11.65 **9.60 10.00 4th Q 11.38 10.40 10.30 1975 1st Q 11.25 10.46 10.42 2nd Q 11.25 10.46 10.42 3rd Q 11.25 10.46 10.43 4th Q 12.38 10.46 10.46 1976 1st Q 12.38 11.51 11.51 2nd Q 12.38 11.51 11.51 3rd Q 12.38 11.51 11.60 4th Q 12.38 11.51 11.90 1977 1st Q 13.00 12.09 12.50 2nd Q 13.00 12.09 12.45 3rd Q 13.66 12.70 12.63 4th Q 13.66 12.70 12.68 1978 1st Q - - 12.70 12.66 2nd Q — 12.70 12.70 3rd Q - - 12.70 12.79 4th Q — 12.70 13.50 10 Table I I .1 continued __________Mideast Light Crude-34_________ Year Posted Price O ffic ia l Price Spot Price 1979 1st Q — 13.48 18.35 2nd Q — 16.15 27.35 3rd Q - - 18.89 32.90 4th Q - - 22.84 38.17 1980 1st Q — 27.17 36.58 2nd Q — 28.82 35.52 3rd Q - - 30.21 33.30 4th Q - - 31.33 38.63 1981 1st Q - - 32.60 37.32 2nd Q — 33.00 33.58 3rd Q — 33.05 32.06 4th Q - - 34.16 33.73 1982 Jan. — 34.00 34.00 Feb. - - 33.85 30.40 March — 33.55 28.20 ♦Market prices varied considerably during this period when very few transactions occurred. Figures shown only indicate broad trends. ♦♦Actual contract sales prices for Arabian Light. Although an o ffic ia l Saudi price existed, i t applied only to the re la tiv e ly few direct sales deals by Petromia. From Petroleum Intelligence Weekly, April 12, 1982, p. 11. eliminate the differences in posted or tax reference prices of crude o il; to establish a uniform general increase in the posted or tax reference price in a ll member countries (O PEC Chronological Events, 1980, pp. 9 ). The result was an agreement between six member coun trie s and 23 international o il companies on Feb. 14, 1971 whereby the posted price of Arabian lig h t was raised to $2.18/barrel. The agree ment also provided for a gradual increase iri the price of o il to adjust for the in fla tio n rate and other factors. The posted prices of o il continued to grow throughout this period, reaching a price of $3.03/barrel by the third Quarter of 1973. * 2.2 The 1973-1974 Oil Price Quadrupling Throughout the early 1970s, and prior to the A rab-Israeli war which took place on October 6, 1973, the demand for o il continued to increase rapidly, thereby tightening the market. The planned demand for o il was rising faster than its supply (Petroleum Intelligence Weekly, August 20, 1973). During the Arab-Israeli war, public opin ion in most Arab countries favored an embargo on oil exports to the USA, and the nationalization of a ll American oil companies in re ta l iation for what was seen as U S A involvement in the war on the side of Israel and against the Arabs. Therefore, the Arab governments needed to do something to ease the popular pressure on them to take action. Accordingly, ten days la te r, on October 16, 1973, the m inisterial committee of the aforementioned six O P E C Gulf Countries met in Kuwait and u n ila te ra lly declared that the posted prices of a ll crude o ils be raised by 70% above the prevailing levels. I t was the f ir s t time 12 I ; that O P E C raised its crude price u n ila te ra lly without negotiating I with the o il companies. Three days la te r, on October 19, 1973, the • members of O A P E C met in Riyadh and imposed an o il embargo on U S A and i the Netherlands for allegedly taking the side of Israel during the | war. O A P E C countries agreed to reduce th eir October production of I : o il by 25% relative to the September levels, and then by an addi- i ! tional 5 % each successive month until the U S A adopted a more 1 j j friendly attitude toward the Arab countries (Johany, 1980). A s shown ! in Table 11.2, O A P E C countries reduced th eir production of o il from j j 19.555 m illion barrels per day in September 1973 to 17.995 m illion ! barrels per day in October. Although i t increased s lig h tly in Novem- ! ber and December, the December 1973 production was s t ill well below i | the September 1973 level. | W hen O A P E C countries started cutting back th e ir production of o il, there was fear on the part of the companies and the importing countries that the embargo would last for a long period of time and, as a result of the programmed further reductions in production, that the market would get even tig h ter. Therefore, future supplies of oi were expected to decline sharply in the future, bearing in mind that | j the o il market was already very tig h t prior to the Arab Embargo. t j Hence, i t was believed that there would be a great shortage of o il in 1 the future. A s a resu lt, the price of o il started going up. The ! problem was aggravated by the behavior of o il importing countries, : o il companies, and o il producing countries during the c risis. The j uncertainty about the future supplies of o il led both the o il Table 11.2 World Production of Oil during the 3-month Embargo Period from October to December 1973 in Thousands of Barrels per Day.* September October November December Abu Dhabi A 1 geria Iraq Kuwait Libya Qatar Saudi Arabia 1381.0 1100.0 2167.0 3479.0 2286.0 608.3 8533.5 1340.0 1020.0 1797.0 3058.4 2383.7 598.0 7797.9 1155.0 880.0 1923.0 2615.0 1766.0 505.0 6269.4 1016.4 860.0 2136.5 1555.8 1769.0 460.4 6615.8 O A P E C 19554.8 17995.0 15113.4 15413.9 Ecuador Indonesia Iran Nigeria Venezuela 470.0 1350.0 5793.0 2101.9 3395.0 210.0 1406.0 5978.0 2190.4 3371.0 210.0 1391.0 6009.0 2200.0 3384.0 230.0 1400.0 6070.5 2255.2 3330.0 O P E C (excluding O APEC) 13109.0 13155.4 13194.0 13285.7 O P E C % 32664.7 31150.4 4.636% 28307.4 13.33% 28699.6 12.14% Free World (excluding OPEC) % 15142.1 15031.4 3.4% 14934.6 9.5% 14956.9 8.7% Communist 9678 9774 9845 10050.0 World Total 57.484 C l 55955.8 53087 53706.5 ♦From Oil and Gas Journal 14 companies and the o il importing countries tq^increase th e ir dem and f.qr^crude oil regardless of the consequences of their actions for the world^o il market (Schneider. 1983). The demand for o il increased so rapidly due to higher precautionary and speculative demand for o il. For example, the oil importing countries increased th eir precaution ary demands for oil in the form of greater inventories of crude oil to secure themselves against any future disturbances in the supply of o il. In addition, i t was expected that the price of o il would increase sharply in the near future; therefore, the o il companies increased their demand for o il to satisfy th e ir speculative motives. Regarding the o il producing countries, i t was expected that the non- O P E C producing countries would take advantage of the opportunity to export at higher prices by increasing th e ir production of o il. How ever, ju st the opposite occurred. From Table I I . 2 i t can be seen that am ong the non-OAPEC members of OPEC, only Iran and Nigeria increased th eir production of o il between September and December 1973. Also, to ta l non-OPEC production of o il in the free world declined from 15142.1 thousand barrels per day in September to 15031.4 thousand barrels per day in October and again to 14934.6 thousand barrels per day in November. All these actions led to an explosive increase in the price of o il. By early 1974 the spot price of o il reached $13 per barrel, and in som e cases even higher. For example, the U SA paid $15 per barrel for the o il i t imported from Canada. All this translated into a 211% 15 increment in the market price of o il in 1974, even though the posted price of O P E C o il remained somewhat below this at $11.65 per barrel. Therefore, i t was neither the deliberate action of O P E C nor the sharp reduction in supply of o il by O P E C countries that were the main factors behind this explosion in the price of oil as commonly believed by the specialists in the fie ld . Rather, i t was the behav ior of both o il importing countries and o il companies in increasing both th eir precautionary and speculative demands for o il. This explains why both MacAvoy (1982) and Schneider (1983) argue that the behavior of both the o il importing countries and the o il companies was the major cause of the 1973-1974 o il price quadrupling. 2.3 The 1974-1978 Period After the sharp price rise of 1973-1974, most of the industrial countries experienced a stagflationary recession during 1974 and 1975. Most economists believe that this recession was in large part caused by the o il c ris is , leading to both higher in fla tio n and higher unemployment, i . e . , s tag flatio n . In other words, the P hillips curve had become positive (Friedman 1978). A s a result of the recession,, the demand for O P E C o il fe ll by about 15% below its 1973 le v e l. By 1976, however, O P E C production of oil was back to its 1973 le v e l, and continued to ris e through 1978. As a resu lt, the price of o il remained rather stable, increasing only slig h tly in nominal terms, and fa llin g in real terms. Although much research has been done regarding the price of oil in general and the role of O P E C in particular, unfortunately there 16 remains wide disagreement as to the cause of the c ris is , the role of OPEC, the future price of o il, and how to model O P E C behavior. For example, som e economists have argued that the price of oil would go back to its original level prior to 1973-1974. In p articu lar, Milton Friedman (Newsweek, March 4, 1974) argued that both O P E C and the price of o il would collapse since O P E C is a cartel and cartels are inevitably unstable. Others believed O P E C to be irrelevant and that the price of oil should be explained through the forces of demand and supply, the price jump of 1973-1974 being a result of changing owner ship (Johany 1978, 1980). Most commonly, O P E C behavior and the future prices of crude o il were explained in terms of the theory of exhaustible resources. 2.4 The 1979-1980 Price Doubling Another p o litical event took place in the Middle-East when the Iranian revolution started in 1978, which led to the overthrow of the Shah and brought about a new p o litic a l system which worked hard to spread the revolution to Muslim neighboring countries. This created widespread uncertainty about the future supply of oil in the Middle East, leading to an increase in the demand for o il. The problem was aggravated when the war between Iran and Iraq started in 1979, caus ing severe reductions in Iranian o il production from 5.275 m illion barrels per day (mb/D) in 1978 to 3.175 mb/D in 1979 and to only 1.48 mb/D in 1980. The outbreak of war between these countries also led to speculation that the war might spread to the other neighboring Middle-East countries. This created uncertainty about both future 17 prices and supplies of o il. The o il importing countries and o il companies responded by building up th e ir inventories of crude o il to satisfy th e ir precautionary and speculative demands for o il. This led to a new rise in the price of o il. Theoretically speaking, this p o litic a l unrest should not have any significant impact on the price of oil because the reduction of o il production by Iran could be easily offset by increasing the o il production of other countries. Actually, as shown in Table I I . 3, this is what happened. The Middle East production of o il increased from 21.42 mb/D in 1978 to 21.91 mb/D in 1979. Also, non-OPEC free world production of o il increased from 17.33 mb/D in 1978 to 18.62 mb/D in 1979. Therefore, this second o il price explosion cannot and should not be attrib u ted , as was commonly believed by specialists, to a severe reduction in the supply of o il due to the Iran -Iraq war. I t is worth mentioning that the Iran-Iraq war led to a new devel opment in the o il market which was an increase in the relative impor tance of the spot market. The spot price of oil could for the f ir s t time be considered to be a leading indicator to or predictor of future o ffic ia l prices. A s a res u lt, o il exporting countries started to determine th e ir o ffic ia l prices based on information about spot prices. I t is estimated that 15 to 20% of the world o il production was sold in the spot market in November 1979, compared to only 3 to 5 % in January of the sam e year (Johany, 1980). The spot prices were changing rapidly during this period. From Table I I . 1 the spot price of Middle East lig h t 34 increased from $13.5 per barrel in the fourth 18 Table I I . 3 World Production of Crude Oil 1971-1981 in Thousand of Barrels per Oay Country 1971 1972 1973 1974 1975 1976 1977 A1geria 603 1061 1035.4 888.8 935 950 990 Ecuador 3.9 59.4 197 232 165 185 180 Indonesia 880 1027 1300 1457 1400 1500 1690 Iran 4514 4900 6000 6128 5600 5875 5650 Iraq 1692 1500 1888.2 1839.3 2400 2070 2150 Kuwait 2895 2750 2890.2 2600 1950 1820 1700 Libya 2800 2230 2116.6 1700 1400 1900 2050 Nigeria 1543.4 1800 2000 2300 1850 2020 2150 Qatar 429 450 555.5 546 410 485 350 Saudi Arabia 4456 5255 7417.9 8400 7000 8570 8950 United Arab Emirate (UAE) 1026 1130 1508.5 1982 1760 1905 2000 Venezuela 3579 3200 3370 3025 2400 2290 2280 Total O P E C % 24421.3 25362.4 30279.3 29786.1 27270 29570 30140 Free World (exluding O P E C ) % 15466.2 15426.2 15621.4 16595.9 14830.6 15280.5 16237.8 Total Free World 39887.5 40788.6 45900.7 46382 42100.6 44850.5 46377.8 C om m unist % 8000 8910 9312 10390 11750 12360 13150 W O R L D T O T A L 47887.5 49698.6 55212.7 56722 53850.6 57210.5 59527.8 19 Table I I . 3, continued Country 1978 1979 1980 1981 1982 1983 1984 Algeria 1260 1240 1000 750 750 686 608 Ecuador 200 220 230 204 215 236 254 Indonesia 1650 1600 1570 1607 1341 1292 1332 Iran 5250 2900 1280 1375 1896 2606 2166 Iraq 2500 3370 2600 892 914 905 1218 Kuwait 1900 2210 1400 916 675 912 925 Libya 2050 2050 1780 1063 1127 1020 1090 Nigeria 1800 2370 2100 1369 1324 1232 1414 Qatar 480 480 470 414 340 270 404 Saudi Arabia 7800 9250 9620 9642 6484 4872 4545 United Arab Emirate (UAE) 1810 1810 1930 1503 1240 1084 1074 Venezuela 2150 2330 2150 2093 1826 1791 1724 Total O P E C % 28850 29830 26130 21828 18132 16906 16754 Free W orld (exluding O P E C ) % 17333.3 18618.4 18963.9 19448.2 20220 21408 22366 Total Free World 46183.3 48448.4 45093.9 41276.2 38352 38314 39120 Com m unist % 13810 14140 14580 14610 14650 14945 14970 W O R L D T O T A L 59993.3 62588.4 59673.9 55886.2 53002 53259 54090 ♦Oil and G as Journal 20 quarter of 1978 to $38.18 per barrel in the la s t quarter of 1979. Accordingly, the o ffic ia l price of o il almost doubled in 1980 compared to 1979 prices. Up to this point O P E C never acted as a c a rte l. I t never forced its members to cut back th eir production of o il, but rather i t was up to the s e lf-in te re s t of each country to determine how much crude oil produce. I t is clear that O P E C was not a cartel in a trad ition al sense. O P E C tried to set prices, or at least the price of one particular kind of crude o il, "the marker," Arabian lig h t 34° Apl, and in one market, the " o ffic ia l" or long-term contract market. Nevertheless, since each member was free to determine how m uch o il to produce, which was regarded to be a matter of national sovereignty, there is no way i t could make its prices "stick." However, inasmuch as most specialists in this fie ld and public opinion as a whole continued to regard the price of o il as being dictated by the O P E C c a rte l, i t is not surprising that the forecasts of o il prices that emerged have been far from accurate. A s w ill be discussed in more detail in the following chapters, the analysts have used the wrong model. This is not to say that O P E C had no power to influence the price of o il, but only that both the 1973-1974 and the 1979-1980 price rises cannot be explained by cartel action in reducing the amount of output, as som e specialists s t ill believe. 2.5 Aftermath of the 1979-1980 Price Doubling A s a result of the 1979-1980 price jump, o il revenues of O P E C countries increased explosively and beyond th e ir absorptive capacity. 21 This explains why most O P E C countries cut back their production of crude o il after the price rise in order to reduce the huge surpluses in th e ir budgets. At the sam e time, the world economy entered another major recession. The world o il demand fe ll by about 20%, because of both the recession and conservation policies in o il importing countries. Since non-OPEC o il production was increasing, this implied an especially sharp decline in O P E C oil production. The demand reductions for O PEC o il a fte r the 1979-1980 price doubling were very deep and lasted for a longer period of time than in the 1973-1974 price ris e . Inasmuch as O P E C kept its o ffic ia l price at $34 per barrel (which was about 20% above the spot p rice), O PEC countries found i t increasingly d iffic u lt to sell th e ir o il. The decline in O P E C sales brought about large reductions in th e ir o il revenues. By early 1983, O P E C production of crude oil was only 53% of the 1979 level of production (Gately, 1984). In 1982 and 1983, many O P E C countries were for the f ir s t time in more than a decade starting to experience budget d e fic its . Eventually, O P E C realized that i t had to lower its o ffic ia l price of o il in order to compete in the international market, and hence to preserve even its already reduced share of the market. Therefore, i t reduced the price of Arabian lig h t 34° to $29 per barrel in 1983, its fir s t reduction in the price of o il. Also, for the f ir s t time since its creation, O P E C adopted a production quota in order to help stab ilize the price of o il. However, such actions did not resolve th e ir marketing problems, leading som e O P E C countries to further reduce prices, and further 22 exacerbating th e ir budgetary problems. In January 1984, O P E C as a whole agreed to reduce its price of o il a second time to $28 per barrel. Since that time until the time of w riting this dissertation (August 1985), O P E C has been under a continuous pressure to further reduce its prices of o il. 23 . C HAPTER I I I A CRITICAL LOOK AT THE LITERATURE ON OIL PRICES AND OPEC BEHAVIOR Since the 1973-1974 price quadrupling, numerous studies have been carried out to analyze the reason(s) behind the price ris e , its influence on economic growth in o il importing countries, how to model O P E C behavior, and how best to predict the future prices of o il. Much less e ffo rt has been devoted to analyzing the effect of the o il price rise on the economies of o il exporting countries. The lite r a ture which exists can be categorized into one or another of the following alternative models: (1) p ro fit maximization models, (2) nonprofit maximization models, and (3) u t ilit y maximization model s. 3.1 The P ro fit Maximization Models All models under this category are optimization models based on the traditional theory of exhaustible resources, fir s t presented in 1938 by Hotelling in his pioneering paper, "The Economics of Exhaus tib le Resources." The idea is to solve for the optimal price path of an exhaustible resource by maximizing the present value of discounted p ro fit. Under this category of models, w e distinguish between two types: (1) the p ro fit maximization— non-competitive model where O P E C is considered as a c a rte l, and (2) the p ro fit maximization-- competitive model or property rights model. 