Fiscal Policy Formulation and Implementation in Oil-Producing Countries
Chapter

15 Issues in Domestic Petroleum Pricing in Oil-Producing Countries

Author(s):
Jeffrey Davis, Annalisa Fedelino, and Rolando Ossowski
Published Date:
August 2003
Share
  • ShareShare
Show Summary Details
Author(s)
Sanjeev Gupta, Benedict Clements, Kevin Fletcher and Gabriela Inchauste1 

I. Introduction

Petroleum product prices are often heavily regulated. Domestic price controls are prevalent, especially in countries that are net exporters of oil. Governments often keep prices well below international levels, resulting in the implicit subsidization of oil consumption. However, as these subsidies are typically not recorded in government budgets as expenditures, their economic cost, as well as the incidence on different income classes, is often poorly understood. The lack of readily available estimates of the size of these implicit subsidies has thus precluded a fuller discussion of their costs and benefits. Good fiscal policy management requires that the cost of all government activities, including such quasi-fiscal ones, be made transparent.2

The purpose of this paper is to shed light on subsidies for oil consumption in oil-producing countries. The paper first assesses the magnitude of the costs stemming from such subsidization. In addition, the discussion addresses the consequences of these subsidies for economic efficiency, income distribution, and fiscal policy.

The paper is organized as follows. Section II describes petroleumpricing policies in oil-producing countries and estimates the implicit subsidy on petroleum products in a wide sample of countries. Section III discusses the effects of pricing policies, including economic efficiency, equity, and macroeconomic performance, with a view to characterizing optimal pricing policy. Section IV then addresses issues in reforming pricing policies and eliminating implicit subsidies, including the preconditions for successful reform, the appropriate pace of reform, and the use and development of countervailing measures. Section V concludes.

II. Current Petroleum Pricing Policies

Description of Current Pricing Polices

Currently, a broad range of petroleum pricing policies exists across the world. In most Organization for Economic Cooperation and Development (OECD) countries, prices are market determined, although high excise taxes are usually levied on petroleum consumption. For example, total taxes on gasoline among the Group of Seven (G-7) countries range from US$0.10 per liter in the United States to US$0.81 per liter in the United Kingdom.3

In developing countries that are net importers of oil, prices are in some cases fixed by the government or by state-owned enterprises. In these cases, as well as in countries that have market-determined prices, excise taxes on petroleum are common; thus, the after-tax retail prices of petroleum products are typically higher than they would be in the absence of any taxes or government intervention.

In developing countries that are net exporters of oil, however, governments typically maintain domestic petroleum prices well below the free-market level. In most of these countries, the petroleum sector is dominated by a few large, state-owned enterprises, and the government typically controls both the wholesale and retail prices of petroleum products, either directly or indirectly, through export restrictions or other such measures. For example, in Kuwait in 1999, the after-tax price of gasoline was set at a level that was only 5 percent of the price that prevailed in the United States.4 These below-market prices result in the implicit subsidization of petroleum consumption; by selling petroleum domestically at a lower price than could be obtained abroad, the government forgoes revenue.

Quantifying the Implicit Subsidies/Taxes

Several studies have attempted to quantify the magnitude and importance of implicit petroleum subsidies in oil-producing and other countries (Rajkumar, 1996; International Energy Agency, 1999; Metschies, 1999; and Gürer and Ban, 2000). However, these studies either focus on only selected countries or petroleum products or are dated. In this study, we attempt to provide an update on petroleum subsidies for a wide spectrum of countries and products. In addition, we propose a methodological improvement over most earlier estimates by including implicit tax subsidies that might be granted in the form of exemptions from value-added taxes (VAT) or other consumption taxes.

Subsidies can be defined broadly as the difference between the reduced price of a good with government support and the price of the good in the absence of such support (World Bank, 1997; Schwartz and Clements, 1999; and Gupta and others, 2000). Implicit subsidy for petroleum product i in country j in time period t (Si,j,t) can then be defined as the difference between the “free-market price” (Mi,j,t) and the aftertax retail price (Pi,j,t) times the volume of consumption (Ci,j,t):

The free-market price is defined as the price that would prevail if there were no government interventions to affect the relative price of the product in question. In other words, the free-market price is the competitive market price plus the level of taxation typically levied on consumption goods. For net oil-importing countries, this free-market price should equal the cost of importing another unit of petroleum. This is calculated as the sum of the world wholesale spot price for the refined product (Wi,j,t), domestic distribution and marketing costs (Di,j,t), 5 and all general consumption taxes (Vi,j,t—general sales taxes, VAT, and so on). General consumption taxes are included because, as a first approximation, these taxes do not distort most relative prices6 and so should be included if the objective is to determine what the price of a petroleum product would be if it were given the same treatment by government policies as any other good. Thus, the free-market price is as follows:

For net oil-exporting countries, the free-market price is calculated in the same manner as for net oil importers, except that the cost of transporting petroleum from one country to another (Ti,j,t) is subtracted from the free-market price in equation (2) (since net oil exporters often do not import petroleum from other countries). Thus, for oil-exporting countries, the free-market price should represent the cost of consuming oil domestically, which is the opportunity cost of not selling the refined petroleum on the world wholesale market (Wi,j,t - Ti,j,t) plus the domestic distribution and marketing costs incurred (Di,j,t) plus the general consumption taxes paid (Vi,j,t).

Data on all of the components in equation (1) for 86 countries were gathered for 1995-2000 from different data sources, including the Energy Information Administration and Datastream, and subsidies for major petroleum product types were calculated. The detailed methodology for these calculations is described in the Appendix. Once the subsidy for each subproduct was calculated, the total petroleum subsidy was simply calculated as the sum.

Several caveats to this methodology should be noted. First, the estimates only measure the subsidy due to underpricing. Substantial subsidies may also arise from nonpayment of bills by consumers to state-owned oil-producing or -supplying enterprises, a practice that is especially prevalent in transition economies. The subsidy estimates in this paper do not include these nonpayments, and thus they may in some cases understate the total petroleum subsidy. Second, the domestic prices on which the subsidy estimates are based are in some cases the average for only a month or quarter rather than an entire year. Thus, the subsidy estimates in these cases should be viewed as a “snapshot” measure of the subsidy that is then annualized over the year.7

Finally, due to data limitations and in order to simplify the calculations, a constant average amount for domestic marketing and distribution costs is taken for each type of petroleum product based on data from G-7 countries. Similarly, a constant average amount for international transport costs is assumed for all destinations. These considerations introduce some measurement error since both types of costs will vary across time and country. However, previous studies have estimated that domestic marketing and distribution costs in developing countries are remarkably similar, on average, to domestic marketing and distribution costs in developed countries (Bacon, 2001), so the assumption of constant domestic costs should not introduce a significant bias. The assumption of constant international transport costs should not in general bias the results since these costs represent only about 2 percent of the retail price of petroleum products. However, it could understate the bottlenecks faced by some oil exporters in supplying additional petroleum output. For example, ice-blocked ports and pipeline capacity constraints can significantly increase marginal shipping costs of some oil producers.

Nonetheless, it should be noted that, using this methodology, the average estimate for net oil taxes is 1.6 percent of GDP for non-major oilexporting developing countries and 2.7 percent of GDP for non-major oil-exporting OECD countries. These estimates are similar to other directly measured estimates of petroleum taxation (Gupta and Mahler, 1995), again indicating that the methodology in this paper uses broadly correct reference prices in estimating net taxes/subsidies.

Table 15.1 shows the average calculated subsidies for 1999 for various groups of countries.8 The results indicate that major oilexporting countries are typically net subsidizers of petroleum products, whereas oil-importing countries are net taxers (negative subsidies). Moreover, the subsidies in major oil-exporting countries tend to be large, relative both to GDP and to government spending recorded in the budget.9 In 1999, the average subsidy in major oil-exporting countries was 3.5 percent of GDP and 15.2 percent of budget government expenditure.

