Let’s imagine a country with a large resource deposit. In this country, the government grants a private company license to extract it. In a given year of production, the company extracts USD 1 billion worth of resources and then pays a 30% share—USD 300 million—through several channels of taxation to the host government.
Citizens who observe similar numbers in company, government and watchdog reports are left wondering whether such a ratio is high or low—essentially, if it represents a good or bad deal for companies or citizens.
This is a question we are asked a lot at NRGI. Our answer can be frustrating: it depends. Without information on the detailed specifics of a project; tax payments and revenues over its full lifecycle; and the conditions under which the full contract was signed; it is very hard to determine whether a deal is “good” or “bad.”
In many countries, companies only pay royalties of between 3 and 10 percent of the value of resources in the first few years of a project until they recover sunk costs. Later in the project, additional levies, such as corporate tax, taxes on dividends, or additional profit taxes may be applicable, often leading to a government share of over 50 percent.
Typically, pre-tax profits are much higher on oil than on gas or most minerals; the government can therefore secure a larger share of those profits in oil projects while companies still get their expected returns. Projects in “frontier regions” (i.e. natural gas in east Africa) might involve more risk and typically bring a lower share of revenues to government. A particular government may also deliberately choose to take bigger risks (i.e. by purchasing a stake in the project), or prefer revenues earlier than later, or negotiate non-fiscal benefits in lieu of taxes. All of these choices affect the average share it gets.
The good news is that more contracts are being published and data on production volumes, prices, revenues and even some costs can now be found across scattered reports. The IMF has made its state-of-the-art financial model available to the public. Organizations like NRGI are using these and available open data to monitor extractive projects. The key remaining challenge? Assembling and reviewing all the information to populate a model is time-consuming and difficult to do at scale.
But a deluge of project-level payment information based on EU mandatory disclosure rules is changing the landscape. Reporting by EU- and U.S.-registered and listed companies could touch on thousands of projects every year. When analyzing similar observations across thousands of projects, one can adopt radically different approaches than those used when looking at one project. One may have much more data gap, but we can also look for patterns in the data.
In Angola, the Ministry of Finance has been providing very granular information on revenues from its oil sector block by block, year by year total value of oil production and total government revenue. (While the data is exemplary in its detail, there are also limitations to the government’s transparency, as discussed in this study.) Using this source, I constructed a dataset consisting of 90 observations (seven years and 15 blocks, where not all blocks saw production over the years). This dataset uses very few datapoints compared to what we expect to be reported in mandatory company disclosures, but it is more homogenous, as it all relates to the same commodity (oil) and same country and resource basin (offshore Angola).
First, I made a plot to display the strong correlation between total value of production and the amount of taxes paid on each block (counting every year as separate observation). The relationship between the two variables is not surprising, but already one can see datapoints diverging from the trend.
On average, government revenues represent 44 percent of total production value. When looking at the distribution of the share of government revenue across all blocks and years, one can see that most are within a 20 percent to 40 percent and 40 percent to 60 percent range. We also see how outliers are distributed.
Next, I looked at which blocks are generating most revenues proportionately. In the oil industry, there are often quite big returns to scale. Massive oil discoveries can attract higher profit margins and can potentially be taxed proportionally more. In order to conduct this analysis, the sample was restricted to the 10 offshore blocks where there was constant production between 2009 and 2015. The remaining five blocks were either onshore (with special contractual terms and project economics) or just starting or terminating production and behaved as outliers: it is not surprising to see large swings in tax revenues across years, as big development and decommissioning costs are offset against taxes.
For these remaining 10 blocks, one can see a very strong association between average production and share of government revenues. The two blocks with large outputs (over 400,000 boe/day), Block 17 and Block 15, had government shares of total revenues of 55 percent and 67 percent, respectively. Blocks with output between 100,000-200,000 boe/day had government shares of revenue between 40 percent and 55 percent. The blocks with smallest production had some of the smallest government share of revenues.
How government shares of revenues change across years is an additional point of interest: the share of government revenue has followed the change in oil prices. The share of government revenue was below 50 percent when oil prices were lower in 2009, 2014 and 2015, but increased to above 60 percent when oil prices were high in 2011. In fact, further analysis shows that for every percent in oil price increase, the government of Angola saw a 1.23 percent increase in oil revenues on average.
These are just some initial baby steps in the types of analysis that could be conducted using project by project payment information once the data becomes more broadly available. One could look at the effect of many more important variables to explain the variance in government share of revenue: whether the production is ramping up, near plateau or declining (affecting cost recovery), what the water depth for each project is (affecting cost) and when the deal was struck (affecting government bargaining potential). Further calibration would also be required to take into account differences between contract types based on risk-reward profile or to account for potential non-financial benefits, such as oil-backed infrastructure. With more systematic data collection such as that undertaken on www.resourceprojects.org, we can start to get a better understanding of the drivers behind oil revenues to government.
By adding more and more contextual information to the analysis, we would anticipate explaining an increasing share of the variance. This also means that we can better identify outliers, which may deserve more scrutiny. Why is total revenue from one project systematically lower than a comparable one? Are payments reported by large oil majors systematically different to those of smaller companies across joint ventures? Do payment patterns differ significantly for projects tied to a particular beneficial owner, to a national oil company or when linked to third-party oil sales deals?
While such analysis will not provide definitive answers on whether deals are good or bad, it will help us better understand the revenue-generating potential of different projects. It might also raise flags on issues that deserve a closer look. This could increase accountability in the resource sector and improve decisionmaking by governments that might not be aware of the financial risks and costs associated with certain licensing practices.
With the extractive data deluge expected this year, mining oil and minerals data will only become more interesting.