3.1.1 The Non-competitive (C artel) Model With the aid of Figure I I I . l , w e can explain how O PEC determines its prices and what incentive i t has to collude and form a c a rte l. Let us assume th a t, before colluding, each o il exporting country is making only a normal p ro fit where the price of o il is given by P c and each country is producing qc. After collusion, O P E C w ill maximize p ro fit by equating M R ^ = MC^, and O P E C w ill produce at Q m instead of Q c. This raises the price level to P m . O P E C then allocates produc tion am ong producers, and now the representative firm is making p ro fit by producing qm at price level P m ; But notice that qm is not the p ro fit maximizing output for the representative country, and there is an incentive for the representative country to depart from the collusive solution and produce at qc . I f other countries do the same, then the cartel w ill collapse, unless the agreement to cut production is honored by a ll members and monitored by the organization its e lf . D ifferent versions of this model have been used. For example, som e specialists regard O P E C as a cartel with a dominant producer, Saudi Arabia, serving as the price leader. A summary of this lite r a ture can be found in Dasgupta and Heal (1979). Several applied models of this type were b u ilt. Prominent am ong the studies d ire ctly pertaining to o il price-setting models is the work of Kalymon (1975). His study uses dynamic optimization techniques, where the criterio n 25 ,c p ' Figure I I I . l . The gain from collusion (the cartel case). 26 function is to maximize the total discounted benefits of oil resources. From that the optimal price trajectories are calculated for both O P E C as a whole and for individual member countries. Sal ant (1976) presented a modified version of this conventional theory of exhaustible resources. H e assumed a Nash-Cournot equilibrium where neither O P E C nor the competitive fringe has any incentive to change th e ir intertemporal behavior. He defined O P E C as a collection of small firms operating with other firm s. Although o il prices are determined completely by OPEC, each competing firm taking the price path set by the cartel as given, each such firm is free to choose a sales path which maximizes the sum of discounted p ro fit. OPEC, in turn, takes the sales path of the competitive fringe as given and chooses the price path which maximizes its ow n discounted p ro fit. The equilibrium price and sales paths are then derived, although under sim plified and sometimes u n realistic assumptions. Pindyck (1978) used such a model to calculate the gains to pro ducers from the cartelizatio n of exhaustible resources in general and applied the model to OPEC. Indeed, he used i t to explain the 1973- 1974 price quadrupling. Other studies undertaken w ithin this framework are those of B litz e r and Stoutjesdijk (1975), and G ilbert (1978). Hence, i t is fa ir to say that the theory of exhaustible resources has been the dominant approach to explaining O PEC behavior and the world o il market during the 1970s. Models of this type became accepted by economists in the 1970s, and, even now, are s t ill highly regarded by 27 som e economists. Prominent am ong the o il industry specialists holding such views has been Adelman (1982). However, gradually the following short-cornings of the theory of exhaustible resources as the basis of O P E C and o il price behavior have been id e n tifie d . 1. Models of this type have reached a practical lim it as anal ytical tools. They are unable to predict world o il prices or even to explain O P E C behavior with any degree of precision whatsoever. Even those economists who are for long periods the advocates of this type of model have lo st confidence in this approach. For example, Pindyck (1982) pointed out that: I t seem s to m e that from a theoretical point of view, models of O P E C o il pricing have reached a practical lim it as tools of analysis. As far as empirical predic tions of o il prices are concerned, som e of these models have already exceeded that lim it . . . . economic ratio n a lity probably applies even less to O P E C producers than to many other real world economic agents, (pp. 175, 179) 2. Moreover, such models are highly sensitive to assumptions concerninig the values of the underlying parameter. For example, Gately (1984) pointed out that in Salant's model in E M F (1982), a change in the assumptions caused an average change in the 1980 and 2000 prices of 18% from the reference case. 3. In addition, such models yield results which are d iffic u lt to explain. As Gately observed in his 1984 paper: "the assumption that O P E C behaves as a wealth-maximizing price-setter often yields an in itia l-y e a r price quite d iffere n t from the actual value" (p. 1109). Then he gave an example of Salant’ s model (1982) which yields an 28 in it ia l price level for 1980 which was 28% above its actual price 1evel. 4. Most such models of this type assumed O P E C to be a cartel with complete control over prices. For reasons given in the previous chapter, serious doubts can be raised about the realism of such assumptions. 5. The additional assumptions, such as homogeneity of O P E C members "made to ensure mathematical tra c ta b ility " (Gately 1984, p. 1109) can also be challenged. Even Pindyck, one of the leading practitioners of these types of models, argued that, Perhaps the most serious problem is that the model assumes that O P E C acts as a single agent to maximize the equity value ( i . e . , the sum of present and discounted future p ro fits) of its o il revenues. But in fact the d iffere n t members of O P E C may have different objectives, and certainly operate under d iffere n t constraints. (1982, p. 178) 6. The evidence from the la s t two decades does not support the proposition that O P E C countries behave as a profit maximizer. For example, i f O P E C rea lly behaved as a p ro fit maximizer, then the following formula should hold: MR = P (1 - i ) ( 1) where e is the price e la s tic ity of demand. From the various authorita tiv e econometric estimates given in Table I I I . l , the short-run price e la s tic ity would seem to lie in the range 0.01 to 0.15. I f w e substitute such values for e into equation (1 ), w e obtain (-100P £ M R <^-7P). Since the profit-maximizing condition is that 29 Table I I I .1 Published Estimates of World Oil E la s tic itie s .* Energy Information Administration EIA 1979 Central Intelligence Agency Model O ne (CIA-1) 1977 Central Intel 1igence Agency Model T w o (CIA-2) 1979 Gately Kyle and Fischer (G-K-F) 1977 Michael Kennedy 1974 M ac Avoy 1982 D em and Price Elasticity Short R un -0.1 -0.06 to -0.15 -0.1 -0.01 to -0.06 - -0.04 Long R un -0.5 -0.17 to -0.28 -0.3 -0.1 to -0.2 ------- -0.4 Income Elasticity Short R un .1 to .17 .5 .35 — •» m m m m 0.1 Long R un .5 to .85 .95 to 1.1 1.00 ------- 1.0 1.0 Supply Price Elastici ty Short R un 0.4 .2 0.01 • - • 0.2 Long R un 0.8 — - 0.22 0.33 0.4 This table is adapted from M ac Avoy, pp. 28, 37 C O o M R = MC, this implies that M C is a very high negative number, which, of course, i t cannot be. Even i f w e should substitute a higher set of numbers for e, such as the estimates of the long-run price elas t ic it y , w e would s t ill get a negative solution for both M C and M R. Secondly, i f O P E C were a p ro fit maximizer, then w e could never obtain the observed short-run backward-bending supply curve (Teece, 1982). Therefore, from both the theoretical and practical points of view, these models do not ju s tify confidence, and w e are le f t without any sensible theory of o il. 3.1.2 The Competitive (Property Rights) Model Johany's (1978, 1980) objective was to show that O P E C as an organization was irrelevant in explaining the price quadrupling in 1973-1974. He argued, to the contrary, that the price of o il is completely determined by supply and demand and offered an alternative explanation for the 1973-1974 price quadrupling, what is called the Property Rights Model. He argued that before 1970, the decisions on how m uch oil to produce and how m uch to charge for o il were com pletely determined by private companies. Those companies were uncer tain about th eir future and feared nationalization. This led them to use a very high discount rate when they calculated the present value of their discounted p ro fit. According to Johany, they produced as i f there were no tomorrow. In the early 1970s, many steps were taken toward nationalization by the host countries. The price of o il and the rate of depletion became host country decisions. Moreover, since the oil exporting countries were concerned to a greater extent with ________________________________________________ ' _________________ 31 long-term objectives, such as economic development and the consump tion o f future generations, they used a m uch lower discount rate, which resulted in a sharp cut in the supply of o il, pushing the price of oil upward. Once again, however, this model can be said to be subject to numerous shortcomings which have become increasingly obvious. 1. F irs t, since this approach is also based on the p ro fit- maximizing theory of exhaustible resources, a ll the criticism s of that theory apply to this one as w ell. 2. Second, Johany’ s model can be c ritic iz e d from the point of view of inconsistency with historical facts. For example, as Adelman (1982) noted when he tested this model against the historical record, there were many cases when host countries asked oil companies to expand th e ir production of o il, In 1950-70, there was constant disagreement over produc tion levels. Countries always wanted more production than companies . . . . Iraq was a leader, for example, in June 1972 i t kicked out the Iraq Petroleum Company for not producing at high enough rates, (p. 39) 3. This model not only fa ils to provide a satisfactory explana tion for the 1973-1974 price quadrupling, but also fa ils to explain the o il price rise of 1979-1980, where no obvious interpretation can be offered. 4. I f i t were true that O P E C was irrelevant and that its m em bers acted com petitively, and since each country is assumed to maxi mize its p ro fit, w e would expect that each country would equate its price with its marginal cost. In re a lity , however, this has not been 32 the case. For a ll O P E C countries, marginal cost is well below the price of o il. For example, Johany (1980) pointed out that "what makes the o il market interesting from a theoretical point of view is the claim that the current price is forty to f if t y times the current marginal cost of production" (p. 1). Subsequently, in another chap te r, Johany said that "Economic theory teaches us that when we are able to determine that a price of a commodity persists well above its' marginal cost, then its producers must have some monopoly power" (p. 21). Therefore, i t would seem reasonable to put to Johany the fo l lowing question: "What is the ju s tific a tio n behind the assumption that the price of o il is determined by the underlying market condi tions and that each country takes the price as given and acts competitively?" 3.2 The Non-profit Maximization or Target Revenue Model Of the non-profit maximizing models, most are p o litic a l models. One non-profit maximizing model of an economic nature is the target revenue model of Teece (1982). Teece's main objective is to show that O P E C is not a cartel in the traditional sense, and that O P E C countries are not wealth-maximizers. This explains why O P E C has not collapsed under the pressure of cheating, and therefore why prices have not fallen back to the pre-1973 levels as was expected by many economists. Teece's view is that: In economic parlance, i t appears that, at least for an important subset of O P E C producers, the relationship between current price and current output is best repre sented by a backward-bending supply curve for the short 33 run. O ne implication is th a t, once a producer is on the backward bending portion of this curve, there is no p ro c livity to cheat on other O PEC members, (pp. 64-65) Teece argues that the behavior of O P E C countries supports this thesis. F irs t, he points to the public statements on various occa sions by O P E C ministers indicating that th e ir production decisions are made with reference to th e ir budgetary needs. Second, again in th e ir speeches he noted evidence suggesting that they respond to price increases by cutting the quantity of o il they produce. For instance, Kuwait’ s minister of o il said that "the big increase in o il prices has given the opportunity to review production,” adding that "the Kuwait Council of Ministers was discussing the po ssibilities of lower production" (October 29, 1976, Middle East Economic Survey). The Algerian Energy Minister also declared that " If terms of trade improve, Algerian exports w ill drop" (Petroleum Intelligence Weekly, May 12, 1980, p. 4 ). Sim ilar opinions have been stated by Libya, UAE, and even sometimes Saudi Arabia. Teece claimed that this behavior by O P E C is not consistent with the neo-classical theory and profit-maximizing behavior. This is because, i f O P E C members were profit-m axim izers, in view of the lim ited p ro fita b ility of investments at home, one would expect a large percentage of th e ir revenues from o il to be converted into foreign assets. However, in general, this has not been the case for various reasons. For example, Teece has pointed out that there are risks of investment in foreign assets, ranging from exchange rate changes to nationalization. R ecall, moreover, that O P E C countries 34 were taught a lesson as to th e ir vulnerability to arb itrary actions in host countries for th e ir investments when, during the hostage c ris is in 1979, President Carter froze Iranian assets in the United States. Further, reliance on investments abroad would be inconsis tent with th e ir struggle to be p o litic a lly independent and th e ir national aspiration for meaningful economic development (as opposed to being dependent on "coupon clipping," even i f this were the wealth-maximization behavior). 3.2.1 The Backward-Bending Supply Curves Teece argued that there is growing evidence that som e O P E C m e m bers determine th e ir domestic investment requirements f ir s t , and then set th eir o il production levels to match this target. Teece explained his proposed backward-bending supply curve by the aid of the diagram shown as Figure I I I . 2. A s shown in Figure I II . 2 a , MEI is the marginal efficiency of investment, which shows the investment opportunities open to an economy for a given rate of return. Let us assume that r is the rate of return picked by planners in that nation; then the desired level of investment w ill be I * . As long as I* is determined, then d iffe r ent combinations of o il prices and quantities can be chosen which give the same investment le v e l. See Figure III.2 b where I*/Q is the iso-revenue curve. Let us assume that this country is producing at less than fu ll capacity as at Qj. N ow assume there is an increase in the demand curve from dxj to dx2, which (ceteris paribus) would raise the price to P2 . Since the country would not wish to lower r , i t 35 Rate of Return Marginal efficiency of Investment r I* Price 2a. I Investment of Oil Iso-revenue curve 2 b. — ----------------Q Quantity of Oil Q P 'ice of 0 1 < 2 c. Short run r curve Quantity of Oil Figure I I I . 2. The backward-bending supply curve (target revenue would stick to I * , and hence with no other options, i t would reduce its production of o il, thereby yielding the backward-bending supply curve shown in Figure 111.2c. This target revenue approach can certainly be considered a step in the right direction. Nevertheless, i t suffers from serious shortcomings. 1. The allegations of Teece to the contrary, i t is s t il l prema ture to say that the behavior of O P E C is inconsistent with neo classical theory simply because of the evidence in favor of backward- bending supply curves. This is because there are several possible explanations for the reduction in oil production following a rise in price. For example, Johany's (1980) hypothesis of a change in the discount rate would suggest that the observed outcome could be the result of a s h ift in the supply curve as shown in Figure I I I . 3 rather than that arising from a backward-bending supply curve. 2. Teece's model always predicts a negatively sloped supply curve in the short run, which is , however, inconsistent with behavior of at least som e O P E C countries. 3. Teece failed to lin k O P E C behavior with economic theory. With a mere waving of hands, he declared neoclassical theory to be at odds with the behavior of O P E C countries. This implies either that O P E C countries behave irra tio n a lly or that other (unspecified) aspects provide the ra tio n a lity for th eir behavior. As we w ill show la te r, O P E C behavior can be seen as being consistent with ra tio n a lity and with neoclassical theory as long as ra tio n a lity is defined more 37 Price of Oil Quantity of o il Figure I I I . 3. Demand and supply and the determination of the price of o i l . 38 broadly as u t ilit y rather than wealth maximization. Therefore, the lack of a p p lic a b ility of the theory of exhaustible resources in explaining O P E C behavior need not imply that neoclassical theory is unable to explain the backward-bending supply curve, but only that the exhaustible resources theory is not the right theory to explain O P E C behavior due to the lim itations of the theory its e lf . 