Table 15.1.Domestic Petroleum Price Subsidies, 1999(Median values in parentheses)
Average Oil Subsidies1
(In percent of GDP)(In percent of government expenditure)
Major oil exporters2,33.5(2.5)15.2(8.6)
Of which: Subsidizing oil exporters44.4(3.7)18.7(10.8)
Other oil exporters5–1.9(–1.8)–7.0(–7.0)
Net oil importers6–2.2(–2.1)–8.3(–8.6)
Of which: Countries not producing oil7–1.7(–1.6)–7.0(–6.4)
Memorandum items:
Subsidy rate by type of product (in percent) for subsidizing oil exporters
Gasoline9(8)
Diesel48(46)
Residential light fuel oil71(78)
Number of countries per category:
Major oil exporters2,31515
Of which: Subsidizing oil exporters41313
Other oil exporters566
Net oil importers64140
Of which: Countries not producing oil72019
Source: IMF staff estimates. Based on exchange rates taken from IMF, International Financial Statistics.

Negative figure indicates net taxation of petroleum.

Based on a comparison of domestic production and consumption volumes.

Includes Algeria, Ecuador, Indonesia, Iran, Kazakhstan, Kuwait, Libya, Mexico, Nigeria, Norway, Qatar, Russia, Saudi Arabia, the United Arab Emirates, and Venezuela.

Comprises 13 countries with positive subsidies—all countries included in footnote 3 except Mexico and Norway.

Includes Argentina, Bolivia, Canada, Colombia, Denmark, and the United Kingdom.

Includes Austria, Barbados, Brazil, Chile, France, Germany, Greece, Guatemala, Hungary, India, Italy, Japan, the Netherlands, New Zealand, Peru, Poland, Romania, Spain, Suriname, Turkey, the United States, Belgium, Costa Rica, the Czech Republic, Dominican Republic, El Salvador, Finland, Grenada, Guyana, Haiti, Honduras, Ireland, Jamaica, Luxembourg, Nicaragua, Panama, Paraguay, Sweden, Switzerland, and Uruguay.

Includes Belgium, Costa Rica, the Czech Republic, Dominican Republic, El Salvador, Finland, Grenada, Guyana, Haiti, Honduras, Ireland, Jamaica, Luxembourg, Nicaragua, Panama, Paraguay, Sweden, Switzerland, and Uruguay.

Source: IMF staff estimates. Based on exchange rates taken from IMF, International Financial Statistics.

Negative figure indicates net taxation of petroleum.

Based on a comparison of domestic production and consumption volumes.

Includes Algeria, Ecuador, Indonesia, Iran, Kazakhstan, Kuwait, Libya, Mexico, Nigeria, Norway, Qatar, Russia, Saudi Arabia, the United Arab Emirates, and Venezuela.

Comprises 13 countries with positive subsidies—all countries included in footnote 3 except Mexico and Norway.

Includes Argentina, Bolivia, Canada, Colombia, Denmark, and the United Kingdom.

Includes Austria, Barbados, Brazil, Chile, France, Germany, Greece, Guatemala, Hungary, India, Italy, Japan, the Netherlands, New Zealand, Peru, Poland, Romania, Spain, Suriname, Turkey, the United States, Belgium, Costa Rica, the Czech Republic, Dominican Republic, El Salvador, Finland, Grenada, Guyana, Haiti, Honduras, Ireland, Jamaica, Luxembourg, Nicaragua, Panama, Paraguay, Sweden, Switzerland, and Uruguay.

Includes Belgium, Costa Rica, the Czech Republic, Dominican Republic, El Salvador, Finland, Grenada, Guyana, Haiti, Honduras, Ireland, Jamaica, Luxembourg, Nicaragua, Panama, Paraguay, Sweden, Switzerland, and Uruguay.

The degree of subsidization in oil-exporting countries varies significantly across products, countries, and time. As Table 15.1 shows, residential fuel oil and diesel fuel are subsidized more heavily than gasoline. Across countries, positive subsidies vary from 16.6 percent of GDP in Azerbaijan in 2000 to less than 1 percent of GDP in Libya in 1999 (Table 15.2). Across time, average petroleum product subsidies tend to rise when world crude oil prices are high and fall when crude oil prices are low. This occurs because governments tend to adjust domestic prices slowly in response to changes in world prices. For example, between January 1998 and January 2000, the spot price of refined gasoline rose by over 40 percent, yet the after-tax retail price of gasoline in Venezuela increased by less than 5 percent.10

Table 15.2.Domestic Petroleum Price Subsidies in Main Oil-Exporting Countries, 1996-2000(In percent of GDP)
Country19961997199819992000
Algeria3.02.22.5
Azerbaijan116.6
Ecuador3.31.4-1.512.6
Indonesia7.76.5
Iran
Official exchange rate10.04.2
Weighted average exchange rate213.212.6
Kazakhstan4.36.8
Kuwait5.15.14.8
Libya5.80.7
Mexico1.0-0.9-0.9-0.4
Nigeria2.5
Norway-4.5-4.8-3.9-4.5
Qatar3.93.22.4
Russian Federation5.0-5.08.0
Saudi Arabia7.47.05.9
United Arab Emirates1.91.41.7
Venezuela5.92.93.74.9
Average subsidy4.24.00.23.57.9
Sample size7116155
Memorandum item:
Average price of crude oil (US$ per barrel, U.K. Brent)20.519.112.717.728.3
Sources: Energy Information Administration (various years); IMF, International Financial Statistics; and authors’ estimates.

Based on IMF staff estimates using a different methodology (Wakeman-Linn and others, 2002). The figure includes only the subsidy due to underpricing; if the effects of nonpayment are also included, the effective subsidy in 2000 is estimated at 21.3 percent of GDP.

Based on a trade-weighted average of the official and market exchange rates.

Sources: Energy Information Administration (various years); IMF, International Financial Statistics; and authors’ estimates.

Based on IMF staff estimates using a different methodology (Wakeman-Linn and others, 2002). The figure includes only the subsidy due to underpricing; if the effects of nonpayment are also included, the effective subsidy in 2000 is estimated at 21.3 percent of GDP.

Based on a trade-weighted average of the official and market exchange rates.

III. Economic Effects of Petroleum Subsidies and Taxes

Is this subsidization appropriate? It depends on the effects of subsidies on economic efficiency and equity, the fiscal costs of these subsidies, and whether they contribute to or dampen cyclical fluctuations in economic activity

Efficiency

In general, economic efficiency requires marginal cost pricing. This implies that, in the absence of market imperfections and other price distortions, it would be most efficient to set the domestic price of petroleum equal to the net price that could be received on the world market.11 As noted earlier, the latter equals the sum of the wholesale spot price for the refined petroleum product (minus any international transport costs) plus the marketing and distribution costs that are saved by not consuming the good domestically. This is also the price that would result from a perfectly competitive free market, as it would eliminate arbitrage between domestic and world prices.

In reality there are several imperfections in this market. First, governments impose taxes in order to raise revenue. Taxation of petroleum may help minimize the efficiency losses of taxation in general, since the revenue collected allows taxes on other products to be lower. While the theoretically optimal rate of taxation on petroleum depends on many factors, a neutral hypothesis would be to assume that petroleum should be taxed at the same rate as other products.12 Under this taxation rule, the most economically efficient policy would be to set the price of petroleum equal to the opportunity cost of selling the product on the world market plus domestic distribution and marketing costs and any general consumption tax, which is the “free-market price” in equation (2).

Second, petroleum consumption is typically associated with several negative externalities, including air pollution and traffic congestion. These externalities generally argue for higher taxation of petroleum products in order to ensure that the price reflects these extra costs to society. How high should these taxes be? The answer depends on how large the externalities are, which is a subject of significant debate and uncertainty. However, a recent review of the research on this issue has estimated that the total environmental and congestion cost of gasoline consumption in the United Kingdom is between US$0.25 and US$0.40 per liter (Parry, 2001), an amount that would imply taxation equal to at least 100 percent of the free-market price. For developing countries, it is likely that the externality costs will be somewhat lower, since incomes are lower and people are therefore less willing to pay as much to reduce the environmental and congestion costs. Nonetheless, the externalities in these countries are still likely to be significant.

Finally, the domestic price of petroleum in some countries is likely to be influenced by their monopoly pricing power in world oil markets. In this case, the marginal revenue lost from not selling another unit of petroleum on the world market is not simply the free-market price; it also depends on the resulting change in the world price and the effect this has on the country’s existing oil export revenues.

To see this, note that the revenue from exporting refined petroleum, R, is equal to the world wholesale spot price of petroleum, W, plus the domestic distribution and marketing costs saved, D,13 minus the costs of international transport, T, times the quantity of exports, X:

Thus, the marginal revenue from exporting another unit of petroleum is the following:

where ηX,P is the price elasticity of demand for world petroleum and Xw is the total demand for petroleum on the world market.

