DISCUSSION PAPE DRD135 ARE OIl, IHNDFALLS A BLESSING OR ,\ CLkSE'1 POUCY EXERCISES HITH AN INDONESIA-Lli~E HGDH by ALL~lB ·. G2.1b Development Research Department Economics and Research Staff World Bank The World Bank does not accept responsibility ror the vievJs expressed herein which are those of the author(s) and shouJd not be attributed to the World Bank or to its affiliated organizations. The findings, interpretations, and conclu.sions are the results of research s~pported by the Bank; they do not necessarily represent official policy of the Bank. The designations employed, the presentation or material, and any maps used in this document are solely for the convenience of the reader and do not imply the expression of any opinion whatsoever on the part of the World Bank or its affilIates concerning the status of any country. territory. city, area, or of its authorities, or concerning tbe delimitations of its boundaries, or nat:lonal [!.f iliation. ARE OIL WINDFALLS A BLESSING OR A CURSE? POLICY EXERCISES WITH AN INDONESIA-LIKE MODEL by Alan.H. Gctb Development Research Department The World Bank November i985 * The World Bank does not accept responsibility for t~e views expressed ~erein which are those of the author(s' and should not be attributed to the World Bank or to its affiliated organizations. The findings, interpretations, and cunclusions are tfte results of research supported by the Bank; they do not necessarily repr~ent official policy of the Bank. The designations employed, the presentatioq of materisl, and any maps used in this document are solely for the convenience of the reader and do not imply the expression of any opinion whatsoever on the part of the world Bank or its affiliates concerning the legal s~atus of any country, territory, city, area, or of its authorities, or concerning the delimitation of it~ boundaries, or national affiliation. ARE OI~ WINDFALLS A BLESSING OR A CURSE? POLICY EXERCISES WITH A. MODEL OF fNDONESIA Alan H. f.;elb Development Research Department ABSTRACT The value of Large, unpredictable terms-of-t~ade gains to mineral exporting countries ha~ been much debated. This paper Uges a computable general equilibrium model of an Indone~ia-like economy to aSgess the "value" of oil windfalls over a 20-year hori~on. The model is simulated and optimised subject to a variet.y of assumptions on macroeco,,"'~:c' d.e'aring, institutional co~straints on economic policy and the ability to predict the world oil market. The results indicate the critical importance of policy in determining the realised value of windfall gains. A "base run" is first constructed, to represent absorption policies characteristic of a representative oil exporter. Used in this manner, the windfall is insufficient to satisfy hopes for accelerated growth such as those common among exporters in the mid-1970s. Permitting optimal lending and borrowing abroad increases its value considerably, particularly if the nonoil economy is not able to adjust to rapidly falling effective demand without experiencing unemploye~ factors of production. The possibility of capital fLight, spurred on by restrictive domestic financial policies and anticipation of the need to realign the real exchange rate when oil revenues fall shortens the adjustment period and can involve heavy cost if frscal policy is not stabili~ing. Overoptimistic predictions of future oiL revenues are shown to have seriously adverse consequences, particuLarly if the nonoil economy adjusts to faLling demand through underempLoyment and capital flight is provoked. Overall, the results indicate a large dow~side riWk attached to reversible terms of trade gair.s. It is not difficult to construct reaListic policy scenarios which result in negative overalL valuation of oil windfalls. These resuLts accord with, and to some degree explain, the conclusions of comparative study which suggest that gains to producer countries frem higher oil prices hav~ been quite limited. AGD/sg16-al/11-1-85 ARE OIL WINDFALLS A BLESSING OR A CURSE? POLICY EXERCISES WITH A MODEL OF INDONESIA Table of Contents Page No. I. Introduction and Overview 1 ···························· (> .... ""·· II. Optimisation and Valuation ................................... 6 :.II· Flexprice Policy Simulations ····························.··· 10 34. A Base Run ·············· ~ ··· O ·························· 10 3b. The "Value" of the Windfall under Alternative Policies ... .................... 12 ~-.' ~. IV. Windfalls when Adjustment is Not SlnOath ····················· 16 4a. Sticky Wages a.nd Prices 16 'I ·························· " ···· 4b. Nigenesia and Trinuela: Valuing Some Institutional Differences ··········.················· 17 V. Optimal Public ~ending and Borrowing Abroad ··.·············· 19 Sa. Flexprice Adjustment . 19 t············· 'I ··· 'I ···· 'I ···· 'I ··· 'I. Sb. Macroeconomic Rigidities ············.·················· 21 VI. Speculative Private Capital Flows ··············.··.······· 22 ~a. Modelling Capital flight and its Effects with Flexprice Adjustment ···········.·····.··.·.····· 22 6b. Capital Flight and Sticky Adjustment ·····.····.····. 24 6c. Optimal Public Lending with Speculative Private Capital Flo~9 ·············.···.··.·.··.·.·..· 24 'fII. Imperfect Foresight: the Consequences of Euphoria · ·· 'I.' · 26 . " ..·. VIII. Conclu9io~ ······························ , ················· 29 .,1\. Use of Windfall Gains .................................. 32 Capi~al night ................................................ ,. ..... 34 Predict ion Errors ............................ "..................... . 35 Su.rnrning Up ................................................. . 36 References 54 ANNEX 1 56 List of Tables Table No. !::age N~, \ 1 Bas~ Run for a Twenty Year Hori~on 38 2 Alternative Uses of the Oil Windfall ··· " Ii) ......... !!Ii...... .. 39 3 Windfalls when Adjustment is Not Smooth ················ 40 4 Optimal Public Savings Abroad .. ".............. Ii ...... I............ . 41 5 Effects of Private Capital Flows ·· e , ············ 42 6 Optimal Government Borrowing with Speculative Capital Flows ········································ 44 7 The Welfare Gains of Optimal Government Borrowing with and Without Speculati,,~ C.:.:p,~,..:!f.l'rlows ···.···.··· 4') 8 Socially Optimal Government BOt"!"0''''';,(':S ~:I;d~'i." Incorrect Expectations ·······················.