3.3 The U t ilit y Maximization Model This model provides an alternative means of explaining O P E C behavior. In this model, each O P E C country is assumed to maximize its social u t ilit y function subject to its budget constraint. The objective of each O P E C country is to produce just enough to satisfy its current and development expenditures. As in the target revenue approach, because of the extra risks involved in investing abroad, regardless of whether foreign interest rates are high or low, invest ment abroad w ill be lim ited . The countries want to be active part ners in development, not mere "coupon clippers." According to our utility-m axim ization theory, OPEC-type countries decide on how m uch crude o il to produce by maximizing th e ir social u t ilit y function, subject to th e ir budget and absorptive capacity constraints. Their u t ilit y function includes both current and future consumption. Oil-producing countries know that one day they w ill run out of o il, and since th e ir revenue from o il is the main, or even in som e coun trie s the only, source of income, they have to make use of this valuable resource. Many oil-exporting countries have developed very ambitious development plans in the la s t decade and, as a re s u lt, there has been a rapid expansion in development expenditures. For example, in Algeria, the level of development expenditure under the 1974-1977 development plan more than trip led that under the 1970-1973 development plan. In Saudi Arabia, development expenditures during the second five-year plan (1975/76-1979/80) were nearly nine times larger than those undertaken during the fir s t five-year plan (1970/71-1974/75). In Nigeria, expenditure rose 447% between 1973 and 1976. Other oil-exporting countries experienced sim ilar expenditure expansions in the 1970s (Wassink, 1978). The objective of oil-exporting countries is universally to achieve the highest level of economic development in the shortest possible period of time in order to achieve self-sustaining growth before they run out of o il, and to depend less and less on o il revenue in the future to finance th eir economic development. As development spending increased, many oil-exporting countries found themselves faced with absorptive capacity constraints. These con straints have arisen from (1) the rapid expansion of domestic expen ditures, (2) in su fficie n t or absent infrastructure, mainly in the transportation and distrib ution sectors, and (3) the in a b ility to augment the skilled labor force as fast as the overall labor stock. Aside from p o litic a l considerations, the major factor affecting the supply of oil in O P E C countries would seem to be th eir capacity to absorb th e ir oil revenues. Let us assume that I* represents the absorptive capacity of a particular O P E C country. Since many O P E C countries aim to achieve _ 4 .0 J the highest level of economic development, but are constrained by the absorptive capacities of th e ir economies, i t is completely valid to assume that I* is also the country's target level. The expenditure planners determine the amount of o il they should produce to achieve this target. Let us assume, as shown in Figure 111.4, that the equilibrium level of saving and investment (I = s') is at point A, where r is the equilibrium level of rate of return. Suppose instead that the planners choose the more ambitious target level of invest ment I * , which also represents the absorptive capacity lim it. But, since s would now be below the target level of investment, in order to achieve th e ir target, planners have to raise th eir savings level up to th eir target I* . Since saving is assumed to be proportional to o il revenue, any increment in o il revenue would, in turn, increase the level of savings; since o il revenue (TR = P 0t% t^ ^as tv'ro com~ ponents, the price of o il (P0^) and the amount of o il produced (q0t)» any increase in one of the above or both would generate higher total revenue, thereby leading to more saving. Let us assume that we are. on the original level ( I,r ) , and assume that there is a jump in the price of o il; this w ill lead to higher total revenue and therefore to more saving. Thus the saving curve w ill s h ift to the right. The magnitude of the s h ift would depend on by how much the price went up. Suppose that savings would s h ift to s', where B is the new equ ilib rium point which is below the target level of investment (I* ). Then the government would increase its production of o il in order to s h ift the savings curve so as to intersect the investment curve at the ____________________________________________________________ 4. 1 _ a r(TR) I = I ( y ) Figure 111.4. Saving and investment equilibrium in the presence of the absorptive capacity constraint. level as at point E. This is the case of positively sloped supply curve. O n the other hand, i f the savings curve would s h ift a ll the way to s" due to the jump in the price of o il, at point D the amount of saving is greater than the desired level of investment. Because of the absorptive capacity constraint, the government in this situa tion would cut back its production of o il, yielding a negatively sloped supply curve. Therefore, in order to know more about the behavior of o il- exporting countries, and p articu larly about th eir production deci sions, considerable e ffo rt should be made to study saving behavior. The objective of the next chapter (IV) is to take on this task. Both consumption and saving functions are estimated and tested under d ifferen t restrictio n s, and also the price of o il has been introduced e x p lic itly in the model in order to test its effect on both saving and consumption. This exploration helps us to know more about O P E C behavior, sp ecifically to find out how i t responds to price changes, and how i t determines how much crude o il to produce. After studying both government current expenditure and government saving (investment) in Chapter IV, we became aware of the main argu ments of the u t ilit y function, and this helped us in specifying the u t ilit y function and the budget constraint, and also in explaining our results. 43 Chapter IV SAVINGS FUNCTIONS OF SOME OPEC COUNTRIES (PERM ANENT IN C O M E A PPR O A C H ) The objective of this chapter is to test the saving function for O PEC countries so as to determine the effect that a change in the price of o il might have on saving in both the short and long runs, and how any such e ffe c t is related to the decision of how much crude o il to produce. In the process, w e estimate both the marginal pro pensity to save (MPS) and the marginal propensity to invest in order to see both whether or not O P E C countries were running a budget surplus (and hence whether or not they are faced with absorptive capacity constraints), and how the quantity of o il produced has responded to the budgetary needs of O P E C countries as would be suggested by the u t ilit y maximization model. In contrast to existing studies, our tests of the permanent income hypothesis are applied to a particular set of countries which have witnessed rapid and irregu lar changes in th e ir measured income and hence in the permanent and transitory components thereof. In addition, to our knowledge, this is the f ir s t study to test the permanent income hypothesis for the government sector (a ll previous studies having tested i t for the household sector only). In accom plishing these objectives, however, i t is most important not to take 44 the existing techniques for estimating permanent income for granted. Indeed, in this chapter, substantial doubts are expressed about the v a lid ity of using adaptive expectations techniques for estimating permanent income. In p articu lar, we test the v a lid ity of the fo l lowing restrictions which have been imposed by Milton Friedman and others in estimating permanent income: (a) that the marginal pro pensity to consume out of transitory income is equal to zero; (b) that the marginal propensity to consume out of permanent income is constant and stable over time; and (c) that M P C out of permanent income is the sam e as the average propensity to consume (APC). Before carrying out these tests and specifying the model, le t us b rie fly review the permanent income hypothesis and challenge the restrictio n s that are commonly imposed in estimating permanent i ncome. 4.1 The Permanent Income Hypothesis Milton Friedman (1956) was the original proponent of the per manent income hypothesis. H e c ritic iz e d a ll previous theories of consumption, especially the Keynesian view that household consumption is based on the household current income, and suggested instead that i t would be determined by what the household considers to be its « * future expected or "permanent*' income. The permanent income hypothesis, as formulated by Friedman him s e lf, can be described mathematically as follows: (i) 45 (4) where y£ is the permanent income, y| is the transitory income, c£ is the permanent consumption, p is the correlation c o e ffic ie n t, r is the interest ra te , w is the household wealth, and u is the ra tio of non hum an wealth to permanent income. In equation (3 ), Friedman assumed that K q = 0. This assumption was found to be consistent with the empirical evidence from long-run time series data (Kuznets, 1946). Also, Friedman assumed that the marginal propensity to consume (MPc) out of transitory income is equal to zero. Moreover, he assumed that the M P c out of permanent income (K j) would change over time as the values of r , w, and u would vary. However, when he and his followers estimated permanent income, they assumed that r , w, and u were constant over time, implying that K| is constant and has the sam e value over time. I t is found that the M P c out of measured income (b) in equation (5) below is related to the M P c out of permanent income (K j) in equation (3) ct = a + bYt (5) 46 ___________________________ i where the relationship between b and Kj is :* V a r ( Y P ) b = K. --------5 E __ = Kh Var(Y;) + Var(yT) 1 t ‘ (6) where ht < 1. Therefore, b is a downward-biased estimate of Kj. A commonly used means of estimating permanent income is the following adaptive expectations model Y t = e Y t + (i-e)(i+g)YS t t t_1 (7) where e is the adjustment co efficien t between measured and permanent income, and g is the average annual growth rate of income. 4.2 A C ritic a l Review of the Permanent Income Hypothesis and of the Technique for Its Testing In this section, the focus is mainly on the following technique and assumption underlying the permanent income hypothesis. In most empirical time series studies of the permanent income hypothesis undertaken in the la te 1960s and 1970s, permanent income *See Appendix A for the derivation of relation (6 ). ^Equation (7) is a discrete-tim e transformation of Friedman's formulation = 6 / ^ pQr ^ discpgtg.time case> Friedman used a weighted average of n period lags income as follows: ,P - s y + e i ' - v i y + e (B-g). ‘ - 2 - -t-„ 47 Y p - s , t + e ( e - g ) V i + e ( 6 . g ) + _ _ _ + ^ was considered to be a smooth function of measured income. Recently, however, som e e ffo rt has been made to use consumer wealth instead of measured income. For a more detailed discussion, see Darby (1974, 1975) and E llio t (1980). In p artic u la r, E llio t (1980) argued that the estimation of permanent income based on wealth measures provides better f i t than those based on current income: W e have found that permanent income measures based e x p lic itly on stocks of household wealth give an improved explanation of consumer spending compared to measures based on current realized income. A m ong various wealth measures, those that take into account the sometimes sharp cyclical variation in prices of stocks and bonds, generally explain consumer spending better than wealth measures that do not include these fluctuations. This suggests that som e net gain in explaining consumer spending is associated with cyclical movements in the market value of household wealth as compared to m uch less cyclical movement of standard permanent income measures. Our results support the idea that consumer expenditures react systematically and positively to market revaluations in the stock of household wealth. (pp. 529-530) In order to benefit from E llio t's finding, this study uses wealth as a substitute for measured income in estimating permanent income, thereby allowing the price of o il to be included e x p lic itly . This is important because in most O P E C countries, o il is the primary source of government revenues. Therefore, any change in the price of o il perceived by O P E C countries as a permanent change would be seen as affecting the value of th e ir wealth and therefore th e ir permanent income. Since in O P E C countries during the la s t decade incomes have been rather unstable, Income of the previous year would seem to be an extremely poor measure of permanent income. For example, from Table IV .l the government revenue of Saudi Arabia increased about 73% in 48 Table IV .l Government Revenue and Expenditure for O P E C Countries for the Period 1970/71- 1979/80* in Millions of National Currency Units Country 1979/80 1978/79 1977/78 1976/77 1975/76 1974/75 1973/74 1972/73 1972/73 1971/72 Algeria R evenue 36901 Expenditure 36881 Ecuador R evenue Expenditure Indonesia R evenue 3395.1 Expenditure 2482.3 Iran R evenue Expenditure Iraq Revenue 9027.7 Expenditure 9468.3 Kuwait Revenue 3241.3 Expenditure 2250 32565 27910 26190 32365 27750 23285 19054 16453 14653 22986 21889 16813 5301.6 4308.8 3689.8 5299.3 4305.7 3684.2 2796 2267.3 1769.3 2936 2390.2 1675.4 7169.4 5998.4 5045 7462 6339.1 5045 3803.6 3096.7 3033.1 1725.1 1750.4 1375.3 21995 14180 10310 21854 14173 i 10620 12273 11394 8088 11967 11410 9003 2733.5 1985.7 1171.7 2730.3 1977.9 1164.3 1626.8 1427.3 692.7 1614.2 1287.3 692.7 3202.6 2983.5 675.0 3640.8 2604.9 675.0 3960.4 2271.4 568.1 1032.6 1085.2 449.9 8702 7500 5819 8935 7750 6766 5514 4423 3699 7201 6813 5557 748.4 559.1 465.1 736.3 540.6 457.9 548.5 481.4 406.7 548.5 481.6 406.7 574.0 534.8 403.7 589.4 534.8 374.0 536.2 359.6 319.4 416.1 354.8 317.7 Table IV .l, continued Country 1979/80 1978/79 1977/78 1976/77 1975/76 1974/75 1973/74 1972/73 1972/73 1971/72 Libya R evenue Expenditure Nigeria R evenue Expenditure Qatar Revenue Expenditure Saudi Arabia R evenue Expenditure 160000 160000 2480 2480 2103 2103 5275.1 8042.4 9651.8 11695.4 1785 1785 6765.9 9701.5 1562.4 1562.4 5514.7 8258.3 1049.8 1049.8 4537.0 4260.3 736 736 1578.2 2575.0 United Arab Emirates (U A E ) Revenue Expenditure Venezuela R evenue Expenditure 50636 49902 8154.5 8927.2 7134.7 5496.6 1735.1 .7318 5808.9 5302.3 1931.1 1353.7 130000 146493 110935 130000 111400 110935 18410 18205 95847 110935 98247 45743 22810 22810 598.9 598.9 1400.8 2064.8 700 700 13200 13200 51179 52041 43140 39468 41000 40370 42799 40059 16433 15042 12547 12842 632.4 500.7 968.5 834.2 664 664 10782 10782 15015.2 14176.3 4001.8 2180.8 1651.3 11417.7 7763.8 3346.5 1735.6 1103.7 12123 11915 490.0 382.8 755.6 797.8 597 505 6380 6542 861.4 720.1 10252 10295 ♦Source O P E C Annual Statistical Bulletin (198 ) 1973/74, from 13.2 b illio n riy a ls in 1972/73 to 22.81 b illio n riy a ls in 1973/74, and then by 331% in 1974/75 when total revenue reached 98.247 b illio n . A s a result of the second jump in the price of o il in 1979/80, Saudi Arabia government revenue increased from 130 b il lion riy a ls in 1978/79 to 261.516 b illio n riyals in 1980/81. In other years, however, the increases in governmental revenue were modest. A s shown in Table IV .l, other O P E C countries have experi enced sim ilar degrees of income in s ta b ility . Given the u n re lia b ility of lagged income or even a weighted average of past measured incomes for estimating permanent income, wealth would seem to be a much more satisfactory basis for estimating permanent income. In such'a mea sure, the price of o il would have to be included e x p lic itly . Once again, however, because of the v o la tility of o il prices, i t would be important to distinguish between permanent and transitory increases in prices. Let us assume that actual price of oil at time t Pt ^as two components—a permanent component p£ and a transitory component T t pI , i . e . : p = pp + pT t r t * t Then for countries heavily dependent on o il, wealth can be approximated as the value of th e ir o il revenues, i . e . , as M t = pt R t (8) where p£ is the permanent price of o il at time t , and R -t is the quantity of o il reserves at time t . As a resu lt, the government's 51 permanent income can be considered to be proportionate to its wealth YP = a w t wt (9) where a is the average extraction rate . Under this d e fin itio n of permanent income, we would expect permanent income to rise sharply when there is a jump in the price of o il. The adaptive expectations technique spelled out in equation (7) above has been the most commonly used technique for estimating per manent income. A s noted above, the parameters e and g are normally assumed to be constant over time. The effectiveness of this tech nique in predicting the future values of som e variables has been c ritic ize d by many economists, especially by the advocates of rational expectations, such as Muth, Sargent, Wallace, Lucas, and others. For example, Cooley and Prescott (1973,76) and Lucas (1976) have argued that in any relationship which is characterized by adaptive expectation such that vt • F<Yt - i - xt ’ e ’s;t ) , , ( 10) where is a vector of exogenous variables, e* is a vector of parameters, and e^. is an error term. Although the usual assumption is that e and F are constant over tim e, i t m ay be more reasonable to assume that they vary over time. Cooley and Prescott, for example, allowed e to be a random variable, following the random walk: et+ l = 6t + nt+ l (11) 52 likew ise, Lucas has argued that econometric models based on stable coefficients are lik e ly to be inappropriate tools for long-term policy evaluations because, in re a lity , government policy and other exogenous events such as changes in technology would a lte r the structure of the model. Thus, while theory suggests i t would be appropriate to view behavioral relationships as varying over time, in practice empirical models have not done so. The sam e thing applies to the permanent income hypothesis. The estimation of permanent income by the adaptive expectations technique given by equation (7) or any other transformation thereof can be considered inappropriate i f , in re a lity , either the relationship or the parameters g and g are not constant over time, e .g ., i f the parameters were to follow the following random walk: .B + 1 ( 12) et+i = h + nt+i h n = gt + nt+ i (13) With respect to O P E C countries in general, the assumption concerning the s ta b ility of the relationship regardless of the differences in the structure of the economy between countries and th e ir stage of development would seem p articu larly u n realistic. The structure of the relationship such as in equation (7) might well be expected to change both over time (as a result of exogenous events) and among nations (because of differences in th e ir economic structures). I t would indeed be naive to use the same structural equation for two 53 countries having completely d ifferen t economic structures. Unfor tunately, many economists have apparently been unaware of this assumption in testing the permanent income hypothesis for d ifferen t countries and/or over time. Such an assumption would seem p articu larly inappropriate in applying the permanent income hypothesis to O P E C countries. There fore, this study tests the adaptive expectations technique to see i f relaxing the res trictio n of constancy of the parameters of g and & would improve the goodness-of-fit of the model. I t also uses current wealth as an alternative to adaptive expectations in estimating permanent income. Milton Friedman assumed that K 2 (the marginal propensity to consume of transitory income) is equal to zero in the following equation c+ = k.y* + ic y ! + u. t I t 2 t t (14) thereby allowing equation (14) to be reduced to ct - K i Yt + et (15) In the context of equation (14) with K 2 = 0, Darby (1974) showed that the estimate of e (the adjustment co efficien t) w ill be biased upward k2 by approximately (l-e )p — , i . e . , K1 54 *2 E(B*) = B + ( l - e ) * - 3 1 (16! In estimating the permanent income of O P E C countries, i t can be expected that the transitory income would have increased explosively due to the unexpected rises in the price of o il which took place in 1973-1974 and again in 1979-1980. Therefore, considerable doubt can be raised as to whether or not the expected value of tran sitory income was equal to zero, which implies that ^ might be sign ifican t and not equal to zero. This study tests for whether or not the assumption of K 2 = 0 leads to a biased value for 3 as claimed by Darby. In the next section, three d iffere n t models of permanent income are suggested for estimating the consumption function for som e O P E C countries and for testing the v a lid ity of som e of the restriction s which are commonly imposed by economists. The f ir s t two are adaptive expectations models, one of the type most commonly used by econo mists, and the other a maximum likelihood model. The third is a maximum likelihood model of the non-adaptive expectations type, where permanent income is considered to be a function of the two componentsj of wealth: namely, the price of o il, and the amount of o il reserves. The f ir s t model (Model I) has several objectives. The f ir s t is to estimate the consumption function for three O P E C countries, ^See Appendix B for the derivation of equation (16). 55 namely, Algeria, Kuwait, and Saudi Arabia4 by using adaptive expecta tions techniques in estimating permanent income. The second is to see whether or not the consumption function has a zero intercept. The third objective is to see whether or not the M P c out of tra n si tory income is equal to zero, and how that affects the value of the adjustment coefficien t g. The main objective of Model I I is to check the v a lid ity of the adaptive expectations technique in estimating permanent income and to see whether or not the relaxation of the assumption that 3 and g are constant over time improves the goodness-of-fit of the model. In addition, several other tests have been conducted, namely to retest Darby's proposition that the imposition of the assumption that M P c out of transitory income is equal to zero biases the estimate for g, and to retest the v a lid ity of the re s tric tio n imposed by Milton Friedman and his followers that M P c out of transitory income is equal to zero and that is constant over time.^ A s shall be demonstrated below, we find out that: 40ther O P E C countries are excluded from this study because time series data on government budget are not available for most O P E C countries. 5To achieve the above goals, a panel data were used, when w e observed ten O P E C countries over four years, and once we established our results that 3 and g are not constant over a short period of time. Therefore, and most lik e ly they are not stable over a longer period of time. Notice the value of the parameter for the second model are not comparable to those of Models I and I I I since they have d iffere n t time data of those in Model I I . Also in Models I and I I I the parameters are for individual countries, while in Model I I the parameters are for the group countries. __________________________________________________ 56. 1. In the presence of adaptive expectations, the assumption that l<2 = 0 should not be imposed because i t leads to a bias value for 8. 2. The values of both 8 and g are not,constant over time, suggesting that adaptive expectations techniques should not be used as the basis for estimating permanent income. Therefore, in Model I I , w e develop the alternative approach to measured permanent income by using wealth as the basis for estimating permanent income, where the price of o il was included e x p lic itly . A s a re s u lt, w e can test d ire c tly how the change in the price of o il affects consumption decisions and therefore saving decisions. This is then linked to the decision of how m uch crude o il to produce. F in a lly , w e use to test Friedman's proposal that (MPc = APc). 4.3 Model I The data used are annual time series data on government expendi tures and government revenues for three O P E C countries over time. There are 15 available observations for the three countries, namely, A lgeria, Kuwait, and Saudi Arabia (1967 to 1981). 4.3.1 Description of the Model The formulation of the model follows the frequently used frame work o rig in a lly advocated by Milton Friedman, where measured income, Yt , is assumed to have two components, permanent income, Y^, and the transitory income, y| . Yt =Yt +YI (AD 57 Also, measured consumption is assumed to have two components, permanent consumption c£ (which is assumed to be proportionate to permanent income), and a tran sitory consumption c£ (which is assumed to be proportionate to transitory income), P T Ct = Ct + Ct * C z (A2) or ct ■ V t + v l + et . (A2) * 2 where e^N (o ,a) . Permanent income is defined by the following adaptive expectations equation = *Yt + (l-e X i+ g J Y ^ j (A3) where the value of g can be determined by estimating the logarythmic trend logVt = T j + y2t * ut ( m ) 6 P where Yq = exp y^ and g = y2- 4 .3 .2 The Empirical Results of Model I I t has been found that the MPc's out of permanent income for A lgeria, Kuwait, and Saudi Arabia are, respectively, 0.54, 0.408, and ®This equation is used by both Darby and E llio t. Other workers used sim ilar techniques. 58 0.30, which are very sig n ifican t. See Table IV .2. These low values of M Pc m ean that saving out of permanent income is very high. The value of the marginal propensity to save (MPS) out o f permanent income is about 0.46 for A lgeria, 0.59 for Kuwait, and 0.7 for Saudi Arabia. While the M P c out of transitory income is found to be also very significant for a ll three countries, its magnitude varies widely from 0.12 for Kuwait to 0.72 for Algeria, implying that a high per centage but not the entire amount of transitory income goes to saving in the case of Kuwait and Saudi a Arabia, and only a very small percentage goes to saving in the case of A lgeria.7 N ow le t us assume that K 2 is equal to zero, thereby reducing the consumption function to the following form: ct ■ V t + ( A5) The regression results for this modified model are those shown in Table IV .3. Note that the value of K ^ has increased re la tiv e to those given in Table IV .2 in a ll countries by about 3.6% for A lgeria, 0.8% for Kuwait, and 8 % for Saudi Arabia. Also, note that the value of R 2 has fallen in a ll countries, and that the set of values of 6* = (.225, .005, .005) of Table IV .2 no longer minimizes the sum of squared residuals of the consumption function in specification (5 ). 7Notice that K 2 > Kj, in the case of Algeria which implies that when there is a jump in the price of o il which is unexpected, this generates a huge amount of transitory income, leading the country to consume more compared to less consumption when the jump in price is anticipated and considered to be a permanent one. .59 Table IV .2 Regression Results: Consumption Function of the Form, Ct * KjYt + K 2Yt + et* w **ere Values of g Are Predetermined and the Values of g Minimize the Residual Sum-square. A1geria Kuwait Saudi Arabia 6 0.225 0.005 0.005 g 0.2219 0.2317 0.365 0.543367 (48.5677) 0.408209 (31.6622) 0.299455 (17.0115) k2 0.720255 (4.4619) 0.121775 (6.2946) 0.221161 (3.6580) df 13 13 13 R 2 0.9977 0.9881 0.9827 60 Table IV .3 Regression Results under the Assumption That Kj = 0, where the Values of £ Are considered as Predetermined Given from Table IV .2. A1geria K uw a i t Saudi Arabia 6 0.225 0.005 0.005 9 0.2219 0.2317 0.365 h 0.579303 (49.6268) 0.41402 (16.6065) 0.3438 (19.6321) df 14 14 14 R2 0.9943 0.9517 0.9649 .61 Table IV .4 Regression Results under the Assumption that K 2 = 0, where the Values of 8 Minimize the Residual Sum-squares. Algeria Kuwait Saudi Arabia 8 0.95 0.325 0.575 g 0.221940 0.231683 0.365 Ki 0.552513 0.284388 0.276319 (75.6198) (39.7658) df 14 14 14 R 2 0.9976 0.9841 0.9912 Table IV .5 Regression Results: Consumption Function of the Form ct = K„ + KjY? + K 2Yt + et . Algeria Kuwait Saudi Arabia 6 0.225 0.005 0.005 9 0.221940 0. 0.365 K 0 242.0313 (5.9055) 289.4482 (1.2929) 915.56711 (1.686) *1 0.52089 (74.2683) 0.3873 (18.9129) 0.2874 (16.0055) h 0.615816 (7.0922 0.10934 (5.1646) 0.191428 (3.2299) df 12 12 12 R 2 0.9985 0.9676 63 The new values of B's which minimize the su m of squared residuals are 0.95 for A lgeria, 0.325 for Kuwait, and 0.575 for Saudi Arabia. See Table IV .4. But these new values of B's are upward-biased estimates of 3 * ' s when specification (2) is the true one. Our conclusion is that in estimating the permanent income con sumption function, the assumption that K 2 = 0 gives an upward-biased estimate for B*, which in turn affects the value of Kj. Therefore, the assumption that K 2 = 0 should not be imposed when adaptive expectations are used in estimating permanent income. The la s t test in this section is to see whether or not the consumption function has a zero intercept. From Table IV .5, i t can be seen that the intercept is positive in a ll three countries, but significant only in the case of Algeria. Therefore, the assumption of K q = 0 is rejected in the case of A lgeria, but accepted for both Saudi Arabia and Kuwait, thereby implying that the long run consump tion function does not necessarily always s tart from the origin (zero intercept) as suggested by Milton Friedman. Rather, i t might have a significant positive intercept in som e countries. 4.4 Model I I A s w e mentioned e a rlie r, the main purpose of this model is to test the v a lid ity of the adaptive expectations technique in estimat ing permanent income. In addition, som e other tests have been con ducted. To achieve th a t, w e formulate a maximum likelihood model which is estimated with LISREL techniques. See Appendix C. _ _ _ _ _ 64 I t has been assumed (a) that measured income has two components, permanent income and transitory income; (b) that there is no error in measuring income; and (c) that both permanent and transitory income changes over time and between individual countries. Y = vp + vT • , i t i t Yi t 1 = n, t = 1.................. T where is the measured government income for country i at time t , P T Y ^ and Y-jt are, respectively, permanent and transitory income. Also, the measured government current expenditure (consumption) has two components, permanent consumption Cpt and transitory consumption, c |t . ci . ■ + c L 11 U U (B2) where both C^t and are assumed proportionate to Y^t and Y|t , respectively. pP = K YP i t h i t (B3) CT - i/ i t 2 i t (84) where Kj and l<2 are the MPc's out of permanent and tran sitory income, respectively. For the time being, le t us use measured income as the basis for estimating permanent income, and for the sake of testing the adaptive expectations technique, le t us assume that w e have the following discrete time transformation of Friedman's original specification: 65 Yt = eYt + (i-e )(i+ g )Y ? , z z t - i where, as before, 3 is the adjustment coefficien t between measured and permanent income, and g is the average annual growth rate, respectively. By substituting for Y ^ . in equation (5) by using equation (6 ), we obtain ^ * e{yP ♦ v j, + ( i . 6) ( i +g)YP_. (B6) or ( i - b ) y J - b y J + ( l - s X i + g j Y j j Dividing equation (6*) by (1 -3 ), w e obtain (B61) Y t = ir fe T Yt + (1+9>Y£-1 (B7) or Y ^ = 8 Y ^ + « Y ? . t t t-1 (B7') In formulation (B7‘ ) , i t is assumed that there is no error term in case of time series data. But, since in this model w e use panel data for ten countries over time, i t is quite reasonable to assume that each country has a d iffe re n t permanent income. Therefore, specification ( B71) has to be w ritten as: Yi t = 0Yi t + aYi t i + it i,t-l i (B 8 ) where 6^ is a permanent residual capturing the variation in permanent 66 Income between countries. This residual is assumed to be normally 2 distributed with m ean equal to zero and variance o2 where 6,^N(o,a ) . 1 6 The transitory component of income at time t is assumed to be proportionate to the transitory income of the previous year, subject to an error term which is assumed to change over time and also between countries. V T _ yT T i t " Yt Yi , t - l + ni t (B9) where is the proportionate level which is assumed to be changing T 2 2 over time, and n ^N(o,o^) where a^T is changing over tim e, which is highly sig n ifican t. See Table IV .8. The above system of equations can be w ritten in LISREL notation as follows: 3n + 6 where 6^N and 0 0 0 m 0 2 °SP 0 0 0 0 0 0 0 0 0 0 0 0 - — — ----- — ----- — — —- 0 0 0 0 2 a *1 0 0 0 0 0 0 0 0 2 a n2 0 0 0 0 0 0 0 0 2 o n3 0 0 0 0 — ......................... — 0 0 0 0 2 o nt 67 where equation ( ! ' ) can be w ritten as follows: 71 P 72 P 73 P 74 71 T 72 T 73 T 74 71 P 72 P 73 P 74 T 71 T 72 T 73 T 74 T n3 The structural relationship between consumption and income is: Y = r , X 4x4 X 4x4 < 4' C K1X 4x4 | K 2X 4x4 1 or 5' Y C V 4.4.2 The Empirical Results of Model I I The data used in this study are panel macro data containing observations on government income and government current expenditures 68 | for ten O P E C countries over four years, 1971 to 1974.® Those coun- ■ trie s are Algeria, Ecuador, Indonesia, Iran, Iraq, Kuwait, Libia, Saudi Arabia, United Arab Emirates (UAE), and Venezuela. The O P E C countries are omitted from this study because of insu fficien t data are Gabon, Nigeria, and Qatar. O n the one hand, the long-run marginal propensity to consume (MPc 1 out of permanent income) is Ki = 0.787 (see Table IV.6 ) which is i significant at the 1 % level of significance; on the other hand, the ; marginal propensity to consume out of transitory income ^ is very I small, K 2 = 0.089, and insig nifican t, implying that O P E C countries ! save almost a ll th eir transitory income. The result is consistent 1 ! with Friedman's assumption about both MPc's. The value of the | adjustment coefficient is 6 = 0.022, which is very close to Darby's ; estimate (3 = 0.025), but i t is not sign ifican t, while the growth ! j rate in permanent income is very significant at 1% level of s ig n ifi cance, and its value is equal to g * 0.436. This high value of g can be attributed to the jump in the prices of o il which led to both i higher permanent income and transitory income, while the growth rate j of transitory income (y^.) exceeded that of the price of o il and i t is j very significant at 1 % level of significance. * Now le t us test for the assumption made by Milton Friedman that j people usually save a ll th e ir transitory income which means that M P c O The above period is chosen in order to study O P E C behavior over j the period of greatest disturbance, providing a re la tiv e ly strong j test for the v a lid ity of adaptive expectations as a tool for j estimating permanent income. out of transitory income is equal to zero. From Table IV .9, the ^ 2 value of X2 3 at 23 degrees of freedom is equal to 69.56 compared to 69.03 at 22 degrees of freedom from Table IV .6 . Therefore, this res tric tio n is accepted at the 1 % level of significance since the difference between the two values o f X ^ at one degree of freedom is 2 2 ^ x23"*22 ^ “ 6 9 * 5 6 “ 6 9 - 0 3 = 0.53, which is very small. Therefore, imposing this res tric tio n does not sig n ifican tly reduce the model's goodness-of-fit. In other words, during the period of rapidly increasing o il prices, saving was encouraged by two new favorable developments: the sharp increase in transitory income and the fact that most of i t was saved. Does that imply that i t is to the advantage of O P E C countries to have an explosive increase in the price of oil since i t leads to an explosive increase in saving? The answer, of course, depends not only on transitory income but also on adjustments to permanent income and in the M P c out of permanent income. From Table IV .7, i t can be seen that as a result of the price explosion of o il 1973-1974 the M P c out of permanent income increased from 0.728 in 1972 to 0.823 in 1973 to 0.877 in 1974. The interpretation is th at, due to the higher o il prices, government revenue in O P E C countries increased rapidly; since most O P E C countries were faced with absorptive capacity constraints, they would want to save only enough to finance th e ir investment projects. Budget surpluses would be useful only i f they would want to reduce th e ir debt and/or invest abroad. Since the former does not exist and the decision to invest abroad is extremely lim ite d , budget ■ ________________________________ .7 .0 surpluses are not of interest. In addition, to the extent that the O P E C countries were led to expect the price of o il to continue to ris e , th e ir expectations of higher prices would imply greater wealth and therefore higher permanent income, leading