For countries with a small share of the world market, the second term in brackets in equation (4) will be insignificant. For these countries, the opportunity cost of not exporting is simply equal to the world wholesale spot price plus the saved marketing and distribution costs minus international transport costs, as noted above. However, for countries that have a sizable share of the world market, the latter term in brackets is likely to be significantly negative so the opportunity cost of not exporting would be lower than the world price. If all other things are equal, this implies that it would be economically efficient to sell petroleum domestically at a lower price.14

How important might this consideration be in reality? Table 15.3 calculates the marginal revenue from exporting another unit of petroleum for the largest oil producers, using equation (4). Different long-run oil price demand elasticities are assumed,15 and each country’s oil exports as a share of world consumption are taken as given. It is further assumed that distribution, marketing, and international transport costs equal 75 percent of the wholesale price.16, 17Table 15.3 also presents marginal revenue for the Organization of Petroleum Exporting Countries (OPEC) as a whole under the assumption that the cartel could make pricing and production decisions collectively and enforce a single domestic pricing rule among all members. The marginal revenue is expressed as a fraction of marginal revenue in the no-market-power (small exporter) case.

Table 15.3.Marginal Revenue from Exporting for Largest Oil Exporters
Exports as Share of World Consumption1Marginal Revenue2
(η =-0.25)(η = -0.50)(η=-0.75)
All OPEC32.70.250.630.75
Saudi Arabia10.80.750.880.92
Russia5.10.880.940.96
Norway4.30.900.950.97
Venezuela4.00.910.950.97
Sources: Energy Information Administration (various years); and authors’ estimates.

For 1998 (Energy Information Administration, 2001).

As fraction of small exporter (no-market-power) case; assumes domestic distribution and marketing costs minus international transport costs are 75 percent of the wholesale price; η refers to the price elasticity of demand for world petroleum.

Sources: Energy Information Administration (various years); and authors’ estimates.

For 1998 (Energy Information Administration, 2001).

As fraction of small exporter (no-market-power) case; assumes domestic distribution and marketing costs minus international transport costs are 75 percent of the wholesale price; η refers to the price elasticity of demand for world petroleum.

If each country determines its prices independently, marginal revenue is significantly less than in the no-market-power case only for the largest exporter, Saudi Arabia; in the case of that country, marginal revenue is still 75 percent of the no-market-power case, even assuming a demand elasticity as low as -0.25. If, however, OPEC is able to successfully coordinate pricing and export decisions across all members, then the marginal revenue from exporting may be as low as 25 percent of the no-market-power case, which, all other things equal, would argue for lower domestic prices in these countries.

Some caveats to this monopolistic pricing/terms-of-trade argument for lower domestic pricing should be noted. First, this monopoly pricing is only optimal from the viewpoint of the oil-exporting country. From a worldwide viewpoint, welfare is maximized by eliminating differences in prices among countries. Second, the calculations in Table 15.3 ignore the reactions of large oil-importing countries that may have monopsonistic power in world markets. These countries may have a symmetric incentive to influence the world price of oil by taxing consumption in their domestic markets. If monopolistic pricing behavior by large oil exporters encourages large oil importers to engage in retaliatory monopsonistic behavior, this will reduce the incentive to price in a monopolistic manner.

In sum, various considerations enter into the efficient pricing of petroleum products.18 Precise calculations are therefore difficult, as there is significant uncertainty regarding key parameters such as the environmental effects of oil consumption and the long-run elasticity of oil demand. However, for countries with no market power, the negative externalities and the need for government revenue clearly indicate it is most efficient to tax petroleum at a rate that is at least equal to that imposed on other consumption goods. For the very largest oil exporters, the most efficient rate of taxation (from the country’s viewpoint) may be slightly lower than in other countries, but it is likely to still be significantly positive, as the estimates discussed earlier indicate that the factors favoring taxation (negative externalities and the need for government revenue) will still outweigh the factors favoring subsidization (market power/terms-of-trade arguments for protection). For example, Table 15.4 presents optimal subsidy/tax rates for major oil exporters, based on assumed environmental externalities of US$0.10 per liter, an optimal uniform consumption tax of 10 percent, and a price elasticity of demand of -0.5. It can be seen that net taxation is optimal under these assumptions even if one considers OPEC as a unified producer and seller.19Table 15.4 also estimates optimal subsidy (tax) rates excluding all environmental externalities. In this extreme case, significant net subsidization would be optimal only for OPEC as a unified producer and seller; however, the optimal rate of subsidization (31 percent) would still be only half the level of subsidization that actually prevailed in 1999.

Table 15.4.Deadweight Loss from Subsidies in Selected Countries, 19991
Assuming Environmental Externalities of US$0.10/LiterAssuming No Environmental Externalities
CountryActual Subsidy Rate (In percent)Optimal subsidy (tax) rate (In percent)Deadweight loss2(In percent of GDP)Optimal subsidy (tax) rate (In percent)Deadweight loss2(In percent of GDP)
Algeria37-571.4-100.4
Ecuador-14-650.5-100.0
Indonesia78-625.4-102.8
Iran387-6412.4-106.9
Kazakhstan46-654.4-101.5
Kuwait95-716.0-103.5
Libya10-650.8-100.1
Nigeria22-611.7-100.3
Qatar32-571.3-100.4
Russia51-584.9-31.6
Saudi Arabia53-443.041.0
United Arab Emirates30-811.5-100.3
Venezuela72-552.7-51.3
OPEC as a unit62-222.2310.4
Sources: Energy Information Administration (various years); and authors’ estimates.

Assumes an optimal, nondistortionary uniform consumption tax of 10 percent, η = -0.5, and market power effects as estimated in Table 15.3. Negative numbers imply optimal net taxation.

Calculated as area C in Figure 15.1, using an assumed constant compensated elasticity of demand function (Q = PηX, where n is the compensated elasticity of demand and X is a constant calculated for each country using the observed Q and P).

Based on the trade-weighted average exchange rate.

Sources: Energy Information Administration (various years); and authors’ estimates.

Assumes an optimal, nondistortionary uniform consumption tax of 10 percent, η = -0.5, and market power effects as estimated in Table 15.3. Negative numbers imply optimal net taxation.

Calculated as area C in Figure 15.1, using an assumed constant compensated elasticity of demand function (Q = PηX, where n is the compensated elasticity of demand and X is a constant calculated for each country using the observed Q and P).

Based on the trade-weighted average exchange rate.

Figure 15.1.Deadweight Loss from a Subsidy

Despite the apparent optimality of net taxation of petroleum, few oil-exporting countries pursue such a policy, resulting in significant economic losses. Rather, as Table 15.1 indicates, they typically subsidize consumption. This deviation from efficient pricing results in a deadweight welfare loss. This can be seen in Figure 15.1, where the loss in government/exporter revenue from subsidization (area ABC) is greater than the increase in domestic consumers’ surplus (area AB). Table 15.4 presents estimates for the magnitude of this deadweight loss (area C in Figure 15.1) as a percentage of GDP for several major petroleum exporters. For 1999, these losses ranged from 0.5 percent of GDP in Ecuador to 12.4 percent of GDP in Iran.20 High levels of domestic consumption resulting from these price distortions can even turn net exporters into net importers, as some analysts forecast may eventually happen in Iran.21

Equity

Subsidies are not the most efficient means of redistributing income or of improving the poor’s access to energy. The impact of subsidies on equity can be evaluated by their relative efficacy—whether they reach those for whom they are intended—and the ease of administration, particularly given the incentives for smuggling and corruption. This section attempts to assess subsidies in each of these respects.

The available evidence shows that higher-income households consume larger quantities of petroleum products and electricity in oil-producing countries and thus benefit relatively more from subsidies.22 For example, in the mid-1990s in Mexico and Ecuador, over 30 percent of household electricity consumption was accounted for by the top quintile (Table 15.5). Outside Latin America, Saboohi (2001) finds that in Iran the energy consumption of a poor household was 44 percent of that of a rich household. Similarly, in Venezuela, the richest 20 percent of the population received 6½ times more in subsidies per person than the poorest 33 percent of the population; while 38 percent of electricity subsidies went to the top 20 percent of the population, only 16 percent accrued to the bottom third (World Bank, 1995). In Ecuador, the more expensive energy products (electricity and liquefied petroleum gas (LPG)) received the highest subsidies, while household kerosene (kerex), which poor households considered convenient and versatile, was not subsidized (UNDP/World Bank, 1994).