······ 46 List of Figures Figure No. 1 Real Wage Trends over the Oil cycle .................... 47 2 Normal~sed Duals of Domestic Oil Price .................. 48 3 Optimal Public Borrowing ····· /I ...................................... . 49 4 Soci&lly Optimal PUblic'Borrowing ............................... f,........ ~ 50 5 Privately Optimal Public Borrowing 51 .................................. OIl...... 5 cont. and Resu~tlng Real Exchange Rates .................................. ........ '" 52 .. 6 Euphoric:Expe~tatio~s .................................................................... 53 At Base and~ptimal Private Savings ............................................. 61 AGelb:ao 11"1-85 I i ARg OIL WINDFALLS A BLESSING OR A CURSE? pbLICY EXERCISES WITH AN INDONESIA-LIKE MODEL 1/ I. Introduction and Overview. This paper addre3ses the problems of economic management faced by oil producing countries through terms of trade fluctuations and the "value" of f ~~ndfall gain~ when their use is constrained by polLtical and admini.trative factors. Oil exporters, :Jm:·.lg them IndonesIa, have experienced extraordinary fluctuations in foreign exchange availabiHty since 1973. The fh,,;: oil price hii<.e conferred a windfall equivalent, on ll't,'utt'lge, to jj.4uE< ·.i:T."."'\\ini!1~GOP on a sample of six producer9~ Gelb (1984).~/ By 1978 thi~ ha~ fallen to 15%. The second oil price ris~ over 1979/80 raised the windfall back to 27%; it was then halved by the oil glut of 1982-84. Are such windfalls 8 blessing, with potential;o create a diversified, dynamic, modern economy (the universal objectiv~ of developing oil exporters)? Or are the difficulties of economic management through uncertain windfall g8i~s so severe as to erode their large ex-ante value, perhaps to the point of turning them into a net liability? The contribution of high-rent mineral sectors to. development has been much debated: for eXilrr.ple. see Hirschmann (958), Amuzagllr (1982), Nankani 1/ The exceptionally hel~ful cOfltribution of Perry Beider and Arne Drud to this work are gratefully acknowledge.. The model was solved u~ing CONO?T which is supported by the Analytic S'~pport Unit, Development Research Department: see Drud (1983). 2/ Thp. producers are: Alg~~ia, Ecuador, Indonesia, Nigeria, Trinidad and Totago Bnd Venezuela. Expressed as a percentage of GDP the windfall would have been far larger for the capital 9urplus producers. (1979), Harberger (1981), Levy (1978) and Forsyth and Kay (1980). The rnnBe of view9 is wide:, I " ·· I hald a vision of La Gran Venezuela ··· " Carlos Andres Perez, ~resident of Venezuela, 1974-78. " ··We are dying of indigestion ··· r call petroleum the Devil's excrement ···waste, corruption, consumption and debt, debt we shall have for years. We are putting our grandchi ldren in debt. II Juan Pablo Perez Al fonso, . Venezuelan oil minister and founder of OPEC, 1978.l/ The performance of oil exporters after 1972 indicate the critical importance of policy in determining the ultimate value of the windfall. Although different countri~~ adopted considerably different 9trategies, some responses were common. Producer governments received about four~;:i.fth5 (.iftne windf~11, came under intense pressure to spend rapidly, dild ~c~ns(ormed m0'~ of their oil income into domestic publ~c investments. About 60I of these were infrastructural and much also went to heavy industry. Hastily planned and implemented, the average quality of public capital formation appears to have been low, with serious cost overruns and delays: Murphy (1983). During the boom, real exchange rates eppreciated and in most countries oonoil output skewed towards the nontraded sectors. Nonoil exports stagnated, imports grew rapidly and dependence on oil increased. Meanwhile, during 1972-81 the nonoil economi~s of the oil exporters grew only 0.9! faster than had middle income developing countries over the 1960's, and they failed to match their own (admittedly exceptional) growth performan~e over 1967-72 . .! The heightened vulnerability of She exporters was revealed by the downturn in world oil markets. Absorptio~reducing policie3 implemented to 1/ Karl (1982). pp. l6-!9. adjust to falling oil income moved exporters i~to severe receSSlon, a f;onseqllence of Bisynunetric macroeconomic adjustment. , Reul exchange races, I prices and wage~ did not respond flexibly to the drop in terms of trade 89 the, had on the upswing. Labor markets slackened and urban unemployment began to preoccupy governments. The rhythm of public investment was again disrupted, often at the high cost of unfinished 91ant. Capital flows amplified the destabilising swings of oil income. F10jections of future oil prices have tended to follow current trends. Expectations have usually been proven seriously wrong, and have thus destabilised the capital account. Rising prices increased public and private access to foreign loans, but in periods of falling prices anticipation of devaluation prompted capital outflows, particularly from countries whic~ maintained cpen capital markets and fixed interest rate ceilings. In 1982 the equivalent of 7% of Venezuela's GDP is estimated to have been sent abroad. Several exporters including Mexico, Venezuela and Nigeria experienced rlebt crlses as the world oil market turned down after 1980. Despite adjustment costs and destabilising feedbacks, some might argue that oil windfalls must be a bonus because they relax resource constraints and so expand the range of choice. But this view disregards the close connection between resource availability and other constraints on decision-making. In principle, governments have an almost infinite range of choice in disposing of oiL rents; in practice this is not the case. The perceived range of choice has been far narrower, both over fntertemporal a allocation and across uses at ~ point in time. Further, a ~ountry!s perceived choice set is heavily conditioned by its re~~nt history, and shifts with the availability of revenues. Not all options feasible with low revenues remSln pol iti.cally acce?table as revenues increase.'