them to expect that th e ir permanent income would increase rapidly in the future. This would have the effect of increasing consumption and decreasing saving. A s a res u lt, the long-run marginal propensity could well decline as a result of the explosive increase in the price of o il. Therefore, there are two forces whereby o il price increases influence saving behavior: f ir s t , the positive short-run effect through transitory income, and second, the negative long-run effect through higher M Pc out of permanent income. The total effect can be viewed as the net resultant of these two effects. Therefore, to the extent that the decline in the long-run M P S dominates the rise in savings out of transitory income, the effect of o il price increases on savings w ill be negative. The problem would be aggravated i f the price jump were followed by a period of price s ta b ility or even worse i f i t were followed by a period o f price decline, especially i f unanticipated, implying that tran sitory income would be negative. This would lead to dissaving in the short run, while in the long run, lower prices (when expected for a considerable period of time) would mean lower permanent income (c e tris paribus). This would require consumption to go down. But, since consumption is rigid downward, i . e . , what Duesenberry (1949) called the "ratchet effect," ( illu s trated in Figure IV .l) , permanent income would fa ll from to Y^- 71 .p t l .P' ’tz Figure IV .1. Consumption function under expenditure rig id ity . n Since the governments of O P E C countries are reluctant to reduce th e ir consumption expenditures for many reasons, but especially for p o liti cal s ta b ility , consumption would not fa ll back along the long-run consumption function. Therefore, the fa ll in the permanent income due to the fa ll in prices of o il would cause consumption to f a ll not by the fu ll amount from to c£g but only from point A to point D, implying both that consumption fa lls less than proportionately and that saving fa lls considerably. In this part, we test the v a lid ity of the res tric tio n that K g = 0. The res tric tio n that K g * 0 is accepted at the 1 % level of significance since the value of x 2 3 increased just s lig h tly to 69.56 (see Table IV.9) compared to that of x ^ 2 2 ™ Table IV.6, an increase which is well below the c ritic a l value ofx^ at one degree of free dom. Even though this re s tric tio n does not reduce the fitness of the ^ model, i t should not be imposed because i t biases the value of 3 *. (From Table IV.6 we found that 6 * 88 0.25 but when we imposed that K g = 0, the value o f gincreased to 3 = 0.108, see Table IV .9, which is a bias estimate for e*.) Therefore, the results of this model support those of Model I wherein i t was found that the assumption that K g = 0 in adaptive expectation is unacceptable. Next, we test the assumption that Kj is constant over time. In Table IV.7, Kj is varying over time. The value of x § 2 dr°PPed "from 69.03 to xig = 58.75. Therefore, the relaxation of th is assumption that Kj is constant over time is accepted at 5% level of significance since the difference between the two x^ values at 3 degrees of 73 Table IV . 6 Maximum Likelihood Estimates of the Consumption Function c n . K l y P t + where the Values of e and g Are Estimated within the Model. 1971 1972 1973 1974 K1 0.789 (17.664) 0.789 (17.664) 0.789 (17.664) 0.789 (17.664) k2 0.089 (0.811) 0.089 (0.811) 0.089 (0.811) 0.089 (0.811) e 0 . 0 2 2 (0.176) 0 . 0 2 2 (0.176) 0 . 0 2 2 (0.176) a 1.436 (27.419) 1.436 (27.419) 1.436 (27.419) Y t 0.906 (4.173) 1.923 (3.852) 5.799 (12.484) a2 6t 15.33 (3.358) 15.33 (3.358) 15.33 (3.358) 15.33 (3.358) 2 a ° i t 18.833 (1.727) 7.804 (1.743) 52.026 (1.763) 254.599 (1.763) 6 0.0215 0.0215 0.0215 g 0.436 0.436 0.436 X 2 * 69. 03 df = 2 2 74 Table IV .7 Maximum Likelihood Estimates of the Consumption Function, = K^y P ^ . + K 2Yi t ’ w^ere K l * s Allowed to Vary Over Time. 1971 1972 1973 1974 K 1 0.77 (14.981) 0.728 (20.957) 0.823 (19.404) 0.877 (10.621) k2 0.101 (1.042) 0.101 (1.042) 0.101 (1.042) 0.101 (1.042) 0 0.009 (0.093) 0.009 (0.093) 0.009 (0.093) : a 1.342 (17.902) 1.342 (17.902) 1.342 (17.902) Yt 0.873 (4.7) 2.215 (4.659) 5.801 (12.40) 2 6t 9.633 (3.187) 9.633 (3.187) 9.633 (3.187) 9.633 (3.187) 2 a n i t 19.375 (1.778) 5.939 (1.764) 40.992 (1.843) 271.452 (1.874) S 0.00892 0.0089 0.0089 0.0089 g 0.342 0.342 0.342 0.342 X2 = 58.75 df « 19 75 Table IV .8 P Maximum Likelihood Estimates of the Consumption Function C i t = T + I^ Y ^ , where Psg = Psg = Ps7 = Psg. 1971 1972 1973 1974 K 1 0.794 0.794 0.794 0.794 (30.141) (30.141) (30.141) (30.141) K o 0.092 0.092 0.092 0.092 (1.75) (1.75) (1.75) (1.75) 9 0.019 (0.3114 0.019 (0.314) 0.019 (0.314) a 1.436 (28.469) 1.436 (28.469) 1.436 (28.469) ^ t 0.914 (2.73) 1.903 (7.726) 5.802 (38.775) 2 15.113 15.113 15.113 15.113 5 t (3.594) (3.594) (3.594) (3.594) 2 a 84.12 84.12 84.12 84.12 n i t (3.567) (3.567) (3.567) (3.567) 6 0.084 0.084 0.084 9 X 2 = 97.85 df = 25 0.436 0.436 ' 0.436 76 Table IV .9 Maximum Likelihood Estimates + and K 2 = 0* P of the Consumption Function * K].Yjt 1971 1972 1973 1974 *1 0.784 0.784 0.784 0.784 (18.68) (18.68) (18.68) (18.68) k2 0 0 0 0 6 0.122 (5.396) 0.122 (5.396) 0.122 (5.396) a 1.433 (27.438) 1.433 (27.438) 1.433 (27.438) Y t 0.898 (4.159) 1.941 (3.86) 5.794 (12.59) 2 15.626 15.626 15.626 15.626 6t (3.407) (3.407) (3.407) (3.407) 2 0 14.745 6.071 40.852 200.514 ni t (2.11) (2.113) (2.121) (2.121) e 0.1087 0.1087 0.1087 9 x 2 = 69. 1 df = 23 56 0.433 0.433 0.433 1 1 1 i • 1 77 j ........................................................ 1 Table IV .10 p Maximum Likelihood Estimates of the Consumption Function + KgY^., where a Is Changing Over Time where 0 Is Assumed Constant and Equal to 0.022. 1971 1972 1973 1974 K 1 0.792 0.792 0.792 0.792 X (21.874) (21.874) (21.874) (21.874) K o 0.091 0.091 0.091 0.091 c (6.266) (6.266) (6.266) (6.266) a 1.293 (16.699) 1.547 (20.374) 1.438 (28.703) Y t 0.91 (4.192) 1.912 (3.852) 5.801 (12.425) 2 o. 11.171 11.171 11.171 11.171 6t (3.329) (3.329) (3.329) (3.329) 2 a 18.945 7.876 52.217 255.598 ni t (2.090) (2.109) (2.133) (2.133) 6 0.0215 0.0215 0.0215 9 X2 = 64.37 df = 21 0.293 0.542 0.438 78 Table IV .11 P Maximum Likelihood Estimates of the Consumption Function = Ki^ it + I^ Y ^ . where ^ Is Assumed Equal to Zero, and 0 Is Assumed Constant and Equal to 0.022. 1971 1972 1973 1974 K1 0.789 (20.395) 0.789 (20.395) 0.789 (20.395) 0.789 (20.395) k2 0 0 0 0 e 0.022 0.022 0.022 a 1.453 (23.775) 1.453 (23.775) 1.453 (23.775) Yt 0.905 (4.168) 1.9242 (3.852) 5.799 (12.452) 2 6t 27.688 (3.484) 27.688 (3.484) 27.688 (3.484) 27.688 (3.484) 2 a nti 14.809 (2.101) 6.135 (2.113) 40.916 (2.121) 200.240 (2.121) g 0.0215 0.0215 0.0215 g 0.453 0.453 0.453 X2 » 86.31 df = 24 79 __1 Table IV .12 Maximum Likelihood Estimates + K2Y|t , where K2 Is Assumed Changing Over Time. of the Consumption Function Equal to Zero (K2 = 0), and Ci t - K l*U a Is 1971 1972 1973 1974 Ki 0.786 (22.136) 0.786 0.786 0.789 (22.136) (22.136) (22.136) k2 o 0 0 0 9 0.122 0.122 0.122 (6.128) (6.128) (6.128) a 1.298 1.524 1.44 Yt 0.901 1.933 5.796 (4.171) (3.858) (12.588) a2 11.872 11.872 11.872 11.872 t (3.357) (3.357) (3.357) (3.357) a2 14.771 6.099 40.881 200.948 ^ it (2.107) (2.115) (2.121) (2.121) e 0.109 0.109 0.109 g X2 = 65.8 df = 21 0.298 0.524 0.44 80 Table IV .13 Maximum Likelihood Estimates + KjyJj. where K 2 = 0 and both of the Consumption Function a and e Are Changing Over Time* - Ki yn 1971 1972 1973 1974 K, 0.791 0.791 0.791 0.791 (21.017) (21.017) (21.017) (21.017) k2 o 0 0 0 e 0.075 0.125 0.129 (0.286) (1.193) (6.996) a 1.296 1.535 1.438 (17.255) (21.377) (30.74) Yt 0.910 1.914 5.80 u (4.155) (3.843) (12.522) a2 9.378 9.378 9.378 9.378 6t (3.393) (3.393) (3.393) (3.393) 2 14.851 6.171 40.949 200.414 nt i (2.068) (2.114) (2.121) (2.121) S 0.070 0.111 0.114 g 0.296 0.296 0.535 0.438 X2 = 59.41 df = 19 81 freedom is (x22 * * 1 9 ) ~ 69.03 - 58.75 * 10.28 which is greater than the c ritic a l value at the 5% level of significance. In other words, by allowing Kj to vary over time, the explanatory power of the model is improved significantly. To test for adaptive expectations and constancy of both3 and g over time, le t us assume that bothBand g vary over time. As can be seen in Table IV.9, the value of x 2 decreases from 29.56 with 23 degrees of freedom to 59.41 with 19 degrees of freedom in Table IV.13. The difference between the two values o fx ^ is significant at the 5% level. Therefore, we reject the assumption that both and g are constant over time. Next, we relax only the la tte r assumption, allowing g to vary over time (but with 3 constant). This reduces the value of x 2 by only fc | 3 “ x l l ) = 69.56 - 65.8 = 3.76 which is not significant (Table IV.12). Therefore, allowing g to vary over time does not improve the fitness of the model. W hen we relax 3 , the constancy of Bassumption while keeping g constant, the explanatory power of the model is found to be improved significantly. Therefore, we concluded that the value of 3 should not be assumed constant over time and that we should not use adaptive expectations technique as the basis for estimating permanent income. In the next section, we suggest an alternative approach. 4.5 Model I I I The results of the previous model have shown that the use of adaptive expectations is unsatisfactory in the following respects. First, the values of the parameter, such as 3 and g, that are assumed 82 to be fixed over time in adaptive expectations models cannot be assumed to be constant without significantly reducing the explanatory power. Secondly, the assumption of K2 = 0 in the presence of adap tive expectations also w ill harm the fitness of the model since i t gives a bias value fo re *. Thus, the best cure for this dilemma is to avoid using adaptive expectations and to use wealth in estimating permanent income. In addition to what has been said above, Model I I I is formulated to achieve the following: 1. The permanent price of oil w ill be included exp licitly in the definition of permanent income. Therefore, this model should help us understand the permanent effect of a price change on consump tion behavior and therefore on saving behavior, and give us more insight into decisions regarding how much to produce of crude o il. 2. Other tests w ill be conducted, including one to determine the validity of the assumption that K2 = 0. 4.5.1 Specification of the Model I t is assumed that the decision on how much crude oil each coun try is going to produce each year is dependent on the long-run expected yield from its wealth y ^-W^, which is called the permanent income v^ = y W e i t Y1 i t (Cl) where W|t is the amount of expected wealth each country has at time t and is the extraction rate (interest rate). Now le t us define w®t as follows, ---------------------------- 83- w it= p nR n U U (C2) where t P ® is the expected price of o il for country i at time t , and is the amount of reserve of country i at time t. I f w e substitute equation (C2) into (C l), we obtain vn - n ‘ t pi . t - i R« > (C3) Assuming that the expected value of oil reserve is the only source of wealth, in order to capture other sources of wealth, i t is more appropriate to respecify (C3) as: Yi t = Yl^ tPi , t - l Ri t ) + uit (C4) where is a permanent income residual which is assumed to capture other sources of wealth or permanent income. This kind of assumption has been used in the literatu re in testing the assumptions of the permanent income hypothesis (e .g ., Atfield,. 1980). Now le t us assume that each O PEC country chooses its. price of oil (P jt) at tiroe t as follows: pi t ■ pt + ei (C5) where Pt is the O PEC price oil at time t and et is an error term 2 where etn H (o ^ ) . (C6) 84 Now le t us assume that each O PEC country forms its expectation according to the rational expectation criterion as follows: t Pi , t - l = Et^Pi 11Jt-1^ (C7) where Et (Pi t |It _j) is the conditional expectation of based on all available information at time t-1. By taking the conditional expectation of equation (C5), we obtain Et^Pi t l I t-l^ = Et * PtJ Tt-1^ + E*et l I t - l * (C8) P6 = ^ t i , t - l *t (C9) Then substituting equation (C9) into equation (C4) and rewriting in log form, we obtain Vi t = Y1 ln Pt + y2 1n ft1t (CIO) Transitory income is assumed to be a function of the growth rate of oil prices (GPt ) plus some error term nj added to reflect other factors (see Atfield, 1980): v n - Y i< G P t>+ "I (C ll) In order to link saving behavior with the decision of how much crude oil to produce, we specify the following supply function for o il: 85 ln q u = e In Pn (012) wher ei s the amount of oil production at time t . The rest of the model consists of the identities for both con sumption and income (equations 1 and 2 of Model I ) , and a consumption function of the form. Ci t = Kl Yi t + K2Yi t (C13) This model can be written in IISREL notation a mm m . r -i Pi t 1 0 0 in Pt yp i t = al 0 a2 in GPt + f t yT i t 0 a3 0 _m Rn_ or (C14) n = Tx + < 5 while the measurable equations are written in the form of (CIO). 'in q" vi t = _Ci t 0j 0 0 0 1 0 K 1 1 K, " 1 r+ * » pu i t > yt . n - (CIO) 86 4.5.2 The Empirical Results for Model I I I This model has been applied to the same three countries (Algeria, Kuwait, and Saudi Arabia), and the same years (1967 to 1981) as Model I. From Tables IV .14 and IV.18, i t can be seen that the MPc out of permanent income is highly significant in all countries, its value ranging from 0.281 in the case of Saudi Arabia to 0.527 in the case of Algeria. This implies that the M PS out of permanent income for the three countries range from 0.473 in Algeria to 0.72 in Saudia Arabia. From Tables IV.15 and IV.16, i t can be seen that the mar ginal propensity to invest (MPI) out of permanent income are 0.462 for Algeria, 0.134 for Kuwait, and 0.684 for Saudia Arabia. In all three, the MPI was greater than MPS, implying that they were running a budget surplus during that period. The amount of surplus varied widely from 0.011 in Algeria to 0.461 in Kuwait, suggesting that all three countries were faced with an absorptive capacity constraint. Therefore, our thesis that O P EC countries decide on how much crude oil to produce by maximizing their u t ility functions subject to both their budget and absorptive capacity constraints, would imply a negative relationship between the amount of oil production and gov ernment permanent income. In Table IV.17, i t is shown that the value of 02 is negative in a ll three cases and is very significant in the case of Kuwait which had the highest budget surplus, but not s ig n ifi cant in the case of Algeria which had the lowest budget surplus, where 02 is defined as dqt e2 = ^ F /q t t __________________ 87_ This suggests that the absorptive capacity constraint plays an impor tant role in determining how much crude oil each country should produce. It was also found in this model that the expected price of oil (^Pt_i) explains permanent income very significantly, and i t is positively related to permanent income in all three countries, while the amount of reserve in each country contributes l i t t l e in explain ing permanent income and its value is insignificant. Additionally, we found that in all three countries, the MPc out of transitory income was insignificant (see Table IV.15) and the assumption that K 2 = 0 is accepted because it does not harm the goodness-of-fit of the model. In general, this model fits both Algeria and Saudi Arabia very w ell, but is a very poor model in the case of Kuwait (even the model does not converge). The next thing to do is to test whether or not the way we define our permanent income affected the results we obtained, or even affect the goodness-of-fit of the model. Let us define permanent income as just proportionate of expected wealth as follows: yt = ° i wt + 6t where is the extraction rate. The maximum likelihood results under this new definition are shown in Table IV.18. I t is found that the values of in all three countries remain the same, which implies that the way we define permanent income has no effect on the values of K ^ in all countries. From the same table, Kuwait had the lowest exploitation rate (0.8%), then Saudi Arabia (1.7%), while 88 Algeria had the highest rate (4.7%); all these values are highly significant. It is clear that all extraction rates are much lower than the nominal interest rates, and their values change between countries. This does not support the assumption that the interest rate and the extraction rate are identical and fixed for all coun tries as is usually assumed under the theory of exhaustible resources. This model fits the Kuwait data very well, but is not a good model for Saudi Arabia. The implication is that there is no unique definition of permanent income which fits all countries. 89 Table IV .14 Maximum Likelihood Defined as Yt = Zero Results where the 1n + “ 2 ln Permanent Income R t + and K2 Model Is Is Assumed to Be A1geria Kuwait* Saudia Arabia K1 0.527 0.412 0.281 J L (98.529)^* (23.021) k2 0 0 0 e 0.042 0.007 0.363 (0.882) (7.570) “0 1 1 1 °1 4.381 46.497 22.985 X (9.354) (7.49) az 0.578 -298.781 1.559 (0.323) (0.133) 0.214 11.046 4.239 (1.825) ... (2.192) 2 °1 18.069 4237.22 0.355 X (0.435) (2.169) 2 a 2 2.772 266.56 115.566 c (2.642) (2.169) 2 °3 0.028 2268.503 22.369 J (2.646) (2.646) X 2 16.6 69.09 11.39 df 7 7 7 ♦The model does not converge in the case of Kuwait. ♦♦Figures in parentheses indicate T values 90 Table IV .15 Maximum Likelihood Results where the Permanent Income Model Is 6 Defined as Yt = In t pi , t - l + a2 " *n Rt + 6t Algeria Kuw a i t * Saudia Arabia K1 0.527 0.361 0.280 1 (92. 37)** (21.868) k2 0.281 -0.08 0.030 0.972) (0.156) e 0.042 0.007 0.363 (.882) (7.570) °b 1 1 1 4.349 54.174 23.047 X (8.973) (7.477) °2 0.342 -204.051 1.601 C m (0.193) (0.136) 0.461 13.185 3.828 (0.744) (1.256) 2 18.069 4216.843 0.355 X (0.435) (2.169) 2 °2 0.130 1570.677 18.283 u (0.422) (0.777) 2 °3 0.130 1570.677 18.283 (0.422) (0.777) 7? 