Table 15.5.Mexico and Ecuador: Share of Total Expenditures by Quintile(In percent of total spending)
QuintileElectricityGasoline1Gas/LPGPetrol/KerexOthers2
Mexico
15.40.96.715.126.1
210.23.913.818.324.0
315.99.318.520.220.4
421.619.425.121.717.4
546.866.535.924.812.1
Ecuador
19.215.243.2
215.323.220.6
316.220.113.8
426.422.17.6
532.919.414.8
Sources: UNDP/World Bank (1994); and IMF staff calculations based on the 1996 Mexican Income and Expenditure Household Survey (INEGI, 1996).

Gasoline refers to gasoline, diesel, or gas purchased as fuel for vehicles.

Includes coal, firewood, heating oil, candles, and other items such as paper or cardboard.

Sources: UNDP/World Bank (1994); and IMF staff calculations based on the 1996 Mexican Income and Expenditure Household Survey (INEGI, 1996).

Gasoline refers to gasoline, diesel, or gas purchased as fuel for vehicles.

Includes coal, firewood, heating oil, candles, and other items such as paper or cardboard.

Relatively better-off groups also benefit more when quantities well above what the poor need are subsidized. In Yemen, for example, the lifeline electricity rate was set at a level of consumption that covered more than 75 percent of the population (Barnes and Halpern, 2000).

The pro-rich bias of subsidies is further compounded by smuggling and corruption. Subsidies create the incentive for smuggling petroleum products to markets where prices reflect market conditions. In Nigeria, for example, petroleum prices are a third of those prevailing in neighboring Niger or Cameroon, leading to widespread smuggling and chronic fuel shortages in many parts of the country.

Despite their pro-rich bias, the sudden removal of subsidies can have adverse social effects in the short run. This is particularly the case when a country lacks adequate mechanisms for shielding poor households from the resulting higher prices. Poor households may spend a significant share of their incomes on energy, and in the absence of compensating mechanisms, the ensuing loss in their consumption could be large. For example, in Ghana kerosene, diesel, and gasoline prices rose in real terms by 161, 214, and 156 percent, respectively, from 1983 through 1987. Since kerosene was a major fuel for the poor (as shown by the expenditure shares), estimates of the loss in consumer surplus for different magnitudes of price elasticity show that households in the bottom two quintiles lost the most (Hope and Singh, 1995).

Fiscal Costs

Subsidies also have substantial fiscal costs for oil-producing countries. In many such countries, the cost of subsidies exceeds the overall fiscal deficit, which averaged 2.1 percent of GDP in the second half of the 1990s in subsidizing oil-exporting countries. Given high levels of public debt in these countries (61 percent of GDP, for the four countries for which data are readily available), the potential gains (in terms of reduced fiscal vulnerability) from further fiscal consolidation could be sizable.23

If fiscal consolidation is not required, the revenue forgone by governments through the subsidization of petroleum consumption could be used to reduce tax rates or increase more productive spending, such as that for infrastructure and human capital formation.24 For example, if subsidizing oil-exporting countries eliminated petroleum subsidies, they could increase spending on health, which is low in relation to the average in middle-income countries (Figure 15.2 and Table 15.6). This should contribute to boosting indicators of health status in oil-producing countries, which in some cases are below those in other countries with similar or lower per capita income.25 In Venezuela, for example, immunization rates are below the level achieved by lower middle-income countries, despite per capita income that is about double the average for this group. Education spending is also low in some oil-producing countries (e.g., Nigeria).

Figure 15.2.Composition of Government Expenditure in Major Subsidizing Oil-Exporting Countries1

(In percent of GDP)

Source: IMF country documents.

1Unweighted averages for subsidizing oil-exporting countries. Oil subsidies are for 1999; expenditure data reflect averages over the period 1997-99. Number of countries in parentheses.

Table 15.6.Social Indicators and Social Spending in Subsidizing Oil-Exporting and Other Countries, 1997-99
Life ExpectancyIlliteracy Rate (In percent of adult population)Infant Mortality Rate (Per 1,000 live births)Under 5 Mortality Rate (Per 1,000 live births)Immunization Rate Against Measles (In percent of children aged less than 12 months)GDP Per Capita (Constant 1995 US$)Education SpendingHealth Spending
(In percent of GDP)(In percent of total expenditure)(In percent of GDP)(In percent of total expenditure)
Subsidizing oil-exporting countries (13)169202636883,2595.015.72.05.8
Of which:
Algeria71353539761,5348.828.01.54.9
Ecuador6992937871,5213.115.60.94.8
Indonesia65144354801,0250.63.3
Nigeria493982149482531.47.60.73.8
Saudi Arabia72251926926,8418.823.12.97.6
Venezuela7382124793,4513.819.71.47.0
Low-income countries (56)1554175126674234.015.02.06.8
Lower-middle-income countries (42)168163344861,6424.816.62.37.5
Upper-middle-income countries (26)171111826905,3064.714.43.39.7
Sources: World Bank, 2001, World Development Indicators. Expenditure data from IMF staff reports.

Numbers in brackets refer to the numbers of countries in the different groups in 1999.

Sources: World Bank, 2001, World Development Indicators. Expenditure data from IMF staff reports.

Numbers in brackets refer to the numbers of countries in the different groups in 1999.

Cyclicality

The policy of maintaining below-world-market prices for petroleum products has important implications for macroeconomic management. As noted earlier, many oil-exporting countries tend not to adjust domestic prices fully in response to changes in world prices. For net oil exporters, this means that implicit subsidies will be procyclical—subsidies will increase when world oil prices increase, which also tend to be periods of economic expansion for these countries. The procyclicality of subsidies will thus exacerbate the effects of oil price shocks on economic volatility.

Available data suggest that subsidies have a substantial procyclical bias. Figure 15.3 plots the changes in subsidies against real GDP growth per capita across countries. The coefficient implies that a 1 percentage point increase in per capita GDP growth is associated with an increase in oil subsidies of 0.6 percent of GDP.

Figure 15.3.Change in Subsidy and Real GDP Per Capita Growth1,2

1 Excludes one outlying observation when real GDP growth per capita was in excess of 20 percent. Robust White (1980)t-statistics in parentheses of regression.

2Dates provided in parentheses indicate the period over which the change in subsidy was calculated.

One way to eliminate the procyclicality of oil subsidies is to tie domestic petroleum prices to fluctuations in international markets. This can, however, lead to wide swings in domestic prices, thereby increasing uncertainty about government revenue and investment decisions. Under these circumstances, governments could adopt partial pass-through rules and use financial hedging instruments.26

Partial pass-through rules can help smooth out the path of domestic petroleum prices. These include moving-average rules (which base retail prices on a moving average of past spot prices), trigger rules (which update only if spot prices change by a predetermined trigger amount), or max-min rules (which place a ceiling and a floor on the level of retail petroleum prices). For example, Chile has a max-min price stabilization law under which reference prices are updated weekly.

Hedging instruments can also mitigate oil price fluctuations. For example, the state of Texas hedges its heavy reliance on oil revenue by buying and selling options in order to narrow the range within which its revenue stream fluctuates. Oil-producing countries do not disclose whether they use hedging instruments because this is market-sensitive information; however, it is believed to be a common practice in a number of countries. Since risk management activities require considerable knowledge of financial instruments and an appropriate institutional framework, the human and institutional requirements to implement and monitor hedging operations are significant. In particular, if institutions in a country are weak, there may be scope for rent seeking. It is important that government officials be held accountable for their hedging decisions.

IV. Reforming Petroleum Pricing Policies

Petroleum price subsidy reform entails reducing generalized subsidies by raising prices and targeting subsidies to the poor and vulnerable population groups. Subsidy reform has not been easy for most countries because of the opposition from losers of subsidization, the likely impact of higher prices on the poor and on production costs, and the inability of the government to protect the poor because of weak administrative capacity.

What lessons does the past provide for successful subsidy reform? In particular, what are the preconditions for successful reform and what are the appropriate timing and sequencing of reforms? What countervailing measures could be adopted to address any adverse distributive consequences of reform, and how might poverty and social impact analysis be useful? This section addresses each of these questions in turn.