!:'! I The prerent analysis therefore has two objectives: (a) to identify constraints most likely to reduce the realised value of oil windfall gains, and assess how they might be relaxed; (b) given such systemic constraints and macroeconomic adjustment asymmetries, to consider "seconci-best" use of windfall gains and their value when their use is "realistically" constrained. The quantitative framework used to analyse these issues is a SlX sector computable general equilibrium (CGE) model of an Indonesia-like economy, described in Gelb (1985). The model economy :3 dUiit~::;tli::~with value-added sharing HI the "traditional" seCLu,n;. (Ivet:' a 20 v~'''''''~),!d'0d government disposes of oil windfalls "optimally" or according to rules which represent political and administrative constraints. simulation is considered as totally constrained optimisation.~1 The private sector. consisting of firms and one household, reacts to government policies and relative prices. The model has several distinctive features. It permits endogenous regime-shifting, to capture macroeconomic adjustment asymmetries i,n the reaL exchange rate and real wage. Certain wages and prices rise more easiLy than they fall, so that the resyonse of the nonoil economy to increasing and l' For this resson certain governments have sought to mask the true extent of their windfalls while others have obscured the extent of1their saving abroad by redefinicg expenditu~es to in:lude anticipated future spendfng. Ge1b (1986) reviews the main priorities of si; capital impor'ing oil exporters and the impact of windfall gain9~n public choice. 2 An important methodological difference between the present work and some which have preceded it such as Dervis, Martin and van Wynbergen (1983) is the treatment of optimised runs as simulation with progressive slacking- off constraints. This assists interpretation of results and per~its the use of models with greater structural and institutional detail at the cost of Ii less theoretically "clean" treatment. falLlng oil prlces may be different. It embodies a richer set of intert~mporal li~kages thon most CGEs. In addition to the conventional updating o.f capi~al stocks through investment and of foreign debt through borrowing and lending abroad, production functions are shifted by "infrastructural" public investment which also locks government in to fut\~re recurrent spending obligations. The efficiency of capital formation falls as investment decelerates or accelerates from its trend growth path. Section II de9crib~s how model runs are evaluated, using a criterion function of discounted private consumption and terminal nonail output. This 1S treaeed in more detail in Annex I. Dynamic simulations are reported in Sections III - VII. InitiaL~y we assume: (a) perfect foresight; (b) no endogenous private capital flows, (c) constraints on public saving of oil revenues abroad; (d) smooth, symmetric (neoclassical) adjustment of the nonoil economy to changes in effective demand. A "base sif"lUlation" represents the nuLL trajectory under poLi.cy considered to be empirically representative. Over the oiL boom government , does not lend abroad or borrow heavily against future oil income and allocat~9 revenues to public investment. The "values" of the windfalL 1n il r3nge of uses but!subject to the above assumptions are then compared. The experi~ent5 , also indrcate policy determinants of the extent to which oil booms are Likely to accen~ate economic duaLism. Section IV relaxe:3 assumption (d), allol..ling for both "Keynesian" ar.d "Cla!;gicill" underempLoyment of factors through (i) dO\<.'T\ward Htickines'l 0 modern-sector wages and (ii) 910w adjustment of the relative price structure (the real exchange rate), , , In sec¥on V (c) is relaxed; government can now lend or borrow abroad. !t optimises its absorption time-profile according to two criteria, one "sociaL" and the other "private," the latter with a higher rate ot ;:ime preference. Section VI also relaxe~ (b). It is assumed that private capital ~lows respond to changes in the real exchange rate. Rapid depreciation of the real exchange rate (which will require nominal devaluati~n if pricGs , e sticky downwards) is pcstulated to send private saver:1 scurrying to place funds abroad. This accelerates the contract:i:'il 3£ dc:::\'::"~.tc 2!H~ctlve demand Section VII also relaxes assumption (a). Enphor \.,:·,H"0jtcctons of future oil prices (6% real growth, the projection which underlay Mexico's decision to borrow heavily against future 011 income in 1979, 1980 and 1981) are revised following the peak of the oil boom. Policies are optimised subject to the initial projection and must then be re-optimised when they turn out to be wrong. This permits the impact of large prediction errors to be distinguished from those of political and other constraints on lending abroad and the use of windfall gains. I rinslly, Section VIII summarlzes conclu\ions. While some (notably those concerning income distribution) relate rather closely co such Indonesia- specific features of the model a~ the specification of labor ma~kets most are more widely applicable. II. Optimisation and Valuation It is hard to a~gue that economles evolve ac~ording to the criteri~n of optimising a Long-run objective function. ~any futures markets are !ackinc; uncertainty increases as the time horizon extends. There is aLso the question of whos~ objective function society i~ presumed to maX1mlze. To link · I simuLation and Oftimisation it 1S nece~sary to be clear on the relationshtp between a particular objective function and a given simulation. Within a given set of constraints, simulation run x is termed compatible with vaLuation function u(.) if X lSthe optimJ11 solution with <.:riterion u or "closel' to it. Otherwise, we can consider u(.) to be an imposed valuation criterion. Policy assessment 19 quite diffeLent for compatible and imposed criterion functions. For example, suppose that ul is compatible with a base run of the model but that u2 has a far lewer rate of time preference. Policies which then allocate an oil windfall to investment will be judged more favorably by u, b~~use the saving9 rate 1n th~ a~aence of oil income is "suboptimal." But care must th,"n be taken in using u2 as an indication of the priorities of those responsi~le for the savings decision; it represents an "imposed judgment. 1t Annex II pre3ents details of optimisation procedures and the i:.verse optimum probLem of determining a function ul within a given cLass of valuation functions which tS ~ompGtible with a reasonabLe base run, Only matn points are noted here. The Objective 19 taken to be to mSXlm1se the discounted value of the log of priva~e consumption per head valued at base year prices.!