NOTE NUMBER 240 P U B L I C P O L I C Y F O R T H E privatesector SEPTEMBER 2001 Petroleum Taxes Robert Bacon Robert Bacon is manager Trends in Fuel Taxes (and Subsidies) and the Implications of the Oil and Gas Policy Division at the World Recent World Bank analysis shows that taxes on petroleum products are Bank. He works on a critical source of government revenue for low-income countries. And if energy sector reform and the links between reform the experience of industrial countries ser ves as any guide, the rates of T H E W O R L D B A N K G R O U P PRIVATE SECTOR AND INFRASTRUCTURE NETWORK strategy and poverty these taxes will have to rise sharply as low-income economies develop. alleviation; the impact of sector reform on the But policymakers must be mindful of how taxes (or subsidies) affect the macroeconomy; and oil relative prices of fuels, since too large a difference in prices between product taxation and subsidies. He has recently products can lead to fuel switching and adulteration, adversely affecting worked on Bank programs the government tax take and pollution levels. in the Islamic Republic of Iran, Mexico, and In 2000 the price of crude oil rose to levels not How important are fuel taxes? Central Asia. seen for more than 20 years, and as a result so Calculating total government revenue from all did the price of oil products. The price hikes taxes on petroleum products in a country prompted public debate about the level of taxes requires extensive data, and even for OECD determining the final prices paid by consumers. countries separate figures on such revenue are By one estimate, taxes accounted for 68 percent generally unavailable. But a 1990–91 Inter- of the final price in the European Union in 2000 national Monetary Fund survey of 30 develop- (OPEC 2000). But what about developing coun- ing and industrial countries does provide data tries, where petroleum product taxes are often on tax revenues from petroleum products, and an important source of government revenue? In thus allows an analysis of their relationship to particular: the government budget and GNP (Gupta and ▪ How important are petroleum product taxes Mahler 1994). Statistical analysis drawing on for developing countries? these data reveals the following relationships: ▪ What principles should govern the level and ▪ The consumption of petroleum products structure of taxes on oil products? rises almost as fast as GNP (elasticity of 0.8). ▪ Is the general level of oil product taxes in ▪ Total government tax revenue from all sources developing countries similar to that in indus- tends to rise faster than GNP (elasticity of 1.2). trial countries? ▪ Tax revenue from petroleum products rises ▪ What is the pattern of relative tax rates on dif- more slowly than total tax revenue (elasticity ferent products? of 0.7). P E T R O L E U M T A X E S TRENDS IN FUEL TAXES (AND SUBSIDIES) AND THE IMPLICATIONS ▪ Thus tax revenue from petroleum products consideration, goods accounting for a larger rises slightly more slowly than GNP (elastic- share of budgets for the rich than for the poor ity of 0.9). should be taxed more heavily. Goods that pro- Regression analysis indicates that at a GNP duce large negative externalities (such as emis- per capita of US$316, predicted total tax rev- sions from automotive fuel use or congestion enue per capita was around US$40, while that from excessive road use) should also be taxed at from petroleum products was US$8 (20 per- high rates, to discourage their consumption cent). At a GNP per capita of US$10,000, pre- and reduce social harm. dicted total tax revenue per capita was For fuels, these principles do not always 2 US$2,500, and fuel tax revenue per capita was work in the same direction, so the relative US$126 (5 percent). importance of each principle needs to be eval- These results illustrate the importance of uated in each case. For example, emissions downstream taxation of petroleum products, from diesel fuel are more harmful than those especially for low-income countries. The reason from gasoline, but encouraging the use of is that taxing fuel is one of the easiest ways to get diesel-powered mass transit may be desirable as revenue: collecting fuel taxes is relatively a way of relieving congestion. Kerosene is par- straightforward, and the consumption of fuels ticularly problematic in developing countries, as a group is relatively price inelastic and since it can be used to adulterate both gasoline income elastic, ensuring buoyant revenue as and diesel. As a result, setting lower taxes on income rises and tax rates are increased. So kerosene (to reduce the cost of lighting and while petroleum product taxes are important in cooking fuels for the poor) can erode the total developed countries, they may be even more tax collected (necessitating an increase in the important in developing countries. general tax level to produce a given revenue). And where kerosene replaces gasoline, lower What makes good tax policy? taxes lead to higher emissions and worse vehi- Tax theory suggests that where the govern- cle performance. Higher taxes on kerosene can ment’s sole aim is to raise revenue for public hurt poor households, however, which tend to expenditure, goods for which demand is least spend a larger share of their budgets on this sensitive to price increases should tend to bear fuel than do better-off households. But this also the highest tax rates. Goods that are close sub- means that if governments wish to offset the stitutes should be taxed at similar rates to pre- effect of higher kerosene taxes on poor house- vent demand from switching from the higher- to holds, they can do so through targeted assis- lower-taxed good, reducing government rev- tance rather than across-the-board kerosene enue. In addition, where equity is an important subsidies. Table Average petroleum product taxes and prices in OECD and non-OECD countries, 1999 1 Fuel and country group Tax as share of final price (percent) Tax (U.S. cents per liter) Net price (U.S. cents per liter) Gross price (U.S. cents per liter) Gasoline OECD countries 67 58.1 25.2 83.3 Non-OECD countries 44 22.9 26.9 49.7 Automotive diesel OECD countries 59 42.4 25.0 67.4 Non-OECD countries 40 16.6 22.8 39.4 Kerosene Non-OECD countries 23 5.1 12.3 17.4 Note: The table shows data for 22 OECD countries (excluding the Czech Republic, Hungary, Mexico, Poland, and Turkey) and 37 non-OECD (mostly developing) countries. Source: For OECD countries, IEA (2000); for non-OECD countries, World Bank staff estimates. Figure Gasoline tax rate and GDP per capita in selected OECD and non-OECD countries, 1999 1 Tax per liter (U.S. dollars) 1.00 Oil producer Nonproducer 0.80 3 0.60 0.40 0.20 0 0 10,000 20,000 30,000 40,000 50,000 GDP per capita (U.S. dollars) Note: This figure covers the same countries as table 1. Source: Author’s analysis based on World Bank data. How do tax rates compare? the higher-income countries two and a half There are few published data on petroleum prod- times that in developing countries for both gaso- uct taxes for developing countries, so World Bank line and diesel. This difference is reflected in the estimates for 1999 were collated.1 Collecting higher gross prices in OECD countries. direct data on the full range of petroleum prod- The net prices for gasoline and diesel are uct subsidies, used in many developing countries similar, reflecting the approximately equal pro- at some point in the value chain, would have been duction costs for the two fuels. But the average a substantial task and so was not attempted. But tax per liter is 40 percent higher for gasoline prices net of tax are available, and in the absence than for diesel in both groups of countries, of subsidies these could be expected to be fairly resulting in substantially higher gross prices for similar across countries, since petroleum prod- gasoline. The difference in taxes might encour- ucts are highly tradable. Thus when a country’s age a switch to diesel in the long run through price net of tax is near the global average, the sub- the use of mass transit, but it will also tend to sidy element is likely to be small. reduce government revenue and thus require For gasoline and automotive diesel, the aver- an offsetting increase in general taxation. age net price in developing countries is fairly sim- Kerosene has a lower net price and much ilar to that in OECD countries (table 1). Since lower tax rate than the other fuels, producing OECD countries do not subsidize petroleum a gross price that is less than half that for diesel products, this suggests that the average net price and a third that for gasoline. Since kerosene in developing countries includes little subsidy. can replace up to 5 percent of the volume of By contrast, the difference in tax charged per gasoline and 20 percent of the volume of unit is very large, with the average tax per liter in diesel without consumers noticing, this large P E T R O L E U M T A X E S TRENDS IN FUEL TAXES (AND SUBSIDIES) AND THE IMPLICATIONS price difference can lead to a substantial diver- for government revenue. Nevertheless, even sion of kerosene in countries with poor sys- for high-income countries, fuel taxes are tems for monitoring and enforcing fuel among the most important commodity taxes. quality standards, with a resulting loss of tax revenue. viewpoint Conclusion Notes The data on petroleum product taxes in 1999 1. Metschies (1999) provides data for many coun- is an open forum to provide strong evidence that rates in developing tries on prices including taxes. An earlier paper analyzing encourage dissemination of countries as a group are much lower than those the share of taxes in final prices is Saito (1975). public policy innovations for in high-income countries, even though these private sector–led and taxes are a more important source of revenue References market-based solutions for for low-income countries. Oil producers, which IEA (International Energy Agency), 2000. Energy development. The views can obtain substantial revenue from upstream Prices and Taxes. Paris. published are those of the royalties, are likely to charge lower commodity Gupta, Sanjeev, and Walter Mahler. 1994. “Taxation authors and should not be taxes than other countries at the same level of of Petroleum Products: Theory and Empirical Evidence.” attributed to the World GNP and government expenditure. The results IMF Working Paper 94/32. International Monetary Fund, Bank or any other affiliated of a regression confirm these conclusions. Fiscal Affairs Department, Washington, D.C. organizations. Nor do any of The regression, which distinguishes oil pro- Metschies, G. P. 1999. Fuel Prices and Taxation. the conclusions represent ducers and nonproducers, finds a strong posi- Eschborn, Germany: Deutsche Gesellschaft für official policy of the World tive correlation between GDP per capita and the Technische Zusammenarbeit. Bank or of its Executive tax per liter of gasoline: a US$1,000 increase in OPEC (Organization of Petroleum Exporting Directors or the countries GDP per capita is associated with a 1.5 cent rise Countries). 2000. “Who Gets What from Imported Oil?” they represent. in the tax per liter (figure 1). An increment of Public Relations and Information Department. Vienna. 1 million barrels a day of crude oil production [http://www.opec.org/NewsInfo/WhoGetsWhat/ To order additional copies is associated with a 6 cent decline in the tax. For WhoGetsWhat.pdf]. contact Suzanne Smith, automotive diesel, a US$1,000 increase in GDP Saito, K. W. 1975. “An Examination of Changes in managing editor, Room I9-017, per capita is associated with a 1 cent rise in the the Retail Price and Taxation of Petroleum Products (July The World Bank, tax per liter, while domestic oil production has 1973–July 1974).” Studies in Domestic Finance, no. 9. 1818 H Street, NW, no significant effect. World Bank, Development Economics Department, Washington, DC 20433. A comparison of net gasoline and diesel Washington, D.C. prices across countries shows that the prices Telephone: are similar in virtually all countries, reflecting Robert Bacon (rbacon@worldbank.org). 001 202 458 7281 the near equality of production (plus trans- Fax: port) costs. By contrast, a comparison of gross 001 202 522 3181 prices shows that in virtually all countries the Email: gasoline price is much higher than the diesel ssmith7@worldbank.org price. This difference reflects a general ten- dency to encourage the use of diesel, although increasing diesel emissions in the megacities of Printed on recycled paper developing countries may result in a push toward equality in prices. The data also suggest that as incomes rise and the ability to finance public expenditure improves, countries will tend to increase auto- motive fuel taxes rapidly, though not quite proportionately because of their ability to broaden the tax base. Thus developing coun- tries can be characterized as moving to tax rates that are higher but of lower significance This Note is available online: www.worldbank.org/html/fpd/notes/