Preconditions for Reform

The decision to liberalize the petroleum market, if met with strong protest and social unrest, can undermine reform efforts. For example, in June 2000 Nigeria increased petroleum prices by 50 percent, leading to a general strike and riots, followed by a subsequent reversal of government policy. In Indonesia, an attempt to limit the fuel subsidy in 1998 was met with violent protests, forcing the government to partially roll back planned price increases. Similar protests have occurred in Venezuela and Ecuador.

The risk of political disruption is highest when rapid reform is attempted without credible social protection mechanisms in place and without adequate attention to building political consensus on the need for reform. To assess the political risks associated with price-subsidy reform and inform its design, policymakers could, to the extent feasible, do the following:

  • Identify winners and losers in price-subsidy reforms. This can be done by examining the benefits of existing subsidies for different income classes and then identifying the characteristics of winners and losers.27
  • Assess the political strength and the magnitude of the losses or gains of each group.
  • Assess the feasibility and cost of alternative measures to protect the consumption of the poor or politically vocal groups.
  • Generate political support.

Subsidy reductions should be embedded in a reform program that engenders broad support and yields widespread benefits. This stakeholder approach (Graham, 1994) implies that governments should avoid reforms that impose an unfair burden on a narrow group of socioeconomic or ethnic categories. Compensatory measures for the poor can help to gain the support of an important constituency. For example, Indonesia was able to reduce fuel subsidies in October 2000 and April 2001 after earlier failed attempts by compensating poor consumers through a combination of public works programs, microcredit, and cash assistance. This process was helped in part by growing acceptance among the population that fuel subsidies were not well targeted. In the longer term, sustained support for reforms should be crafted in a consultative process, with adequate poverty and social impact analysis (PSIA) to inform policy choices (see below).

Finally, publicity campaigns to discuss the trade-offs involved in providing subsidies can be useful in fostering support for reform. Successful reforms have often been accompanied by effective government communication to the population regarding the trade-offs and rationale for reform. This includes presenting the cost of subsidies in concrete terms and/or explaining how the reform package affects real household income.28 For example, in Egypt the budgetary burden of subsidies was compared with revenues from the Suez Canal.

Timing and Sequencing of Petroleum Pricing Reforms

The optimal speed of subsidy reform depends on several factors. First, the government must take into account fiscal considerations. There is a trade-off between rapidly cutting budget-financed subsidies and avoiding an adverse impact on the poor. Although a one-time adjustment of prices to eliminate subsidies can yield immediate budget savings and quickly correct distortions in resource allocation, it can also result in a sudden and significant decline in household consumption, especially for low-income households. Although there may be a potential for rapid budgetary savings from the elimination of implicit subsidies, a sustainable reform will also require substantial outlays to protect the poor.

A second consideration regarding the speed of reform is the availability of social protection instruments. Such instruments require not only resources, but also a system to deliver compensation to those who need it. In general, if existing social protection instruments can be adapted to the needs of the poor during reform, then reform can be rapid. However, if new social safety net instruments need to be established, then the speed of reform will be affected by the administrative capacity to design and implement adequate and well-targeted social protection. In this regard, gradual adjustment may minimize the adjustment costs faced by businesses and individuals. A sudden increase in prices of petroleum products can affect the viability of businesses, potentially leading to job losses.

Third, governments must consider their political constraints. Rapid reform is feasible only when governments are politically strong and social disruption from implementing reforms is unlikely. Gradual adjustment has the benefit of giving the government time to assess and react to unintended consequences, including adverse political repercussions. Furthermore, insufficient institutional and administrative capacity to protect the poor can weaken political support for reforms.

Finally, the external environment may also affect the pace of reform. Specifically, more rapid reform may be possible under favorable external circumstances, such as low international prices. For example, in Guinea, prior to the adoption of an automatic adjustment mechanism for the retail prices of petroleum products, the price of gasoline at the pump was reduced in 2001 to reflect the decline in international oil prices.

Despite these advantages, a gradual pace of reform also has drawbacks. First, it extends the time frame for reaping budgetary gains and, by slowing the adjustment of prices, it diminishes the incentive for the private sector to rapidly switch to energy-efficient technology. It may also lead to policy reversals, especially if it is adopted to postpone politically difficult reforms.

In the absence of a formal price-setting mechanism, subsidies can quickly reemerge regardless of whether reform is rapid or gradual. For example, in Ecuador one-time adjustments were undertaken, but no automatic adjustment mechanism was put in place, leading to the reemergence of subsidies (Box 15.1). Exchange rate changes can also lead to the reemergence of subsidies when domestic prices are fixed. For example, in Indonesia, the resurgence of massive fuel subsidies was due to the large depreciation of the exchange rate coupled with the lack of an automatic price-setting mechanism.

The problem of policy reversal with gradual reform can be overcome by adopting and making public a detailed timetable of reform measures. For example, in Cameroon the authorities decided to phase out the petroleum subsidy by cutting it in half in 2001/02 and completely eliminating it in 2002/03.

Distributive Considerations and Countervailing Measures

The adverse social effects of reforming energy subsidies noted earlier can be mitigated through appropriate social protection mechanisms. Poverty and social impact analysis (PSIA)—which aims to assess the ex ante, during, and ex post consequences of policy interventions on the well-being of the poor—can help weigh the costs and benefits of alternative strategies prior to implementation and therefore inform the design of the reform program itself.29 Such an analysis could help evaluate the timing and the sequencing of reforms as well as the need for countervailing measures.

BOX 15.1.Ecuador: Experience with Energy Price Subsidy Reform

Ecuador provides a good case study of how the lack of a formal pricesetting mechanism can hinder the implementation of lasting reforms of petroleum subsidies. Traditionally, Ecuador has granted subsidies to consumers for cooking gas, electricity, and some other commodities (mainly fuels and other utilities) and subsidies have continued to reemerge. The government grants these subsidies by fixing consumer prices below the opportunity cost—sometimes even below the cost of production—of these goods and services. In September 1998, the government decreed an increase of more than 400 percent in the price of cooking gas, thereby eliminating the existing subsidy. However, the new price was fixed in sucres and the subsidy quickly reappeared as the price of petroleum increased with exchange rate depreciation. In May 2000, the government introduced price increases of 65-92 percent for diesel and gasoline and of 90-333 percent for other derivatives. This sharply lowered the subsidy on fuels from an average of 35 U.S. cents per gallon to an average of 8 U.S. cents per gallon, but it left unchanged the subsidy on cooking gas. The government also implemented several increases in electricity prices: in September 1998, in January 1999, and in May 2000, when prices were raised by an average of 70 percent (combined with 4 percent monthly adjustments thereafter). The direct subsidy to small consumers was replaced with a system of cross-subsidies between large and small consumers. This increase was intended to allow electricity companies to cover their own operating costs, but prices were still significantly below long-term marginal cost. In 2001, a decree doubling the price of cooking gas and increasing petroleum prices was reversed—following massive protests in February of 2001—and instead the government agreed to freeze petroleum prices for a year and cut cooking gas prices.

Source: Offerdal and others (2000).

For example, an ex ante social impact study for Iran (Saboohi, 2001) estimated that an increase in energy prices with the aim of eliminating energy subsidies would have a serious impact on household consumption and inflation. However, estimates showed that if part of the additional resources obtained through the elimination of energy subsidies were allocated to strengthen the budgetary balance (thus containing inflation) and the rest of the resources were allocated to targeted programs for the poor, then rural populations and the poorest three deciles of urban households would be better off.30

Countervailing measures that can be implemented to minimize poverty and social impacts include cash transfers or limiting subsidies to a subgroup of the population. Cash compensation, in some cases, has taken the form of a separate benefit in lieu of the subsidy. In 1998, for example, Ecuador eliminated the cooking-gas subsidy and instituted a cash transfer program for poor families and the elderly. In the wake of the 1999 financial crisis, this cash transfer program became the backbone of the government’s safety-net strategy. In other cases, cash benefits have been merged with existing social benefits.

Cash transfers have several advantages: they allow for consumer choice, their cost to the budget is explicit and known with greater certainty than generalized subsidies, and they can be targeted to the poor. However, their real value may erode quickly during periods of high inflation and they are prone to corruption. For example, although the safety-net program in Ecuador has been quite successful in providing relief to a large segment of the poor population, more than a quarter of the current recipients are not eligible (World Bank, 2000a). In addition, since these transfers are disbursed through the banking system, which is far better developed in cities and urban areas than in rural areas, the distribution tends to favor the urban population. The coverage of these transfers is, therefore, well below target in rural areas.

Other countervailing measures include limiting price subsidies to a subgroup of the population. For example, in Ecuador, the cash transfer program was complemented by targeted electricity subsidies to poorer consumers financed through higher prices charged to wealthier consumers. A more common approach among oil producers is to tax relatively inelastic products such as gasoline and subsidize so-called “social products” such as kerosene and diesel. This practice is common in Algeria, Egypt, Iraq, and Syria (Al-Faris, 1997). In Yemen for example, gasoline prices were increased every year from 1996 through 1999, so that by 1998 they were 80 percent above world market prices. This allowed for cross-subsidies for diesel and, to a lesser extent, for fuel oil and kerosene. This policy, however, added to the discrepancy between domestic and border prices and suffered from the efficiency problems discussed earlier; furthermore, it provided incentives for industries and the nonpoor to shift their consumption to the subsidized products.

V. Conclusions

This paper finds that major oil-exporting countries tend to be net subsidizers of petroleum, while oil-importing countries tend to be net taxers. Implicit subsidies in major oil-exporting countries are large, equaling 3.5 percent of GDP and 15.2 percent of explicit government expenditure, on average, in 1999. Subsidies in major oil-exporting countries vary over time, with a marked procyclical bias. With respect to variation by product, residential fuel oil and diesel are subsidized more heavily than gasoline.

Subsidization does not appear to be a wise use of resources. From an efficiency point of view, it would be best for oil-exporting countries to set the domestic price of petroleum equal to the world market price (assuming no market imperfections). Taking into account market imperfections complicates the analysis but generally supports the notion that prices should be no lower than world market levels. From an equity point of view, subsidies tend to be an undesirable method of redistribution, either because they benefit all users (including the rich) or because they are indiscriminate, allowing for consumption well above what is needed by the poor. The pro-rich bias of subsidies can be further compounded by smuggling and corruption. From a fiscal perspective, the opportunity costs of these subsidies are substantial.

Despite the substantial costs of implicit petroleum subsidies, reform is difficult, as there is strong popular opposition to their elimination. Subsidy reform should be embedded in a reform program that engenders broad support and yields widespread benefits. This could include the use of countervailing measures and vigorous publicity campaigns to educate the population on the trade-offs involved in providing subsidies versus other social services. The speed of subsidy reform will depend on the required size of fiscal adjustment, the availability of social protection instruments, the strength of the government, and the administrative capacity to implement reforms. The adverse social and political effects of reforming energy subsidies can be mitigated by undertaking poverty and social impact analyses and establishing safety nets. These safety nets can include cash compensation for the most vulnerable, limiting price subsidies to a subgroup of the population, or tax exemptions for the poor. To inform the design, pace, and implementation of reform, PSIA should be undertaken prior to, during, and after a reform to ensure that appropriate mitigating measures are in place.

Appendix. Calculation of Implicit Subsidies

This appendix explains how implicit subsidies are calculated. Conceptually, the implicit subsidy for petroleum product i in country j in time period t (Si,j,t) is defined as the difference between the “free-market price” (Mi,j,t) and the after-tax retail price (Pi,j,t) times the volume of consumption (Ci,j,t):

To calculate this subsidy, data on each of these three variables are calculated as follows. Data on the after-tax retail prices of major petroleum products are obtained from various editions of the International Energy Annual (Energy Information Administration, various years). Data are obtained for gasoline,31 motor diesel, residential light fuel oil, kerosene, residential liquefied petroleum gas (LPG),32 industrial light fuel oil, and industrial heavy fuel oil.33 Data are collected for the years 1995-2000. For most countries, however, data are available for only some years. Prices for each year come in various forms (either from a particular month, a quarter, or an annual average). There are 86 countries in the sample and all prices are expressed in U.S. dollars.

The data on consumption are obtained from the Energy Information Administration website.34 These data are divided into gasoline, jet fuel, kerosene, distillate fuel oil, heavy fuel oil, LPG, and other. For most countries, data are available for 1995-98. For some countries, data are also available for 1999. Missing consumption data for 1999 and 2000 are imputed by assuming that consumption grows at the rate of real GDP, measured in domestic currency. All GDP and exchange rate information are from International Financial Statistics (International Monetary Fund, various years).

The free-market price is defined as the price that would prevail if there were no government interventions to affect the relative price of the product in question. For net oil-importing countries, this freemarket price should be equal to the cost of importing another unit of the good. This is calculated as the world wholesale spot price for the refined product (Wi,j,t) plus domestic distribution and marketing costs (Di,j,t)35 plus all general consumption taxes (Vi,j,t)—general sales taxes, VAT, etc.). General consumption taxes are included because, as a first approximation, these taxes do not distort most relative prices36 and so should be included if the objective is to determine what the price of a petroleum product would be if it were given the same treatment by government policies as any other good. Thus, the free-market price is as follows:

Data on world wholesale spot prices are obtained from Datastream. For each of the following retail prices, the corresponding wholesale spot price is used: (i) premium gasoline—New York premium unleaded nonoxygen gasoline; (ii) regular gasoline—New York regular unleaded nonoxygen gasoline; (iii) motor diesel—New York low sulfur (0.5 percent) diesel fuel; (iv) light fuel oil (both residential and industrial)—New York No. 2 fuel oil; (v) kerosene—Singapore jet kerosene; (vi) LPG—Mount Belvieu propane; and (vii) heavy fuel oil—Northwest Europe heavy fuel oil (3.5 percent sulfur). There are four or five major world markets for each type of product (mainly New York harbor, the U.S. Gulf Coast, Los Angeles, Singapore, and Rotterdam). The choice of market is based on data availability from Datastream. However, prices in these markets usually do not differ widely for long periods of time, so the use of only one market should not introduce significant bias into the estimates. For each observation, the frequency of the wholesale price (i.e., whether it is a monthly, quarterly, or annual average) is chosen so as to match the frequency in which the corresponding retail price is reported. All prices are in U.S. dollars.

For gasoline, motor diesel, light fuel oil, and heavy fuel oil, domestic distribution and marketing costs are measured as the difference between the average before-tax retail price across eight OECD countries taken from Monthly Price Statistics (International Energy Agency, 2001b)37 and the wholesale price of each of these products in November 2001. For kerosene and LPG, costs are calculated as the difference between the average before-tax retail price to end users in the U.S. from the Annual Energy Review (Energy Information Administration, 2000) and the average annual wholesale price for 2000. The resulting estimates for domestic distribution and marketing costs in U.S. cents per gallon are 36.4,44.3,35.6,19.8,59.1, and 15.2 for gasoline, motor diesel, light fuel oil, jet kerosene, propane, and heavy fuel oil, respectively.

To simplify the calculations, these domestic distribution and marketing costs are kept constant over time and country for each product. This introduces some measurement error since these costs will vary across time and country. However, previous studies have estimated that domestic marketing and distribution costs in developing countries are remarkably similar, on average, to domestic marketing and distribution costs in developed countries (Bacon, 2001), so the assumption of constant costs should not introduce a significant bias.

Data on general consumption taxes are obtained for each country from Corporate Taxes: Worldwide Summaries (PricewaterhouseCoopers, various years) and internal IMF databases. The taxes include the standard rate of any VAT or general sales tax that is applied at the central government level.

For net oil-exporting countries, the free-market price is calculated in the same manner as for net oil importers, except that the transportation cost of shipping petroleum from one country to another (Ti,j,t) is subtracted from the free-market price since oil exporters need not import petroleum from other countries.38 These transportation costs are estimated as US$9.71 per metric ton, which was the average for all routes between 1996-2000, based on data from the OPEC Annual Statistical Bulletin 2000 (Organization of Petroleum Exporting Countries, 2001). The assumption of constant international transport costs should not in general bias the results since these costs represent only about 2 percent of the retail price of petroleum products. In some countries and under some conditions, however, they may be substantially higher. In the former Soviet Union, for example, ice-blocked ports and pipeline capacity constraints may significantly increase marginal transport costs.

Another complication that arises in calculating subsidies is that available consumption data are divided into somewhat different categories than the price data. This problem is most pronounced for distillate fuels, which are lumped into one group in the consumption data but for which three different retail prices exist (motor diesel, residential light fuel oil, and industrial light fuel oil). For most OECD countries, however, the breakdown of distillate fuel consumption into the various subproducts does exist in the Energy Statistics of Non-OECD Countries 1997-98 (International Energy Agency, 2001a).39 Thus, it is assumed that all countries divide their consumption of distillate fuels into the three categories in the same proportions as the OECD average for 1998. Similarly, jet fuel and kerosene are provided as separate categories in the consumption data, whereas only data on kerosene are provided in the retail price data, and only data on jet fuel are provided in the spot price data. Thus, kerosene and jet fuel are lumped together as one consumption category, and the retail price for kerosene is compared to the market price for jet fuel. In addition, retail prices for some products are missing for many countries. In these cases, it is assumed that the subsidy rate on missing prices is equal to the subsidy rate on motor diesel. If the retail price of motor diesel is missing, it is assumed that the subsidy rate on missing prices equals the subsidy rate on residential light fuel oil.

Bibliography

    AlbouyYves and NadiaNadifi2000“Impact of Power Sector Reform on the Poor: A Review of Issues and the Literature’ Energy Sector Management Assistance Program Paper No. 002 (Washington: World Bank).

    • Search Google Scholar
    • Export Citation

    Al-FarisAbdul-Razak1997“Energy Pricing Policies in Arab Countries: Impacts of Structural Adjustment Programmes,”OPEC Review Vol. 21 (December) pp. 24560.

    • Search Google Scholar
    • Export Citation

    Asia Pacific Economic Cooperation Forum World Bank Asian Development Bank Inter-American Development Bank and IMF2001“Social Safety Nets in Response to Crisis: Lessons and Guidelines from Asia and Latin America” (unpublished;Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    AtkinsonA.B. and J.E.Stiglitz1976“The Design of Tax Structure: Direct Versus Indirect Taxation,”Journal of Public Economics Vol. 6 No. 1-2 pp. 5575.

    • Search Google Scholar
    • Export Citation

    BaconRobert2001“Petroleum Taxes: Trends in Fuel Taxes (and Subsidies) and the Implications” Viewpoint Note No. 240 (Washington: World Bank).

    • Search Google Scholar
    • Export Citation

    BarnesDouglas F and JonathanHalpern2000“The Role of Energy Subsidies”inEnergy and Development Report: Energy Services for the World’s Poored.byPenelope J.Brooke and SuzanneSmith (Washington: World Bank Energy Sector Management Assistance Program).

    • Search Google Scholar
    • Export Citation

    ChuKe-young and others 1995Unproductive Public Expenditures: A Pragmatic Approach to Policy Analysis IMF Pamphlet Series No. 48 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    DinmoreGuy2002“Iran’s Oil Crisis Threatens to Drain the Country’s Lifeline” Financial TimesMay7.

    “Economic Theory African Reality”2002EconomistJanuary24.

    “Ecuador Battles Over Cooking Gas”2001EconomistFebruary8.

    Energy Information AdministrationInternational Energy Annual (Washington: various years). Available via the Internet: http://www.eia.doe.gov/emeu/iea/contents.html.

    • Search Google Scholar
    • Export Citation

    Energy Information Administration2000Annual Energy Review (Washington). Available via the Internet:http://www.eia.doe.gov/emeu/iea/contents.html.

    • Search Google Scholar
    • Export Citation

    FreundCaroline L. and Christine I.Wallich1995“Raising Household Energy Prices in Poland: Who Gains? Who Loses?,”World Bank Policy Research Working Paper No. 1495 (Washington: World Bank).

    • Search Google Scholar
    • Export Citation

    GolombekRolfCathrineHagem and MichaelHoel1995“Efficient Incomplete International Climate Agreements,”Resource and Energy Economics Vol. 17 pp. 2546.

    • Search Google Scholar
    • Export Citation

    GrahamCarol1994Safety Nets Politics and the Poor: Transitions to Market Economies (Washington: Brookings Institution).

    GuptaSanjeev and WalterMahler1995“Taxation of Petroleum Products: Theory and Empirical Experience,”Energy Economics Vol. 17 (April) pp. 10116.

    • Search Google Scholar
    • Export Citation

    GuptaSanjeev and others 2000Equity and Efficiency in the Reform of Price Subsidies: A Guide for Policymakers (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    GuptaSanjeevMarijnVerhoeven and ErwinTiongson2001“Public Spending on Health Care and the Poor,”IMF Working Paper 01/127 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    GürerNadir and JanBan2000“The Economic Cost of Low Domestic Product Prices in OPEC Member Countries,”OPEC Review Vol. 24 (June) pp. 14383.

    • Search Google Scholar
    • Export Citation

    HarbergerArnold1964“Taxation Resource Allocation and Welfare”inThe Role of Direct and Indirect Taxes in the Federal Revenue System: A Conference Report of the National Bureau of Economic Research and the Brookings Institution (Princeton, New Jersey: Princeton University Press).

    • Search Google Scholar
    • Export Citation

    HopeEinar and BalbirSingh1995“Energy Price Increases in Developing Countries: Case Studies of Malaysia, Indonesia, Ghana, Zimbabwe, Colombia, and Turkey,”World Bank Policy Research Working Paper No. 1442 (Washington: World Bank).

    • Search Google Scholar
    • Export Citation

    HwangM.J. and C.W.Yang2001“Unstable Price Elasticity and High World Oil Prices” paper presented at the 52nd International Atlantic Economic Conference Philadelphia PennsylvaniaOctober11-142001.

    • Search Google Scholar
    • Export Citation

    International Energy Agency1999World Energy Outlook: 1999 Insights: Looking at Energy Subsidies: Getting the Prices Right (Paris). Available via the Internet at http://www.iea.org/statist/index.htm.

    • Search Google Scholar
    • Export Citation

    International Energy Agency2001aEnergy Statistics of Non-OECD Countries 1997-98 (Paris).

    International Energy Agency2001bMonthly Price Statistics November 2001 (Paris). Available via the Internet at http://www.iea.org/statist/index.htm.

    • Search Google Scholar
    • Export Citation

    International Monetary FundInternational Financial Statistics (Washingtonvarious years).

    LampiettiJulian A. and others 2001“Utility Pricing and the Poor: Lessons from Armenia,”World Bank Technical Paper No. 497 (Washington: World Bank).

    • Search Google Scholar
    • Export Citation

    LoveiLaszlo and AlastairMcKechnie2000“The Costs of Corruption for the Poor”inEnergy and Development Report: Energy Services for the World’s Poored. byPenelope J.Brooke and SuzanneSmith(Washington: World Bank Energy Sector Management Assistance Program).

    • Search Google Scholar
    • Export Citation

    MackenzieGeorge A. and PeterStella1996Quasi-Fiscal Operations of Public Financial Institutions IMF Occasional Paper No. 142 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    MetschiesGerhard P.1999Fuel Prices and Taxation (Eschborn: Deutsche Gesellschaft fur Technische Zusammenarbeit).

    OfferdalErik and others 2000Ecuador—Selected Issues and Statistical Appendix IMF Staff Country Report 00/125 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    Organization of Petroleum Exporting Countries2001OPEC Annual Statistical Bulletin 2000 (Vienna).

    ParryIan2001“Are Gasoline Taxes in Britain Too High?,”Challenge Vol. 44 (July) pp. 6781.

    PricewaterhouseCoopersCorporate Taxes: Worldwide Summaries (New York: John Wiley and Sons, various years).

    “Put Your House in Order”1998EconomistFebruary5U.S. Edition, Survey.

    RajkumarAndrew Sunil1996“A Study of Energy Subsidies”Environment Department Background Technical Paper (unpublished;Washington: World Bank).

    • Search Google Scholar
    • Export Citation

    RamseyFrank P.1927“A Contribution to the Theory of Taxation,”Economic Journal Vol. 37 No. 145 (March) pp. 4761.

    SaboohiY2001“An Evaluation of the Impact of Reducing Energy Subsidies on Living Expenses of Households,”Energy Policy Vol. 29 No. 3 (February) pp. 24552.

    • Search Google Scholar
    • Export Citation

    SchwartzGerd and BenedictClements1999“Government Subsidies,”Journal of Economic Surveys Vol. 13 No. 2 (April) pp. 11947.

    • Search Google Scholar
    • Export Citation

    United Nations Development Program (UNDP) World Bank1994Ecuador: Energy Pricing Poverty and Social Mitigation Report No. 12831-EC (Washington).

    • Search Google Scholar
    • Export Citation

    Wakeman-LinnJohn and others 2002Azerbaijan Republic: Selected Issues and Statistical Appendix,IMF Staff Country Report 02/41 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    WhiteHalbert1980“A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity,”Econometrica Vol. 48 (May) pp. 81738.

    • Search Google Scholar
    • Export Citation

    World Bank1995“Venezuela: Efficiency Repricing of Energy” Sector Report No. 13581 (Washington).

    World Bank1997“Expanding the Measure of Wealth: Indicators of Environmentally Sustainable Development” Environmentally Sustainable Development Studies and Monograph Series No. 17046 (Washington).

    • Search Google Scholar
    • Export Citation

    World Bank2000a“Ecuador: Crisis Poverty and Social Services” Sector Report No. 19920 (Washington).

    World Bank2000bMaintaining Utility Services for the Poor: Policies and Practices in Central and Eastern Europe and the Former Soviet Union (Washington).

    • Search Google Scholar
    • Export Citation

    World Bank2001World Development Indicators (Washington).

    World Bank Energy Sector Management Assistance Program1994a“Ecuador: Energy Pricing Subsidies and Interfuel Substitution” ESMAP Report No. 11798 (Washington).

    • Search Google Scholar
    • Export Citation

    World Bank Energy Sector Management Assistance Program1994b“Ecuador: Energy Pricing Poverty and Social Mitigation” ESMAP Report No. 12831 (Washington).

    • Search Google Scholar
    • Export Citation
1We are grateful for helpful comments frem Muhammad Al-Jasser, Ulrich Bartsch, Philippe Callier, Juan Pablo Cordoba, Ramon Espinasa, Mangal Goswami, Eliot Kalter, Vladimir Klyuev, Jean Le Dem, Edouard Maciejewski, Edouard Martin, Melhem F. Mel-hem, Rakia Moalla-Fetini, Bright Okogu, Stephen Schwartz, Nicola Spatafora, Siddarth Tiwari, Bert van Selm, John Wakeman-Linn, and participants in the IMF Conference on Fiscal Policy Formulation and Implementation in Oil-Producing Countries, June 5-6, 2002. Shamit Chakravarti and Erwin Tiongson provided invaluable research assistance. The usual disclaimer applies.
2For a general discussion of quasi-fiscal activities, see Mackenzie and Stella (1996). For a discussion of quasi-fiscal activities in the energy sector specifically, see Petri, Taube, and Tsyvinski, Chapter 18 in this volume.
3Data refer to November 2001 Monthly Price Statistics (International Energy Agency, 2001b). As a comparison, the tax-exclusive price of gasoline in the United States was US$0.21 per liter.
4Energy Information Administration (various years).
5This includes the cost of transporting the product from the world wholesale market to the importing country.
6Because some goods, notably leisure, are often excluded from consumption taxes, some distortion of relative prices is still likely to occur, especially to the degree that goods are complements to, or substitutes for, the excluded goods.
7This may be relevant, for example, in Ecuador in 2000. In this case, the domestic price was measured in January 2000 but prices were adjusted significantly later in the year. Thus, the large subsidy estimate for Ecuador shown in Table 15.2 represents the annualized subsidy in January 2000, which may be different from the actual subsidy for 2000 as a whole.
81999 is the latest available year for most countries; the world oil price in that year was also near the average for recent years (US$18 per barrel).
9Much of the subsidy here does not appear in government expenditure—only in the case that an explicit government subsidy was provided to an oil producer would these outlays be reflected in total government spending.
10Energy Information Administration (various years).
11A similar result holds in an intertemporal context where the first-order condition for optimization requires that the marginal revenue from exporting today be equal to the discounted marginal revenue from exporting tomorrow. Thus, the domestic price would be set equal to the marginal revenue from exports in each time period.
12It has been shown that, under some conditions, it is most efficient to tax all commodities at the same rate (e.g., Atkinson and Stiglitz, 1976). A more general rule would be to tax goods so that the percentage tax-induced change in the quantity demanded (measured along the compensated demand curve) is the same for each taxed good (Ramsey, 1927).
13The distribution and marketing costs of domestic consumption are “saved,” since these costs are not incurred if petroleum is exported instead of consumed domestically.
14The argument for creating a wedge between export and domestic prices is essentially identical to the classic terms-of-trade rationale for protectionism. This argument notes that trade tariffs may benefit a country if the changes in world supply or demand induced by the tariff positively affect the country’s terms of trade.
15Golembek, Hagem, and Hoel (1994) estimate price demand elasticities of around -0.9 for OECD countries and -0.75 for non-OECD countries. Hwang and Yang (2001) find somewhat lower elasticities (around -0.25) in their study of U.S. data.
16In other words, equation (4) is evaluated at the point at which D -T = .75W.
17This number represents an average of distribution, marketing, and international transport costs in OECD countries observed over a period of time. If these costs were lower, the marginal revenue estimates would be somewhat lower. For example, if these costs were 50 percent of the wholesale price, then marginal revenue for Saudi Arabia would be 86 percent (rather than 88 percent) of the small exporter case when n = -0.50.
18Gupta and Mahler (1995) survey the reasons behind petroleum taxation policies.
19Domestic oil consumption would also fall in oil-exporting countries if prices were increased, leading to pressures to increase exports. Given the small share of world oil consumption accounted for by these countries, the impact on world prices would be small.
20Note that the deadweight loss in Table 15.4 is in some cases larger than the subsidy in Table 15.2. This is because the deadweight loss also takes into account the externality costs (from pollution and congestion), whereas the subsidies in Table 15.2 are calculated relative to the “free-market price.”
22Oil is a major input for electricity production.
23To the extent that subsidies are implicit, removing them may not automatically increase government revenue. Changes in tax policy may be required to capture higher profits of domestic oil-supplying or -producing enterprises. In addition, corruption in state oil companies (SOCs) may result in the government’s effectively having less than a 100 percent claim on the SOCs profit, thus reducing the fiscal benefits of subsidy reform (see McPherson, Chapter 7 in this volume).
24For a related discussion of unproductive spending, see Chu and others (1995).
25See Gupta, Verhoeven, and Tiongson (2001) for an examination of the impact of public spending on health care and the health status of the poor.
26See Federico, Daniel, and Bingham, Chapter 17 in this volume, and Daniel, Chapter 14 in this volume, for a more extensive discussion.
28In this respect, increased transparency in the management of public finances is critical for gaining public support for higher petroleum prices. It would demonstrate that resources realized from reducing or eliminating implicit price subsidies would be used for the benefit of the population in need.
29See PSIA concept note (http://www.imf.org/external/np/exr/facts/sia.htm) and the recent Poverty Reduction and Growth Facility (PRGF) review (http://www.imf.org/External/NP/prgf/2002/031502.htm) for an assessment of recent experience on PSIA.
30As another example, an ex post social impact analysis for an oil-consuming country—Armenia—showed that reforms in the energy sector had led to a more reliable electricity supply, improved the financial viability of energy companies, reduced cross-subsidization, and improved payment discipline. However, the poor cut consumption more (relative to the nonpoor), partly because the elimination of the increasing block tariff raised the average price of electricity more than expected (Lampietti and others, 2001).
31For most countries, prices are for premium gasoline; for a few, however, prices are reported for regular gasoline instead.
32LPG includes primarily residential propane or a mixture of propane and butane.
33Heavy fuel is also sometimes referred to as residual fuel oil.
34The website is http://www.eia.doe.gov/emeu/world/mainl.html. This same information is published in the International Energy Annual (Energy Information Administration, various years).
35See footnote 5.
36See footnote 6.
37The eight countries are Canada, France, Germany, Italy, Japan, Spain, the United Kingdom, and the United States. Data are missing for Canada for light and heavy fuel oil and for the United States for heavy fuel oil.
38Note that insurance costs are not taken into account, which could lead to an overes-timation of the subsidy. However, these costs are likely to be very small and would not substantially change the results.
39Data taken from the table on page 11.11.

    Other Resources Citing This Publication