16.41 66.62 11.35 df 6 6 6 ♦Figures in parentheses indicate T values Table IV .16 Maximum Likelihood Defined as Yt = o tj Function Is Defined Results where the ln p? ,t-l + “2 1n as I = with Permanent Income + 6t and the k2 = 0. Model Is Investment A 1 geria Kuwait^ Saudi a Arabia K1 0.462 (54.717)♦♦ 0.152 0.684 (22.885) k2 0 0 0 e 0.042 (0.882) 0.007 0.363 (7.57) % 1 1 1 °1 4.359 (8.363) 43.573 23.384 (7.412) °2 1.071 (0.542) -99.318 -2.271 (-0.189) °l3 -0.258 (-1.219) 26.207 6.271 (3.09) 2 °1 18.069 (0.435) 4286.879 0.355 (2.169) *>2 3.384 (2.633) 370.003 120.938 (2.572) CM 0.093 (2.646) 2153.186 24.716 (2.646) i? 25.85 57.43 16.16 df 7 7 7 ♦♦Figures in parentheses indicate T values 92 Table IV .17 Maximum Likelihood Results where Yt = P and In qt = In P + 62*t» Hhere K2 = °1 ln t p? ,t-l 0. + ln Rt + 6t2 Algeria Kuwa i t* Saudia Arabia K1 0.527 0.405 0.281 (98.529)* (4.969) 23.021 k2 0 0 0 el 0.079 0.15 0.578 (0.593) (1.213) (4.872) 0 2 -0.009 -.009 -0.009 c (-0.295) (-3.087) (-1.933) a Q 1 1 1 a l 4.381 47.009 22.985 (9.354) (4.567) (7.49) a o 0.578 -302.072 1.559 c (0.323) (-1.876) (0.133) a. 0.214 13.01 4.239 0 (1.825) (0.693) (2.192) 2 a l 5.153 1.381 0.11 (0.295) (0.591) (2.192) 2 2.772 305.316 115.566 C (2.642) . (1.011) (2.579) 2 a o 0.028 2226.594 22.369 0 (2.646 (2.646) (2.646) X 2 16.52 41.89 (2.646) 8.05 (2.579) df 6 6 6 ♦Figures in parentheses indicate T values 93 Table IV .18 Maximum Likelihood Results under the Assumption that K 2 P Permanent Income Is Defined as Yt = + 6t . = 0, and that Algeria Kuwait Saudi Arabia K i 0.527 0.405 2.81 1 (100.090)* (4.993) (23.422) k2 0 0 0 al 0.047 0.008 0.017 (10.671) (4.863) (13.565) 0 2 0.214 13.01 4.239 (1.853) (0.696) (2.23) 2 al 2.348 118.354 43.22 (2.642) (1.816) (2.581) 2 a2 0.028 2226.594 22.369 (2.646) (2.646) (2.646) 2 °3 7.15 0.38 13.36 df 2 2 2 ♦Figures in parentheses indicate T values 94 Table IV .19 = vl Maximum Likelihood Results of the Investment Function (I * K*Yt ) 1 2 under the Assumption that K* = 0, and that Permanent Income Is P Defined as = a^wt + 5t* Algeria Kuwait* Saudia Arabia *1 0.462 (55.464)* 0.134 (5.263) 0.684 (22.239) k2 0 0 0 “1 0.048 (12.356) 0.008 (4.444) 0.017 (11.861) a2 -0.258 (-1.236) 27.68 (1.513) 6.271 (3.140) 2 °1 1.748 (2.634) 717.614 (1.866) 59.429 (2.574) 2 °2 0.093 (2.646) 2137.922 (2.646) 24.716 (2.646) 2 X 3.46 2.11 17.49 df 2 2 2 ♦Figures in parentheses indicate T values 95 CHAPTER V THE UTILITY MAXIMIZATION MODEL There remains no sensible theory which might offer a satisfactory explanation of O PEC behavior. Most of the suggested explanations for O PEC behavior have, in the last decade, proven unable to even approx imately predict O PEC behavior, and as a result, there remains wide spread disagreement on how to model O PEC behavior. Gately (1984) argued that, I f the 1970's were not good years for macro economists, they were not alone. Microeconomists specializing in world oil market fared l i t t l e better. Their predictions were not particularly accurate, and the orthodox theor etical approach proved not very useful. Ten years after the 1973-1974 oil price quadrupling there remains much disagreement about what happened and what we can expect in the future . . . . There are numerous important issues, both empirical and theoretical, that remain unsolved. With regard to empirical questions, there are important uncertainties about both demand and supply. With regard to theoretical issues, i t remains an open question how best to design a model of the behavior of OPEC. This is true especially for the behavior during disruptions but also for decisions in more normal times, concerning production levels, pricing, and capacity changes. There are a large number of alternative the ories, but a m uch smaller number of sensible applied models, (pp. 1100, 1113) In this chapter, an attempt is made to suggest a more sensible theory which might offer a better explanation of O PEC behavior. As 96 we stated in previous chapters, O PEC countries are concerned about maximizing their u t ilit y function subject to their budget con straints. The main objectives of this chapter are to derive the supply curve, to determine the conditions for having a backward- bending supply curve, and then to determine the factors which influ ence the decisions on how much crude oil to produce. A u t ilit y maximization model is used in order to achieve the above objectives. In the firs t section a u t ility model is displayed and the supply curve is derived, determining the condition to obtain a negatively sloped supply curve. In the second section, the supply curve is derived under the assumption that the representative country's total expenditure is less than or equal to its total revenue, and hence that i t is faced with an absorptive capacity constraint. In the third section, the country is assumed to be running a budget d efic it and having irreducible or subsistence level of consumption expenditure; once again, a supply curve of crude oil is derived. 5.1 The General U tility Maximization Model It is assumed that the representative O P EC country at each period of time t is concerned with maximizing its current consumption Cl t , and future generation consumption, in the form of future investment income rt C2^, and future oil wealtht t R t+j. Then the objective of each country is to maximize its u t ility function of the form u = u(Cl t ,r t C2 t,Tt Rt+1) (Dl) 97 subject to its budget constraint where oil revenue is the main source of government revenue. I t represents about 75% of the total revenue in most O PEC countries, while only about 25% comes from other sources such as taxes, ta r iffs , profits from previous investments, etc. Therefore, the budget constraint for period t can be written as follows. Cl t + C 2t = Tt (Rt " Rt+1) + Yt which can be written as Cl t + C2t + Tt Rt+ l ” Tt Rt + Yt ( 02) (D3) where F i x R t + r t (D4) and R t+1 5 Rt " Q t (05) where Cl t = real current consumption at time t C 2t = real investment expenditure at time t = the amount of oil reserves at the beginning of period t which is assumed to be given Rt+1 = the amount 0 1 *1 reserves at period t+1 Q |. = total production of oil at period t = the price of oil at time t rt^2t = 'flJture rea^ investment income 98 rt = the interest rate at period t Yt = non-oil revenue in period t . Then the objective of the representative O PEC country is to maximize its u t ility function subject to its budget constraint. M a x u = u (C u,r - tC 2t,T tR ttl) s .t Cl t + C2 t + Tt Rt + 1 = F w h e r e C lt’C 2t,R t+l > 0 Then the solution to this maximization problem w ill give the following three demand functions. C lt = fl^ T t*rt,F ^ C 2 t = f2 ^ T t,rt’F ^ Q t = Rt " ^ Tt ,r ,F ) (D 6 ) (07) (D8) (D9) (010) Then from equation (DIO), substituting for Rt+j in equation (D5), the supply curve is obtained. Rt+i = ♦ < V rf F) (O il) 99 To know the slope of the supply curve, take the partial derivative of equation (A.11) with respect to obtaining 8Q 3t i = . c + i t 3F 1 fa# =- 3d,' t k SF ' k R‘ > 0 (D12) aQ. The sign of t is indeterminate. I t might be positive or 8Tt negative, depending on whether Rt+1 is a normal good or an inferior good. The necessary but not sufficient condition to have a nega tiv e ly sloping supply curve is that must be a normal good. Otherwise, the supply curve w ill be positively sloped. I f the country were running a budget surplus, then Rt might well 3 ( { ) r\ be considered a normal good, so > u allowing a negatively sloped supply curve to be generated. However, i f i t were running a budget a <i > p . d e fic it, then R t+j is considered an inferio r good so gf < u , which 3Qt i implies that Then, we have a positively sloped supply curve.1 dTt * I f the country were running a budget surplus, then when govern ment income increases, this would generate more surplus; then i t is expected that the country would reduce its surplus by reducing which means increasing Rt+j . Therefore, we expected that Rt+j is positively related to government income. By the sam e token we can show Rt+i is negatively related to government income i f the country is running a budget d eficit or, in other words, R t+1 is an inferior goods. 1 - 0 0 Roughly speaking, the higher the price of o il, the more likely the representative O PEC country would be running a budget surplus and hence the more likely i t would have a negatively sloped supply curve of crude oil as shown in the upper portion of the supply curve, S. in Figure V .l. O n the other hand, the lower the price of o il, the more lik e ly a budget d eficit and a positively.sloped supply curve as shown in the lower portion of the supply curve, S. 5.2 Derivation of the Supply Curve for a Cobb-Douglas U tility Function The objective of this example is to show that having a normal good is a necessary but not sufficient condition to have a backward-bending supply curve. Let us assume that the u t ility func tion facing a representative OPEC country is of Cobb-Douglas form where Rt+j is a normal good. Then this country will maximize its u t ilit y function. u = a ln Cu + 3 In r t C2t + 6 In xt Rt+1 (E l) s -t (E2) The first-order conditions are a v = 0 C It (E3) - y = 0 2 t (E4) 101 Tt Figure V .l. The backward-bending supply curve (when wealth is considered as a normal good). D yr, = 0 Kt +1 1 Then solve for the three demand functions:^ d - «F " t+1 Tt (a+g+5) r a F I t Ta+pfiJ C = BF 2t Ia+6+6) (E5) (E6) (E7) (E8) The supply curve is: 0 = R . ^ t t x^TB+a+fiT but since 8Q 3t (E9) the slope of the supply curve is 8Q 37 t _ - ( a+B+6)F . (ct+e+<S)<STt Tt (a+B +6)2 •^(a+B+S)2 \ -(a + B + 6 ) ( F - t R^.) 4Yt ] _T(a+B+6)2 t ^ ( o+B+5)_ > 0 (E10) See Appendix A for the derivation of the demand functions. 103 This generates a positively sloped supply curve. I t is not always true that being Rt+j as a normal good generates a negatively sloped supply curve, because i t is a necessary condition, not a sufficient one. To have a negatively sloped supply curve the income effect must dominate the substitution effect.3 5.3 Absorptive Capacity and the Supply Curve Most O PEC countries had either balanced or surplus budgets during the period 1973 to 1982; most experienced huge increases in spending on their development plans. Their aim was to achieve the highest possible level of economic development in the shortest possible time, but their efforts were soon crippled by their absorptive capacity constraint. As was pointed out in the previous chapter, in all three OPEC countries investigated (i.e., Algeria, Kuwait, and Saudi Arabia) the marginal propensity to save (MPS) is greater than the marginal propensity to invest (MPI). The greatest difference between M PS and MPI, namely, more than 45$, was observed in Kuwait, implying that this country was running huge budget surpluses with revenues being generated far in excess of its absorptive capacity. Not surpris ingly, therefore, Kuwait has reduced its supply of oil considerably in the last decade. Accordingly, le t us assume that the government's total revenue is equal to or greater than its total expenditure. In addition, le t us * Notice that u t ility functions of the Constant E lasticity of Substitution type where Cobb-Douglas is a special case with o = 1 do not give rise to a backward bending supply curve (see Hanoch, 1965). 104 assume that this country is faced with an absorptive capacity con straint; therefore the amount of C 2 £ should be less or equal to the absorptive capacity I* . Then the optimization problem (Kuhn-Tucker) condition is that Max u = . In Cl t +Bln rt C2t + «xt Rt+1 (FI) s .t. (F2) (F3) "Clt* -C2 t * "Rt+1 = 0 (F4) or = a In + 3 ^n^2t + ^ Tt Rt+l + yl F “ Cl t " C2t ' Tt Rt+l ] + -2 [l* - C2 t] (F5) The solution for this problem is that (F6) C 2t - 0 (F7) 105 Rt+1 3R^t l = 0 Rt+1 = 0 Now le t us assume that Cjt , C 2 ^» ^t+1* an(1 y i are a^ s tric tly positive; then the first-order conditions are: ■c2- - yl = ° ul t 1 8 r v > i - yo = 0 4 t L * Rt+1 "1H Cl t + C 2t + Tt Rt+l ~ F = 0 C2t = ** Then, solving the demand function for R^+x* we obtain^ < s (’rt Rt + Yt " Rt+1 Tt(a+6) 4For derivation of that, see Appendix B. (F8) (F9) (F10) (F ll) (F12) (F13) (F14) (F15) 106 The supply function is given as follows with slope given by 3Qt 6(a+6)(Y. - I*) 1--------- $ o 8xt T^.(a+6)^ The condition for having a negatively sloped supply curve is then I* > Yt . Because the percentage of Y t in the total revenue is around 25% in most O PEC countries, while the percentage of I* in the total budget is about 50%, we expect th a t(Y -I*)< 0 which implies that •« t ^ < ° a negatively sloped supply curve.5 Therefore, in the presence of the absorptive capacity constraint, most lik ely the country would be faced with a negatively sloped supply curve as long as the country is dependent on oil revenue to finance its budget; or more precisely, its non-oil revenue is less than its planned investment expenditure. 5.4 Budget D eficit with a Lower Limit on Government Expenditures and the "Supply Curve" The problems of OPEC countries were aggravated when the price of oil started to decline in 1983, making O PEC countries reduce their price of oil to $29/barrel and causing their total revenues to go 5This is true only when the constraint is binding ( i . e . , when desired investment is more than the absorptive capacity one I * ) . 107 down. Because government expenditures are rigid downward, at best they reduced their consumption by less than in proportion to their income reduction, leading to a higher M Pc and therefore a lower MPs. The total government revenues of most O PEC countries have fallen by more than government spending, thereby bringing about deficits begin ning in 1983. In such circumstances, individual O P E C countries tended to feel their fiscal deficits could be treated only by increasing their production of o il. In order to do so, they fe lt they would have to lower the price of oil s till further in order to compete more effectively in the international market. This reduced the a b ility of O PEC as an organization to influence the price of oil and i t became an even more d iffic u lt task for O PEC to maintain the price of oil at its original level of $29/barrel. Nigeria was the fir s t O PEC country to announce its intention to lower its price of oil on October 15, 1984, in order to compete with Great Britain and Norway which had lowered their prices for the same gravity of crude oil the week before. O PEC as an organization called for an "emergency meeting" in Geneva, trying to convince Nigeria to cancel that move. At the end of that meeting, they issued a com munique in which they agreed to cut their production of crude oil by 1.5 mb/d from 17.5 mb/d to 16 mb/d to maintain the price of oil at $29. I t proved d iffic u lt for som e O PEC countries to adhere to the agreement and as a result, production remained at the same level as before. Other O PEC countries were upset about the non-adherence to the agreed-upon production cuts by the others and realized that the 108 lowering of the price of oil was necessary in order to compete in the international oil market and enable its members to f u lf ill their financial obligations. This is what happened and O PEC lowered its price of crude oil on January 31, 1985 by $1 to $28 per barrel for the Arabian light. In order to model what we mentioned above, le t us assume that we have a Stone-Geary u t ilit y function where the representative country is running a budget d e fic it, and at the same time that i t has to spend at least Yjj. on current expenditure, and Y2t on investment expenditure. Then the u tility function facing the representative country is u = ax ln(Cl t - yu ) + a 2 ln rt (C2t • y 2 t ] + a3Tt Rt+ l (Gl) where 1> a^> 0, y j t , and C2t— Y 2 t * Y l t subsistence level of consumption goods, and Y2t is the subsistence level of investment goods. Then the objective of the representative country is to choose Cit» C 2 £, and at time t to maximize the above preference function subject to its budget constraint Cl t + C2t ± Tt (Kt ' W + Yt (G2) By solving the above optimization problem, the following demand functions are obtained:^ %ee Appendix C for the derivation of the demand functions which implies that the supply curve is negatively sloped. Hence, i f 109 ^ lt Yl t + T“i +a2+0l3^ ~ ^ Yl t + Y2t^ (G3) C2t r 2 t + l a ^ a ^ a , ) ^ " ^ Yl t + Y2t^] 1 2 3' (G4) ^ t + l T i.(a1+ a s>+ a ~ ) 1 2 3 (G5) Then the supply curve is " Rt ' Tt (ai+o2+a3) [ F ‘ (Yl t +Y2t>] 9Q t _ 9t , < “ l + _ “2+“3)“3TtR t‘ ( c ' l +“2+a3)"3 [F ' [ Tt ( “l +02+“3)] a3^al +a2+a3^ lTt R t ‘ t t Rt ‘ Vt + “3l(Yl t +Y2t> - Yj > 0 T ( a 1+ a 2+ a 3 ) Given the importance of oil revenues to most O P E C countries, we expect that (Yl t + Y2t) > Y t 110 which implies that the supply curve is negatively sloped. Hence i f their budget is sufficiently tight as to cover what they regard as their subsistence level of expenditures only with d iffic u lty , O PEC countries would have to increase, their production of oil when the price of oil goes down. The demands for both cl t and C 2 t are positively related to the price of o il, and negatively related to their own prices as shown in Fi gure V.2. Now let us assume also that O PEC countries try to stabilize the price of oil at its new lower level by cutting back the production of o il. This policy is unlikely to work i f most of the O PEC countries are running a budget deficit and at the same time i t is d iffic u lt for them to meet their subsistence level of expenditure. The problem w ill be aggravated when a country has already reduced its expenditure to what i t regards as the subsistence level. To get out of this dilemma, an individual country would be lik e ly to violate its price agreement so as to increase its production of oil and to raise its government revenue. This would seem to be exactly what has been happening in 1985. Our conclusion is that O PEC can be a powerful organization when its members are running a budget surplus; on the other hand, when its members are running deficits, i t is hard for O PEC to enforce either a price or production policy, le t alone both. Ill Figure V.2. The demand curve for both consumption and investment goods for a Stone-geary u t ilit y function. 112 Chapter VI SUMMARY, CONCLUSIONS, AND SUGGESTIONS FOR FUTURE RESEARCH 6.1 Summary and Conclusions Existing models have failed to provide satisfactory explanations of O PEC behavior. As the theoretical and empirical shortcomings of existing explanations have become more and more obvious, i t has led to increasing frustration am ong specialists in this fie ld . For this reason, an alternative " u tility maximization" model of O PEC behavior is developed and applied in this dissertation. Our study of the historical record of O PEC countries' behavior has led us to the following conclusions. 