/ There 1S no allowance for the pattern of ;ncome distribution. The first part of the ,. valuation function is therefore: 1/ An alternative would have been tv value the integral o~ the under~ying utili~y function. Althougn this is mGre consistent frOM the per9p~clive of optimisation, base year price valuation was ~h09~n to permit a clo~er link between valuation; simulations and the comparacive analysis ill Gelb (1984) · AGD/ag16-A!11-1-8S - 8 - [91n(c/n)(l-r)t t:l where r is a rate of pure time preference and n IS exogenous population. Eac ., 11 per10d 1S two years.- Because of different rates of technical change across sectors and constrained oil sector 0utput the model does not reach a true !.teady state &\~n when world oil price~ are constant. The terminal condition problem is solved in a crude but atiparently effpctive way. From a rall~"" "f: ite little net debt. ~or an analysis of e;;"i: flight which emphasizes the importance of overvalued exchange rates an. restrictions on domestic interest rates se'e Cuddington (lS8S). The present analysis assumes away the possibility of lecal investment in fn~Aicr~-rlAnAm;"arort onn~~~ AGO/ag16-a/ll-l-eS - 23 - private investment. Large depreciations cause a maXlmum percentage outflow 22% private s~vings!1 Termin4 ~tocks of fnreign private assets are mostly undeclE1re.d and may never be repatriated. They are assumed to have zero value. Table 5 summarizes t~eeffects of private capital flows 1n a number of simulation runs. YA ("yil indicates a run with endogenous capital flows) valued less (socially and privately) than A. YR, with constant oil priceJ, ~:oduces a higher value according to both criteria, than the corresponding r B. Private capital flows are most valuabl~ ~~g stable economic environment and possibly adverse W~lenexogenous shocks are experienced and policy is nor stabilizing. Capital flows reduce the benefit of the oil windfall by 88% as defined by the private welfare function, and results in the windfall having net social cast. Oil windfalls can be a liability if inappropriate (yet empirically representative) policy causes volatile capital flows.~1 11 This bound is based on 'fenezuelan data. This can be interpreted as supposing that private capital which has not been induced back into the economy by year 20 never will be, and neglecting the benefit accruing I a (usually) limited group of citizens with assets abroad. All private capital flow figures, and resulting as~et/debt levels, reported here 6 in addition to an exogeneous inflow identified in the base-year data a maintained at at a constant level in all periods of all runs; this exogeneous fore1gn capital SOurce accounts for 12% of base-year privat; investment, and r~ughly 2% in the last period. The ~ffect of changes i terminal foreign assets over and above Lhose due to the exogeneous flo·; is ~3sentially zero in the base case: with a maximum inflow of 2.5% there is a negligibl; terminal net asset position. It 21 These conclusions need to be qualified if higher terminal private asset stocks confer a later benefit which is not reflected in the valuation functions. However, even without this post-horizon correc~ion, higher private terminal debt does not imply higher valuation. For example, ru YA results in a small net terminal debt but is valued less than A becau the final net foreign asset s~ock is an inadequate summary statistic fo the time profile oi flows. The value of the time profile depends on th rate of ti~e preference and on the rate of retur~ to domestic investmen and its efficiency. Capital flows with a Inrge early outflo~ balanced only later oy gradually increasing inflows Like VA, will be valued less AGD/ag16-a/11-1-8~ - 24 - It cout(j be argued that profit maximising private investors \Jould ne disrupt domestiq inve9tments too severely by switching savings abroad. In tw additional experiments t YA* and A*t devi,'ltions from "normal" privat~ investment growth are allowed without penalty. Capital flows now augment social value slightly, even when world oil prices and the economy are volatile. However, even with an unvalued increase in terminal debt, they .educe the value of the less "patient" private objective function. 6b. Capital Flight and sticky Adjljstu:'~flt Capital flight accentuates the contraction 1n demand, compresses the period of real exchange rate adjustment and increases the costs of wage and price rigidity. Unemployment in run YF is more severe and longer-lived than 1n run F, the corresponding run without capital flight. Terminal-year output is 2.l% lower than in ru~ YA; without capital flight the terminal output loss from rigid wagas was only 0.5% in run F. The value of the windfall 1S now su~stantially negative by either welfare function. The combination of a temporary oil boem, policies which induce capital flight and an inflexible labor market can be disastrous. The economy .. now consists of two groups. "Speculators" are better able to protect their assets because they have flexibility 1n disposing of their savlngs. But the) amplify the fluctuations exp!rienced by "workers," at great cost to the economy as a whole. , 6c. Optimal Pubilc Lending with Speculative Private Capital Flows Figure 4 depict:.s socially optimized government borrowing trajectorie and the resulti~g real exchange rates with and without priVAte capital flows. Y denotes capital flows, X denotes optimal borrowing; and runs with and without the boom in world oil prices are respectively labelled A and 8. AGD/ag16-a/11-1-85 - 25 - Socially optimal borrowing and lending profiles do riot change great with cspital fl~s. They are still strongly countercycli~al. Privately optimal borrowi~ strategies are affected far m0r~- as shown in Figure 5. Tl potential losses to the domestic economy from capital flight induce far stronger counter-cyclical policy even with heavy discounting. Whether one large depreciation is "better" than several small depreciations depends on the determinants of capital flight. When large _d~preciatiQns cannot be avoided and there 1S a natural limit to the share of private savings which can be remitted abroad, it is better to effect large parity changes quiCkly. The four optimal runs with