All governments either tax or subsidize the consumption of fossil fuels, but many of these taxes and subsidies are difficult to observe or infer. Coal taxes and subsidies, for example, are exceptionally hard to determine since in many countries coal is sold directly to government-owned utilities through long-term contracts rather than through retail markets; any subsidies may be hidden in the price of electricity11. It is also hard to determine the supply cost for coal, since there is no single international reference price. Without either the supply cost or the retail price, taxes and subsidies are very difficult to measure.

Our study focuses on gasoline taxes and subsidies, which can be more readily measured: gasoline is sold directly to consumers in all countries, which gives it an observable retail price; country-to-country differences in gasoline quality are relatively small; and there is effectively a single world reference price. Until now, data on gasoline prices have been limited to a single price every two years for most countries, which is too infrequent to observe many policy changes13.

We collected data on local gasoline retail prices for 157 countries from January 2003 to June 2015 using both primary and secondary sources. We included all sovereign states whose populations were greater than one million in 2012, except for four countries for which we could not locate reliable data: Cuba, Eritrea, North Korea and Turkmenistan. For 46 countries our coverage is temporally incomplete for the full time series, with data missing for 1,067 (4.5%) of the 23,550 country-months. A full list of countries, along with the number of country-month observations for each case, is found in Supplementary Table 1.

Primary documents were obtained from national governments and found on websites maintained by ministries of finance, commerce, trade, communications, transportation, natural resources, petroleum, energy, and mining; central statistical agencies; central banks; energy regulatory bodies; state-owned oil companies, including oil refining companies; and official government decrees and public announcements. In 17 countries we employed local researchers to obtain primary data that were not otherwise accessible.

Secondary sources included the European Commission, the International Road Transport Union, the Food and Agriculture Organization, the Famine Early Warning System, CITAC Africa, the Organization of Arab Petroleum Exporting Countries, and documents from the IMF and World Bank. Whenever possible we cross-validate price data across multiple sources to alleviate concerns about measurement bias.

To measure implicit taxes and subsidies we use the price gap method, which compares the observed retail price in each country with a global benchmark price14. This method yields a single figure that represents the net per-unit value of all taxes and subsidies, both implicit and explicit, and that can be readily compared across countries and over time. Below we refer to these implicitly measured net taxes or subsidies as ‘net taxes’ or ‘net subsidies.’

The price gap method allows us to measure what the IMF calls ‘pre-tax subsidies’, which represent the difference between the retail price and the international supply cost; we do not attempt to estimate ‘post-tax subsidies’, which are defined as the difference between the retail price and the sum of the international supply cost, a basic consumption tax, and a Pigouvian tax that offsets the costs of local pollution, congestion, and carbon emissions4. Hence, the price gap should be interpreted as a lower-bound estimate of total fossil fuel subsidies.