1. O PEC as an organization did not act as a cartel in the traditional sense in the last two decades. Therefore, both the 1973- 1974 price quadrupling and the 1979-1980 price doubling cannot be explained in terms of deliberate actions on the part of O PEC to regulate the production of o il. The decision of how m uch crude oil each country should produce has, at least until very recently, been le ft to the self-interest of each individual country. 2. The behavior of O PEC countries is inconsistent with the theory of the exhaustible resources, and there is growing evidence 11 that O PEC countries are not maximizing the discounted value of their profits. Therefore, most of models of this genre are inappropriate. 3. A more plausible alternative would seem to be that O PEC countries maximize their u t ility functions defined in terms of both present and future consumption subject to their budget constraints. I t is in this way that O PEC countries determine how much crude oil to produce and how much to keep in the ground for future generations. 4. The absorptive capacity constraint plays an important role in the decisions of how much crude oil to produce. As a result of the price shocks in 1973 and 1979, som e O PEC countries found them selves with amounts of savings which were beyond their capacities to absorb effectively, thereby leading them to cut back their production of oil so as to eliminate any surplus in their budget. Therefore, attention is paid to government saving behavior in O PEC countries and for the following reasons, to the permanent income hypothesis. 1. This hypothesis is consistent with the lif e cycle u tility maximization approach advocated in this dissertation. 2. The distinction between short-run and long-run saving func tion can be made much clearer by using the permanent income hypothesis approach. 3. The price of oil can be included e x p licitly under this type of model, and the effect of the price of oil on saving can be studied and tested directly. 114 6.2 Main Findings 6.2.1 Saving Behavior (A Permanent Income Model) Three different models of permanent income are used in empirical estimation of the consumption function for som e O PEC countries. The fir s t two (Models I and I I ) are adaptive expectation models, each with different features and using different techniques for estimating permanent income. The results of these two models are as follows: 1. The assumption that the marginal propensity to consume out of transitory income is equal to zero (K£ = 0) led to a biased e s ti mate for g* (the expected value of the income adjustment coefficien t). 2. The adaptive expectations technique for estimating permanent income is rejected, and therefore both the income adjustment coeffi cient (e) and the growth rate of permanent income (g) should not be assumed constant over time. The above results lead us to the following conclusion: in order to improve the s u itab ility of the model and to avoid the shortcomings of using adaptive expectations for estimating permanent income, a new approach should be used where the expected price of oil could be included explicitly in the model. Therefore, in Model I I I we have used a maximum likelihood technique for estimating permanent income, where permanent income is defined as the log of the two components of wealth, namely, the expected price of oil and the amount of oil reserve and where the expectation is formed according to the rational expectation criterion. This model fits the data for both Algeria and 115 Saudi Arabia very well, but is not a good model for Kuwait. O n the other hand, when permanent income is defined as a proportion of total wealth, the model fits the Kuwait data very well. W e found that the results of this model support the results obtained from Models I and II that the marginal propensity to consume out of transitory income can be assumed to be zero. In addition, i t was found that, at least during the period under investigation (1967-1981), the marginal pro pensity to invest out of permanent income (MPI) was higher than the marginal propensity to save (MPS) in all three countries. This led us to believe that these countries were faced with absorptive capa city constraints. Also, we found that the quantity of oil produced responded negatively to permanent income which suggested that O PEC countries determine their production of oil with reference to their budgetary needs which are determined by u t ilit y maximizing behavior. W e also found that the assumption that Kj is constant over time should be rejected. W e found that the jump in the price of oil in 1973-1974 had a negative effect on long-run savings since i t led to a higher M P C out of permanent income. Therefore, this explains why many O P EC countries started witnessing a budget deficit in the last three years. Specifically, when price started going down after 1983, this reduced permanent income. With government expenditures rigid downward, however, the effect was a reduction in the long-run MPS. 6.2.2 The U tility Maximization Model The objective of Chapter V was to determine the conditions in flu encing the decision of how much crude oil each country should pro duce. The oil supply curve has been derived under different conditions, and the following conclusions have been reached. 1. The necessary but not sufficient condition to have a nega tively sloped supply curve is that each country must consider its future reserves of oil ) as a normal good. 2. W hen an O PEC country is faced with an absorptive capacity constraint ( i . e . , the desired level of investment I is greater than its absorptive capacity constraint I * ) , we expect that the supply curve is negatively sloped as long as ( I* - Y) < 0; or i f a country is highly dependent on oil revenues to finance its investment expenditure. 3. Additionally, i f a country is running a budget d e fic it, where i t is hard for i t to finance its minimum requirements of both current expenditure and investment expenditure (subsistence le v e l), then the country has to lower its price of oil in order to expand its produc tion of oil to generate enough revenues to f u lf ill its financial obligations, which generates a negatively sloped supply curve. 6.3 Suggestions for Future Research Admittedly, this dissertation makes but one small step in what we believe to be the right direction; there is certainly much room for extending and improving the model. One desirable extension could be to study the effects of changes in the interest rate on oil 117, production decisions. Also, this model can be extended so as to constitute an overlapping generations model for use in studying saving behavior over time; where the distinction between foreign and domestic saving can be made. Also, the marginal propensity to con sum e could formally be derived from a life cycle u t ility model which would be an improvement over the Friedman permanent income hypothesis model used here. Different versions of this model could be used to determine a price path for O P EC which maximizes the u t ility function of its members, subject to their budget constraints, where the demand for consumption goods (C^t ) is a function of both the price of oil and permanent income. Also, the marginal propensity to consume could be derived as the partial derivative of consumption with respect to permanent income. MATHEMATICAL APPENDIX APPENDIX IV.A Y = t Tt Yt ( I ) 1 rP = k y* 5 Lt V t ( 2)' Ct = a + bYt Then to get the value of a and b which minimize the least squares criterion function 4 ', J(a,b) = E(Ct - a - byt ) The fir s t order conditions for minimizing 4 are (3)' |i = 2E(Ct - a - byt ) = 0 | i = 2E([ct - a - b y jy ,} = 0 Equation (4)' implies that a = ECt - bEYt Substitute (7)' into (6 )', we obtain e( [Ct - Ect + bEy - byt]yt] ■ 0 = 0 E ((ct - E C t]yt) - b E ([yt - E y t]yt or E[ l Ct ' ECt ] [ yy ' Eyt ] ) ' bE([yt - Eyt ] [ yt or Cov (Ct ,yt ) - 6 Var(yt ) -EytJ = 0 (4)' (5)1 ( 6)' (7)1 120 which Implies that cov(Ctyt ) b - var(yt ) (8)' From equation (1)' var(Yt ) = var(Y^) + var(Y^ ( 9 )’ Substituting (9)' into (8)' yields b _ cov(Ctyt ) var(Y^) + var(Y^T) (10)' co»(Ctyt ) = e( [Ct - E C J [y, - E yj) = k| [ V t ; Ki Eyt i [ yt + f t - Ey a l . where E(Ypj = 0 E(Ki [ yt - Ey3 ] |tyt - Eyt 3 + yl since p (Yt iYt )= 0, then c°v (ct yt ) = Kx var(Y^) + 0 ( I D 1 Substituting (11)* into (10)' yields var(y^) b - K, p j var(yt ) + var(Yt ) (12)' where ht < 1. 121 APPENDIX IV.B In this appendix, we adopt Darby proof (1974). From that chapter, we have the following equations: ct ■ Ki vt + v l * «t r < ■ K l Y t + Et 2' Yt ■ i + Y I 3' i - eYt + (1-6)(l+g) yJ.j Now differentiate 1' with respect to Yt , we obtain 4’ dYt dY2 dYt dVY dYt dY^ dYj = Kn — - + K, — ^ ^ * dYt dYr 5‘ From equation 4' -rn^- = 6, substitute that into (5 )', we obtain t dC ® dYl — ^ = KnB + K 9 — dY. 1 * dY. t t 61 N ow substitute for Yt from equation 3' into equation 4 '. W e obtain 122: v j - ( i - e ) Y t - ( l - e l d + g j v j . ! 7' dVt From equation 7 ', we obtain = (1-B), then substitute that into equation 6 ', then we obtain dC ® dV7 = + ^ ( i - b) w k2 = KjB + (1-3) 8' or dCj K„ , = K. ($+ 1 ^ -— ( 1-B) dY. 1 h t g i I f we differentiate 2' with respect to Yt , we obtain < „ dc£ dYt or dVt dVt dYt dc! — i = K.(E(s*) , dYt 10' Darby denoted the estimated value of-£ using specification 2 where K 2 = 0 by 6* where E stands for expectation. 12' Then he equated 9' with 10'. Then we obtain k2 Kj E ( b* ) = Kj ( b+ i ^ . d - e ) ) o r k2 E(e*) « a + U" K 2 where is what he called i t , the approximate amount of bias. K1 124 APPENDIX IV.C The LISREL (Analysis of Linear Structural Relationships) model was firs t introduced by Joreskag (1973). The variables in the equa tion system m ay be either directly observed variables or unobserved latent variables. The LISREL model has two parts: the structural equation model and the measurement model. The measurement part shows how the latent variables are measured in terms of the observed vari ables, while the structural equation model shows the causal relation ships am ong the latent variables. The LISREL model can be described mathematically as follows. Consider the following system of linear structural equations: n = 6n + r$ + c where both rf = ( n j,^ . . . and c = (S j,^ * • • • 5^) are respectively random vectors of latent dependent and independent variables, and both e(mxn) and r(mxn) are coefficient matrices andt is a random vector of residual (random disturbance terms). I t is assumed that t is uncorrelated with K and that ( 1 - 8 ) is non-singular where I is the identity matrices. Since both n and ? are not observed but instead vecotrs, y' = ( Y j ^ , . . ., Yp) and x' = (x j, . . . , x g ) are observed such that Y = Ayn + e and X = A X C + 6 125 where e and 6 are vectors of errors of measurement in y and X, respectively. The matrices A (pxm) and A (gxn) are regression ~y ~a matrices of y on n and x on c, respectively. This above model can be written in different submodels discussed above. For a detailed explanation see Joreskag and Sorbom, 1981. 126 APPENDIX V.A T h e first o r d e r c o n d itio n a _ n r— - y - 0 u lt (1 ) ^ - y = 0 °2 t (2 ) R - y x . = 0 K t+ 1 1 (3 ) F r o m e q u a tio n (3 ) s o lv e fo r y; th e n w e o b ta in < 5 ^ T R T tK t+ l T h e n s u b s titu te fo r y in e q u a tio n (1 ) a n d (2 ); t h e n w e o b ta in o _ 6 C lt T tR t+ l 3 _ 6 C 2 t tR t+ l B y s o lv in g fo r a n d C 2t» w e o b ta in , . a T tVi U lt 6 r _ 0 T tR t+ l U 2 t 6 T h e n s u b s titu te th e a b o v e v a lu e s o f a n d C2 1 in th e b u d g e t c o n s tra in t a n d s o lv e fo r 1 2 7 nTtVl + + R = F 6 6r t t+l Then d = ,<SF t + l T^.(a+B+6) and the other two demand functions are r = aF I t (o+g+6) p _ 8F____ 2t (a+3+6) Then the supply function is Q t = \ " Tt (a + e + 6 ) APPENDIX V.B The fir s t order condition is that “ yl = 0 L lt 1 3 - n C 2 t - <*1 - v2 - - * i Tt = 0 R t+ 1 1 * C lt + C 2t + T tR t+l = F (1 ) ( 2) (3) (4) C2t = ** Zt (5) Solve for pj in equation (3); then we obtain 1 T tR t+l ( 6) Substituting for pj in equation (1), w e obtain r a T tR t+l L lt ' 6 Then substitute the value of and C 2 1 in the budget constraint; we obtain 129, ♦ '*+ TtRt+l ■ or aTt Rt+l + < S l* + 6xt Rt+l = Then solve for R^+j; we obtain R t +1 - or ^ t V V ‘ ) t + l T^.(a+S) Then the supply curve is _ sC tJ L + V 1*) Q - n t L w t - R - ~ rt ( a+6) — 0 - n ^ t W 1** Q t ' R t " T t'(a+ '« )--------- APPENDIX V.C u = 0 1 ln(Cl t -Yl t ) + a2 In rt (C2 t -Y2 t) + a3 ln R t + 1 " Pl t Cl t “ P 2 t C 2 t ‘ Tt Rt+ l ^ " F Then the firs t order condition is al 7 7 ----- -— r - A = 0 (cn ' Yi t 7 ( 1 ) a 2 7 7----------r - A = 0 ^ 2 t~Y2 t ( 2 ) “3 t r 2- - At. = 0 V i 1 (3) Cl t + C 2 t + Tt Rt+ l = F * (4) From equation (3 ), solve {xj; then we obtain A - “ 3 Tt Rt+i Substitute for X in equations (1) and (2) and then solve for both Cjt and C 2t < • a i ^Cl t _Ylt^ Tt Rt+l a 2 a3 ^ Q ^C2t~Y2t^ xt Rt+l 131 or _ V t Rt+l + a3rlt u ' “3 _ “2TtRt+l + “3Y2t " °3 Then substitute the values of Cl t and C 2t in the budget constraint; then we obtain the demand function for R^+i . a3 S t + t+l Tt l+a2+a3 132 REFERENCES 133 Adelman, M. (1972). The World Petroleum Market. Baltimore: Johns Hopkins University Press _______ . (1982). OPEC as a c a rte l. In D. G riffin & J. Teece (Eds.), O PEC Behavior and World Oil Price. London: George Allen and Un-win (Publishers) Ltd., pp. 37-63 Al-Qudsi, S. (Spring 1984). A n empirical evaluation of consumption behavior in oil societies: The case of Kuwait. The Pakistan Development Review, 23(1), 81-94 Atfield, C. (1980). Testing the assumptions of the income model. Journal of the American Statistical Association, 75(369), 32- 35 B litzer, C„ and Stoutjesdijk, K . A. (1975). A dynamic model of O PEC trade and production. Journal of Development Economics, 2(4), 319-355 Cooley, T. (1973). Tests of the adaptive regression model. Review of Economic Statistics, 55, 248-256 _, and Prescott, E. C. (1973). An adaptive expectation regression model. International Economic Review, 14, 364-371 _, and Prescott, E. C . (1973). Estimate in the presence of stochastic parameter variations. Econometrica, 44, 167-184. Dasgupta, P., and Heal, G . (1979). Economic Theory and Exhaustible Resources. Cambridge, England: Cambridge University Press Darby, M . (1974). The permanent theory of consumption. Quarterly Journal of Economics, 88 (2), 228-250 . (1975). Postwar U.S. consumption, expenditure, and saving. American Economic Review, 65, 217-222 Duesenberry, J. (1949). Income, Saving, and the Theory of Consumer Behavior. Cambridge, M A: Harvard University Press E llio tt, J. (1980). Wealth and wealth proxies in a permanent income model. Quarterly Journal of Economics, 95(3), 509-530 Friedman, M. (1973). A Theory of the Consumption Function. Princeton, NJ: Princeton University Press^„ ________ . (1974). Newsweek, March 4 . (1977). Nobel lecture: Inflation and unemployment. Journal of Political Economy, 85(3), 451-472 134 Gately, D . (1984). A ten-year perspective: O PEC and the world oil market. Journal of Economic Literature, 22, 1100-1114 Ghabrial, G . (1977). OPEC: Its international economic significance, 1974-79. In R. Stone (ED.), O PEC .and the Middle East. New York: Praeger Publishers, pp. 65-101 G riffin , D., and Teece, J. (1982). O PEC Behavior and World Oil Prices. London: George Allen and Unwin (Publishers) Ltd. Gilbert, R. (1978). Dominant firm pricing in a market for an exhaustible resources. Bell Journal of Economics, 9 Hannoch, G . (1965). The "backward-bending" supply of labor. Journal of Political Economy, 73(6) Heckman, J. (1974). Life cycle consumption and labor supply. A n explanation of the relationship between income and consumption over the lif e cycle. American Economic Review, 64(1), 189-194 Hotelling, H. (1931). The economics of exhaustible resources. Journal of Political Economy, 39, 137-1975 Jacoby, N. (1974). Multinational Oil. New York: Macmillan Co. Johany, A. (1982). The Myth of the O PEC Cartel. New York: John Wiley and Sons, Ltd. Joreskag, K . (1973). A general method for estimating a linear structural equation system. In A. S. Goldberger and A. D Duncan (EDs.), Structural Equation Models in the Social Science. New York: Seminar Press, pp. 85-112 Joreskag, K., and Sorbom, D. (1981). Analysis of Linear Structural Relationships by the Method of Maximum Like!ihood. Chicago: National Educational' Resources, Inc. Kalyman, B. (1975). Economic incentives in oil pricing policy. Journal of Development Economics, ,2(4) Keynes, J. (1936). The General Theory of Employment, Interest, and Money. New York: Harcout Brace Jovanovich Kuznets, S. (1942). Uses of national income in peace and war. New York: National Bureau of Economic Research 135 Lucas, R. (1976). Econometric policy evaluation: A critique. In K. Brunner and A. Meltzer (Eds.), The Phillips Curve and the Labor Market (ed. 1, vol. 1 of Carnegie-Rochester Conferences in Public Policy, A Supplementary Series to the Journal of Monetary Economics. Amsterdam: North Holland Publishers Macurdy, T. (1981). A n empirical model oflabor supply in a l i f e cycle setting. Journal of Political Economy, 89(6), 1059-1086 MacAvoy, P. (1982). Crude Oil Prices. Cambridge, MA: Ballinge Mayer , T. (1972). Permanent Income, Wealth and Consumption: A Critique of the Permanent Income Theory,- tfie L ife Cycle Hypothesis, and "Related theories. Berkeley: University of California Press Nordhaus, W . (1973). The allocation of energy resources. Brooking Papers on Economic A ctivity, 4(3), 529-570 O PEC Chronological Trends. (1980). p. 9 Petroleum Intelligence Weekly, August 20, 1973 Pindyck, R. (1978). Gains to producers from the cartelization of exhaustible resources. Review of Economic Statistics, 60(2), 238-251 ~ _________ . (1982). O PEC oil pricing and the implication for consumers and producers. In D. G riffin and J. Teece (Eds.), O PEC Behavior and World Oil Prices. London: George Allen and Unwin- XPubTTshers) Ltd. Salant, S. W . (1976). Exhaustible resources and the industrial structure: A Nash-Cournot approach to the world oil market. Journal of Political Economy, 84(5), 1079-1093 . (1982). Imperfect Competition in the World Oil Market. Lexintgon, M A: D.C. Health Schneider, A. (1983). The Oil Price Revolution. Baltimore: Johns Hopkins University Press Teece, D . (1982). O PEC behavior: An alternative view. In D. G riffin and J. Teece (Eds.), O PEC Behavior and World Oil Prices. London: Georqe Allen and Unwin (Publishers) Ltd., pp. 64-93 Wassink, D. (1978). Economic development with unlimited finance OPEC. Rivista In t. S ci., Econ. Com., 25, 1086-1095 136

