The World Bank publishes a useful time series entitled “oil rents.” It is a variable defined as the difference between the cost of extracting oil and the value of oil production. It’s utility is obvious. The variable gives a measure of how much value there is to fight over along the value chain in the country — between governments, upstream producers, midstream transporters and (if applicable) downstream refiners. You can use it to test all sorts of hypotheses about oil politics and the industrial organization of the sector. All there, right on the World Bank website.
Only not. I checked the metadata, which explains how the variable was constructed. It sent me to a (fascinating!) article by Michael Ross of UCLA on “Oil, Islam, and Women.” There I went to the equally fascinating “Manual for Calculating Adjusted Net Savings” by Katharine Bolt, Mampite Matete and Michael Clemens.
Which is where I learned that oil rents by the World Bank are not the data you are looking for.
The definition of oil rents, remember, is revenue minus cost. Revenue is more complicated than you might think, since oil prices vary according to the quality of the crude and the difficulty of getting it to a port. Still, the price of oil can be estimated well enough. The problem is cost.
For most countries, the World Bank cost data come from one year: 1993. Only the costs for Latin American countries come from another year: 1981. The oil rent data assume that these real costs over time are constant, which, to put it mildly, bears no resemblance to reality. It is a particular problem for countries that produce heavy oil, such as Canada and Venezuela; or countries that have invested heavily in secondary recovery of declining fields, like Norway and Mexico; or countries that have discovered new technology to develop previously inaccessible reserves, like America and Brazil.
Moreover, direct cost data are only available for Ukraine, Russia, Venezuela, Mexico, Libya, Malaysia, Nigeria, USA, Gabon, Egypt, the United Kingdom, Norway (a 1987 study), Ecuador, Peru, Trinidad & Tobago, Argentina, Bolivia, Brazil, Chile, Colombia Iran, Iraq, Saudi Arabia, Kuwait, United Arab Emirates, Oman, Indonesia and Canada.
Other countries are assigned values given geographic proximity and a similar ratio of offshore to onshore production — which is even crazier than assuming that real production costs did not change between 1980 and 2012. Brunei’s production costs are not Indonesia’s, for example, and Chad bears no relation to Gabon. God only knows how they estimated costs in Central Asia.
In short, what seems like an awesome data set turns out to be not so useful once you dig through the metadata. The number simply moves in parallel with total oil revenue: there is no new of variation.
Buyer beware. Sometimes you gotta get a broom and clean up the numbers yourself.