private capital flows are summa1-tsed in · Table 6. This allows valuation differences between the base run and constan oil price simulations to be compared with those in the previous optimized ru, without private capital flows, in Table 4. With optimal public borrowing abroad, the 10s9 due to volatile capital is 4% with ul and 53% with u2; good policy can limit the adverse effects of volatile capital f10ws. However, a temporary oil boom which requires eventual real exchange rate depreciation 13 necessarily less valuable with private investor behaviour as modeled here · .. Is "gOOd" policy more or less important when there is a propensity for capital flight? From Table 7, at either discount rate, the "gap" in val" between the cases with and without capital flows increases (becomes mO~0 positive) as the problem of economic management trecomes more complex--moving · from the constant oil price situation to the oil ~oom and then to the oil boc with rigid wages. In other words, endogenous capital flows increase the premi\~ on optimal policy as the economic environment becomes more complicatE due to external shocks and internal rigidities.!/ ~I - Lb - VII. rmperfectForesight: the Consequences of Euphoria The ev~lution of world oil markets has u~ually confounded projections. Both oil price sho~ks were unexpected. In 1974 and 1978 there was widespread skepticism over OPEC's ability to maintain high oil prices. By 1981 predictions of global e~~rgy shortage led to projected real oil prices growing at 3% until at least 1990. Some exporters were still more ~ptimistic. Mexico, for example, assumed 6% real growth in oil prices 1n formulating its national plans. This section considers the consequences of such prediction errors as for economic man8gC:Il':'!~r..· The socially optimal publi..:1Qnr.ht1q', solution 1S fir'!t r,n;,-:J fur a "euphoric" oil price projection--after the peak of the boom, world oil prices Footnotes from previous page. 11 The sign of a (welfare gain from optimality) 1S ambiguous, but that of a capital flows a2(welfare gain from optimality) positive. a capital f18ws a complexity 1S 2/ Since government borrowing trajectories in YXA2 and YXAl lead to steadily-rising real wages in the formal sector, they are also optimal with downward wage rigidity unlike the comparabl~ case XAl in Table 5. Thus, for each of the two valuation functions, t~ewelfare gains in moving from the exogeneous base public borrowing~path to the optimal trajectory can be calculated for three situation!; (8) with the boom in It oil prices, (b) with constant oil prices, and (cft with the oil boom and rigid wages. In five of the six comparisons in ~able 7, optimal policy is indeed "worth more" with a free capital market, but this result depends on the particular sub-optimal borrowing trajectory chosen for the exogenous base. Essentially this result means that the base path is closer to being "dght" for an environment without capital flo<.Js than [or one with flows which requires more caution. AGD!ag16-aill-1-85 - 27 - are projected to~continue to rlse at 6% per year.!! Public lending and borrowing abrcadi is then optimised with the base oil price trajectory, but · with lending in the first three periods fixed at its values in the euphoric run. Thus, for three periods policy IS optimal but conditional on very optimistic projections. As the boom slackens it must be revised in line with a pessimistic correction. The seriousness of forecasting error depends on the time horizon and borrowing possibilities. Real-world limits on debt serVIce and debt/GDP constrain the ability to smooth out its effects. In the model it is sufficient to impose the same terminl.il darer. ccon~lI·r''11 r.t'".:cause the horizon 1s not sufficient to permit a very long-term strategy.~! ~igure 6 displays optimal public borrowing and real exchange rates for five such imperfect foresight runs. The euphoric projection causes a massive borrowing peak in periods 2 and 3. This augments the oil windfall by 1/ In addition to optimal public borrowing, the domestic oil price is raised exogenously by 1.5 per period, from the base level of l in period 3 to 11.5 by period 10, and a second policy instrument--the level of transfers to households--was also included in the optimization. The optimal transfers is zero in periods 2-5 (transfers in th~ first period are fixed in accordance with the base-year Sfu~). ~or greater comparability with earlier runs transfers are again specified exogenously in the experiments below using the "mistakenft starting point. Instead of the base , trajectory, with transfers ccnstant at 0.139, these later runs set transfers to 0.139 in the first and last periods, and to zero in periods 2-9. This trajectory agrees with the first periods of the high-pric'~ · optimization, and allo~s the governme~t a little more slack in makin~ up the revenue 9hortfalls in later years. Only in period 4 and 8 of rung YZA2 and YZ?2 beLow did the duals on household and governmpnt income indicate that positive transfer ~ould be desirable; thus, the losses due to mistaken expectations would be slightly larger than those c~lculated below if the base path for transfers was retained. 2/ In the runs reported, t~e debt service ratio never rises above 65%. A separate debt serVice r~tio limit is not imposed; its effect is quit~ predictable. - .1:.0 - 66%.. V causing t~ real exchange rate to appreciate by 38%. Runs YZF2 and ZF2 include downwarcl modern wage rigidity, with and without endogenous private · capital flows. The heaviest borrowing is in period 4, as the government tries desperately to reduce the macroeconomic costs of adjustment. Mistaken expectations about world vi~ prices have, however, absorbed borrowing capacity and have also elevated the real wage. Between periods 4 and 8, the formal !>Gctor unemployment rate is 15.1%, 17.1%, 16.0%, 11.4%, and 3.n: in run ZF2, and 17.2%, 23.6%, 22.0%, 16.6%, and 7.8% 1n case YZF2. '!:.,I This indicates the degree cf stress on the nonoil pconomy. The dramatic swing from borrowing to lendin,), 1,:1',1' :'.i-dC\(.' in · v,"'" ,." (tiU' ....iJ;;J_ (socially optimal public bo~rowing with mistaken expectations and private capital flight) results from the fact that a severe depreciation of the real excha~ge rate is inevitablp in these runs and that once large it may be still larger without penalty. Government can partially stabilise real private investment despite private capital flight by severely depreciating then appreciating the real exchange rate; but such violent erratic policies are not optimal with the adjustment costs of runs ZF2 and YZr2. The projections error has disastrous consequences as shown in Table ~ 8. It eliminates 63% of the social value of the windfall in run ZA2; for the 11 Coincidentally, M.xico's increased borrowing over 1980-B1 was a130 ~% of its actual windfatl; Gelb (1984), 21 There is also 4.6% unemployment in period 9 in run YZF2, even though the wage constraint has ceased to bind by that time. This is apparently in order to keep wages uP. and through them the price of investment goods, so as to prevent the level of real investment from rising further. As it is, the real growth rate over the previous period is 24%. This is the same rationale (though using a different mechanism) disc'lssed helow Ear run ¥ZA2. - 4':1 - other three ]:J c~sesthe loss ranges hom 122% to 325%< '!.'he economy may be far worse off refeiving a windfall and badly overestimat~an correctly anticipating no windfall at all. All four runs are also valued less than corresponding simulation runs; A, F, YA and YF are respectively preferred to ZA2, ZF2, YZA2, and YZF2 respectively. The base borrowing profile produces (socially) better results thm the "optimal" trajectory in all cases with badly incorrect expectations. ZF2, YZA2 and YZF2 are even worse than the corresponding runs with the base borrowing profile and witho~! the windfall. Is such an error preferred 'D the full soci,·d.0t"tlWUJD by ~l ~hi~\1 heavily discounts the after-effects? For YZ2, the er~or does cause a 27% rlse in the windfall valued at the higher discount r~te. However, for runs ZF2, YZA2 and YZF2 respectively, the mistake induces losses of 47%, 99%, and 198% of the private benefits. Even though it induces ma9sive borrowing the mistaken trajectory is not preferred by ul when the economy cannot adjust smoothly downward and private valuation of all four cases is also lower than that of the corresponding simulations. VIII. Conclusion. lnis study addresses a much debated question--the value to producers .. of oil windfalls--using a=computable general equilibrium model of an exporting country. It emphasises ~iPium-run adjustment and management through terms of trade gains and reversals, not longer-term depletion. Two valuation functions have been used. One (ter'med "private(') has 11high discount rate, the other, 11 No comparable cptimization runs existed for case YZA2*. However, there lS a corresponding slmulatlon run YA* (see Table 5). "sociaL" d Low one; the maximands are private consumption streams and a terminal conditi9n on nonoil output. Some resuLts, notably those concer.ning distribution (w~ch i9 not given weight in valuation) reflect dualistic · features of the model specific to Indonesia and 9imilar economies, but most are more generally valid. Ev~n if adjustment costs are Large, windfalls cannot but be of value if used "optimally". However, their absorption, both over time and across ~~es, has been constrained by It variety of political and administrative factors which have themselves changed more gradually than the financial re90urce9 availabLe to oil producing economies. Furthgr~ 18r~e thaQ!es.in I¥orld oil markets have been poorly predicted. 'TIlis paoer therefo-.::e assesses the value of windfal19 under a variety of a9sumptions on (i) the range of choice available to producer governments, (ii) key m~croeconomic closures, and (iii) the predictability of the oil market. Optimal Absorption Rates. The representative oil exporter did not save windfall gains abroad; neither did it borrow heavily against future oil lncome. Revenues I¥ere mainly transformed into domestic public investments. The base simulation emb0dies these pol.it:y choices 1n the response to an oil windfall which is perfectLJ :orseen from lhe start, lasts for six years and .. peaks at the equivalent of 17% of nonoil GOP. It is also assumed that the nonoil economy adjusts smoothly through flexible wages and prices, and that private capital flol¥s do not react in a volatile manner to changes in world oil markets or to domestic policies. Downward rigidity of real modern-sector wages and stickiness 1n the real exchange rate when demand is falling are then introduced. These introduce an !lSyn'.metry into the process of adjustment, and resuLt ill & tendency for large booms to t~ followed by receSZlons. The post-boom receSS10n 1S shown to eliminate up to 40% of the windfall's value with tne above policy choices. If the causal relationship between boom and recessiun were more widelTt appreciated, restraint 1n spending windfall gains might be · more politically viable.l/ Assume now that the absorption of windfalls (which are still ~erfectly forseen) can be optimised over time. At t~e 10w disco~nt rate of 3% it is optimal for government to save the equivalent of 71% of the windfall 'l~road during periods 2, 3 and 4 (each period represents two years) and then to run down its foreign assets as oil exports fall. This strategy raises the windfall's value by 61% relativ~ to the base run because the profile of consumption is steadier, the efficiency of public capital formation is maintained and the real exchange rate appreciates far Less. This legt effect maintains the purchasing-power of oil over domestic investment goods and so reduces "crOWding-out" of real capital form.:tion. Because the optimal trajectory is steady it does not run up against macroeconomic rigidities unlike the base run. Taking these into account, fiscal restraint roughly doubles the windfall's value. However, even under the most favorable circumstances the contribution to growth of the windfall 1S modest because of the importance of other factors of production, technical · change and the r~allocation of labor from subsistence to modern sectors. The expectations raised at the timE of the first and second oil price increases appear to have been overoptimistic.~/ 1/ It is interesting to note that the high downside cost of poorly sterilised export revenue fluctuations were central to Keynes' arguments for commodity price stabilisation, which were ~~de on macroeconomic grounds (Kanbur (1~84». 2/ For a single-sector analysis which reaches the same conclusion gee Gelb (1984). Discounting at the high rate of 14.5% creates pressure for heavy additional borr~ing during the oil boolll. This is because muLtiplier and real exchange rate effects cause about one third of investment expenditures to accrue to private consumers; whether the investment has a favorable longer- run impact on supply is less important because of discounting. Such considerations have undoubtedly been important 1n expLaining the speed with which governments have absorbed windfalls. However, recession after the boom (j~eto macroeconomic rigidities is so costly that, even with heavy , discounting, optimised fiscal policy is rather like that in the base scenario, with little tending abroad or borrowing ag~inst future·inco~e. This confirms the importance of macroeconomic closures in determining appropriate nbsorption rates. Use of Windfall Gains. Abstract from the ~ of absorption of windfall gains, and consider the question of how they are absorbed. Several aspects of this issue are addressed in Gelb (1985) using a static ve.sion of this modelj two other aspects are summarised below. The first is the choice between expanding public investment programs or increasing subsidies and direct transfers to the private sector. Poorer oil exporters have tended to follow the first policy. Richer exporters, with a wider range of fiscal options have included the second. The latter have, however, usually had more powerful organised labor movements and larger modern . .1t sectors, so that subsidies and transfers have become institutionalised. This , creates serious adjustment problems when world oil markets turn down. . The model's results suggest that the wider range of fiscal options of the richer exporters 1S valuable. Optimal transfers can raise the windfall's private value by nearly a third without reducing its social value below that of the base run because progressively lower quality and more costly pubLic "' .. _-, -0 ....... ....t-- .... ~~ investment is an inefficient means to disburse the consumption benefits of windfalls to households. But they also indicate that subsidies and transfers c~n be very costfy in the downturn if they become institutionalised. These results suggest that attempts to link any transfer payments explicitly to oil receipts be given serious consideration, despite their obvious administrative difficulties .1.1 The second question lS appropriate pricing of domestic oiL over the boom. Because oil is tradeable, economic efficiency argues for maintaining domestic prices at world levels. However, since pricing policy affects the distribution of income between public and private sectors, the relative value of public and private inco~e streams is a.: important determinant. This relative valuation depends on the discount rate (low discouncing favors public income because of the growth impact of public investment) and also on the feasible set of fiscal options and their flexibility over the oil boom. If government l~ under political constraints to spend windfall gains rapidly and the margin of fiscal adjustment is limited to public investment, results show that there may be a ·'::.hird-best" case for delaying i~creases in domestic oil prices until after the peak of the boom and then raising them rapidly. This is because the marginal social value of public investment can be driven below zero at the pea~ of the boom once account is t~ken of · subsequent recurrent costs and the dislocation of massive cutbacks. Fir~t- best policies would have involved s&ving a larger part of the win~fall abr.oad, second-best po!icies would have included less distortionary ways !f Shifting resources to t~e !II!' private sector such as lower nonoil taxe? or hib~r 1/ Such arrangements have been implemented in certain oil-r~rh reglons and were considered by policymakers in several developing countries but not implemented. transfers. Such a pattern of domestic oil pricing 1S actually not uncommon.l l It~ timing rend~rs the incL~~se particularly difficult since the nonoil ec:onomy ~ simultaneously experiencing the effects of contracti!lg demand. These results also suggest that a phase of lower nonoil tax effort may be appropriate when world oil markets are buoyant but that this should be used to develop a comprehensive tax system able to increase revenues when oil i~come declines. Yonoil tax effort &lackened in several exporters, but only Indonesia appears to have used its fiscal breathing slPce to institute u major tax reform. Capital Flight. Rapid depreciation of t~e Le~l exchange rate is usually only possible with nominal devaluation. If export~r:l ;\l;&' clleir exchange rates a.li.] restrict the return on domestic financial assets, the experience of several exporters shews that capital flight is likely when world oil markets turn down and exchange rate policy is seen as unsustainable. Results show that this shortens the period of adjustment to lower revenues and heightens the premium on good--and cautious--economic management. One group (speculators) with greater flexibility i3 able to throw the burden of adjustment onto less internationally mobile factors (labor). This greatly increases costs of adjustment. When capital flight follows the ~ peak of the boom as simulated in this p2~er, it is possible for the windfall ~U be negatively valued especially if spe~ulators' foreign holdings are not ! glven a positiv~ valuation. It is not clear that this provides a first-best 1/ Ecuador raised domestic oil prices from $0.18 per gallon to $0.60 in mid- 1982, Venezuelan oil prices, which had been constant for 30 years, tripled in 1982 and Indonesia raised internal prices 9hdrply after the end of 198!. argument for capital controls since the other policies above appear to be equally neces3a~y.ll , Prediction Errors. How i~portant for determining their value is the fact that windfa!l gains and reversals have beeu so poorly predicted? Results of the last set of exercises suggest that overopti~istic projection has serious consequences. Goverr~ent spending is first optimised 3ubject to a projection :~at real world oil prices will incre~se at 6% annually after the period of sharp price increase. As in Mexico (where planning embodied the same assumption), government borrows heavily against expec~ed future oil income and augments the impact of the oil windfall on absorption hy two-thirds. The ceitl t exchange rate appreciates by an additional 13%, real wages soar and foreign debt accumulates. At the peek of the boom the euphoric projection 1S corrected but the econvmy is locked into a real exchange rate-wage-debt configuration from w~ich exit is only possible at high cost in terms of prolonged domesti~ recession. Under a wide range of assumptions on macroeconomic closures, capital flovs a~d discount rates the economy is better off in the suboptimal base run. Indeed, receiving (and e,pecting) no windfall at all 1S often preferable. when capital fLight 19 induced and real wl'lges are downwardly sticky, the error results 1n the windfall being valued at -325% of its value in the base run.~1 1/ Similar cap~tal flow effects have been noted for nonoi1 count~ies. Corbo, de Melo and Tybout (1984) discuss the case of Chile. 2/ Rather than drawing up budgets on the basis of expected revenues, producers might therefore adopt a projection such as the "30th percentile rule" in their deve~opment planning. This and other options for budgeting under uncertainty with an asymmetric penalty function are discussed i the oil e~porter contex~ ~y Kreinheder and Stein~erg (1981), I A~O/ag16-a/11-1-85 ,- 36 - ?umming [Je. The above results confirm the critical importance of economic management 1n deteri'!lining the value of windfalls, especially if macroeconomic adjustment lS a~etric, capital flows are volatile and prediction is poor. I They also suggest a number of explanati.ons for the representative base run phasing of pubLic expenditu,e over oil windfalls. Of course, it could also be argueJ--with much weight--that public decisionmakers operate by rules of thumb rather thaD that they attempt to optimise, that poLicies emerge from fragment~d and inconsistent decisionmaking, that simple criteria such as those used in the paper are ina?propriate or that the model 1n the minds of those making decisions is very di£fere~t from the kind used 1n this paper. First, even if its overt social rate of time preference is low, government will come under pres~ure to spend rapidly if the time preference of individual agents is high because of the stimulative impact on incomes and consumption levels, If direct tran$f·,,:s are ruled out in favor of public investment this is effected largely--and inefficiently--through reat exc~~nge rate appreciarion. Second, analyses of the impact 0f windfalls have typically not 8d~ressed the relationship between choices made during the upswing and their consequences in the poss ible later' dOlolTIloITIturn, Therefore the' pol icy debate has not usually included potential costs in terms of recession which appear to be high. '~ird, higQ or rising world oil prices have typically been justified by producers on the:Srounds of scarcity. This implies still higher future prices and so is less easy to reconcile with cautious fiscal policies, If the seriOUS cQnsequ~nces oE major prediction errors are forseen, ~lsk aversion may dampe~ the urge to borrow against anticipated revenues. The base policy trajectory then emerges as a reasonable compromise. It is not so counter- AGD/ag16-a/11-1-85 - 37 - cyclical 3S would be desired at low discount rates given a high probability of the oil market later weakening. However, it satisfies to some degree politlcal pressures to disseminate windfalls rapidly, and over a range of . macroeconomic closures and discount rates it is better than full reliance on an optimistic forecast which might prove to be seriously wrong. - 38 - Table 1 ----- Base Run for a Twenty Year Horizon .----_. ~~-.------ Year 112 5 6 9 10 1 Terms of trarle effect/GDP~ 0.000 0.168 0.1)')] n.n l\eal private consumption inrleK 1.000 1. 380 1. 65-' ],4 Real puhlic consumption i nde'{ 1.000 1.284 1. 717 2.7 Real investment inde:< 1.000 1.670 l.Cfl5 4,1 Pu hlicit otal invest"1ent 0.430 0.518 I) · )76 0.2 Real exchange r.'lte l.000 1. 250 0.99/. 1.01 Private consumption deflator 1.00O 1. .138 0.990 o,9; Public consumption rleflator l.OOn 1.27n 0.980 1.0· Investment rleflator 1.000 1.n8 0.978 1.01 Real wage index: Traditional LO!,)O ,j. <}Y3 1.199 1, Ae· Morlern 1.000 1,201 l.nQ3 1 · '), Modern lahour forceltotal O. ]03 1).33? 0.372 n.4! Trade patterns (volumes) : rood imports -7 -548 -813 -26: Other agricultural exports 465 50 553 25' Oil exports 721 668 791 1 11anufactured imports -2426 -3599 -2410 -25 exports 282 293 989 1./1 Service imports -403 -558 -786 -lB Deht service ratio .252 .232 ,1.04 .J ---------- .. la GDP~ nonoil GLF It Table 2 Afternative Uses of the Oil Windfall A U E Constant world Food Direct 5% Kise in man. oil ccs subsidy transfers _cxeort growth Real Real Real Period niJi nd fa 11 " exchange private private private private private rate consumption consumption consumption consumption consumption 1 0.00 l.00 l.00 l.00 1.00 1.00 l.00 - 1.13 1.13 '1 O. II 1.17 l.lR 1.2 1. 27 J 0.17 1. 25 1. 1R 1. 29 1. 5:i 1. 52 1. 29 4 0.09 l.Og 1.50 1. tiS 1.63 1.63 1. 45 5 0.03 O.qq 1. 65 1.64 1.71 1. 74 1.65 f, o.()3 1.00 logO 1.8R 1. 9~3 1.99 l.89 7 0.02 1.00 2.20 2.16 2. 2.29 2.19 w R 0.02 1. 00 2.55 2.51 2.6,) 2.63 2.54 <.D <) 0.01 1.00 2.9R 2.92 3. Or; 3.05 2.97 to 0.(;1 1.00 3.49 3.43 3.4h 3.53 3.50 Valuation "private" ob;ecttve ul 12.3525 12.2294 12.530', 12.5477 12.2537 'social" objective 112 699.0411 6YB.14H7 696.73(,1 696.931f1 698/83;7 Termlnal real Olltput 42,011.1 41,353.2 39,4!.7,: 39,579.0 td.957.2 Z deviation from base -1.6 -6.1 -5.8 -0.1 ." SOllrce: Series of 7/6/83. - 40 .- Table 3 Windfalls when Adjustment is Not Smooth Run f" G H CL r 1 ----- Base Base with Base with rood with Ioiage sticky sticky wages subsidies subsidies Period rigidity prices and prices over boom Ioiage rigidi. Real Private Consumption 1 1.00 l.00 1.00 1.00 1.00 2 1.18 1.l9 1.19 1. 23 L23 3 1. 38 1. 38 l.J~ 1. c.6 1 ,46 4 1.49 1.49 1.48 1. 54 1.. 52 5 1.6L 1.62 1.-';·C i . !j~\' l.60 6 1. 87 1.83 L86 L89 1.84 7 2.18 2.L8 2.17 2.l8 2.LS 8 2.53 2.54 2.53 2.53 2.50 9 2.96 2.96 2.96 2.'15 2.92 La 3.47 3.48 3.47 3.46 3.1d Value U1 12.3246 L2.3322 12.3173 12.4243 L2.3867 Value U2 698.7597 698.7890 698.6734 698.7937 698.3822 Terminal non-oil real output 41,803.0 41,815.0 41,738.0 41,680.7 I.L,376.6 ! deviation from base -0.5 -0.5 -0.7 -0.8 _. t .6 · . ----- .----"--.-~-- ! ... - 41 - Table 4 Optimal Public Savings Abroad Summary Valuation: Run/ a Private Social A 12. 3525 699.0411 B 12.2294 698.1486 :