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Key factors in appraising development project in Egypt

PDF

A macroeconomic model for Mexico with emphasis on income distribution

PDF

The political economy of financial reform and financial repression in developing countries

PDF

An energy-economy model for Indonesia

PDF

A nonlinear simulation model of the American economy

PDF

North-South trade and the growth of the South

PDF

Does air pollution affect residential property values?

PDF

The Pricing Of Initial Public Offering And Privatizations: An Influence-Seeking Theory And An Empirical Investigation With New Issues In The Nigerian Equity Market

PDF

Exchange policy: A research on Peru's rate of exchange and external balance

PDF

Social and political factors in the Lebanese economy

PDF

Exchange rate volatility and regular dynamics of the exchange rate in general equilibrium models of exchange rate determination

PDF

Three essays on the determinants of firm structure and contractual choice

PDF

Interdisciplinary research and the university reward system

PDF

A theoretical model of the optimal distribution of labor derived income

PDF

Dutch disease economics, rent-seeking theory and the Nigerian food sector

PDF

The management and design of economic development projects: A case study of World Bank electricity projects in Egypt

PDF

China's economic reforms and establishment size, revealed comparative advantage, and policy effects: A new institutional economics approach

PDF

Philippine financial reforms: Aspects and implications

PDF

Interaction between the public and private sectors in regional I-O models

PDF

The Middle Eastern oil exporting countries: Absorptive capacity, market sharing, and investment strategies

Asset Metadata

Creator Adeinat, Mohammed K. (author)

Core Title The price of oil and OPEC behavior: A utility maximization model

Degree Doctor of Philosophy

Degree Program Economics

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag economics, general,Energy,OAI-PMH Harvest,political science, international relations

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Nugent, Jeffrey B. (committee chair), McEachern, Alexander W. (committee member), Niedercorn, John H. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c20-268720

Unique identifier UC11259447

Identifier DP23332.pdf (filename),usctheses-c20-268720 (legacy record id)

Legacy Identifier DP23332.pdf

Dmrecord 268720

Document Type Dissertation

Rights Adeinat, Mohammed K.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA