Public interest models: a powerful tool for the advocacy agenda
Open financial models can clearly put analysis into a genuinely independent public space, and also trigger a rise in public understanding which could enrich the governance debate in many countries.
But there is a third function public models can serve: that of advocacy for targeted disclosure of information.
The stress here is on “targeted”. A lot of transparency debates are generic – the need to disclose data as a matter of principle.
It is striking that as the transparency agenda has advanced, and won many battles, so has a debate about whether it is contributing to an increase in accountability. As Paul Collier said: “transparency has to lead to accountability otherwise we’re just ticking loads of boxes”.
We need all these campaigns to continue, and we need to pursue maximum disclosure. Because while transparency does not guarantee accountability, it is its essential prerequisite. Necessary but not sufficient.
But here’s where modeling can help to provide some examples of how data can be used, in a very specific way, to advance accountability.
Let’s take the example of an oil project in Africa. A financial model has to deal with uncertainty and so provides three scenarios for future production and prices, which all have a radical impact on the revenues the government could expect to see. That’s unavoidable. Under the “God, Exxon and everyone else” principle, future price and to some extent production are hard to foresee.
But then there is a second layer of uncertainty caused specifically by the model having to use public domain data. The company, and the government if it exercised its rights of access to information, does not face this second layer because it has access to real data, whereas the public interest model must use estimates and extrapulations. These can be justified, written out and explained – they can be well-informed guesses, in other words, and in the blog on the analytical power of public models, we argue that you can still arrive at useful analysis and conclusions despite this handicap.
Nevertheless, they are guesses. And unlike the first layer of uncertainty, relating to future prices and the ever-changing global market, this second layer can be directly addressed by information the government already has to hand – or could get under its contractual right of access to information.
In the case of this African project, the estimate for how much the government will get would vary over the lifetime of the project by between $2.3 billion and $3 billion – even using exactly the same price and production scenario. This gap of $700 million could be closed by the provision of three simple sets of information. In this case:
- The historic prices recorded from the field calculated using the formula in the contract. This would resolve a $170 million uncertainty gap. In the absence of hard data, the model has to use the crude grade’s API quality and two-year-old investor documents estimating a discount to the Brent benchmark.
- Details of financing costs approved by the government. The contract simply states the Contractor can pass on finance costs at a reasonable market rate without specifying what percentage of its costs it would seek to borrow against, and at what rate. A reasonable guess of the upper and lower parameters of this gives another $250 million range of uncertainty in revenues to the state, a gap which the government could close simply by stating what terms of finance used in the project are.
- Transport costs. The oil from this field has to pass through a local loop pipeline that was built especially for this project to a trunk pipeline to export which was already built. A reasonable guess based on investor documents and analogous data gives costs of about $9.50 per barrel. But if this was wrong either way by just $2 – say it was $7.50 or $11.50 – this could have another $300 million impact on government finances. The contract states a separate transport protocol will be signed and lays out some general principles to be included in it. So if the government published the protocol, and the actual transportation costs charged, this uncertainty could also be cleared.
Three fairly simple sets of information can close a three quarters of a billion prediction gap in a country whose annual government budget is only about five times that.
Public interest models can add a lot of power to transparency demands because they can demonstrate, quite precisely and ahead of time, why the information asked for is important and how it will be used. Think of it as keyhole transparency surgery.
But beyond individual data points, it will also become clear over time and repeated use that there are entire classes of information missing from the transparency agenda which are needed to build a full picture.
Financing costs allowed to producers are key to what a country sees from its extractive project at the end of the day. There has been little focus yet on this area. But I believe many public interest models will show the same uncertainty and significant impact on state revenues depending on what financing terms have been agreed, and it will emerge then as a key transparency demand – supported by empirical data (or the lack of it) from many real world cases.
Trading too is largely information dark. EITI has begun to pay attention to it, and there are plenty of examples which show how trading is a huge area for rent-seeking and cronyism in many countries. Global Witness’s recent study of SOCAR’s arrangements to trade Azerbaijani oil is a case in point. Modeling can create demand for the historic prices recorded field by field, and used in formulae to calculate direct state revenues such as the base on which royalties and profit splits are calculated.
Above all, accurate and independent reserves estimates would reduce unnecessary uncertainty around revenue flows and share of the profits. In a year or two, with models popping up all over the place showing large ranges of uncertainty caused by this fact, a campaign to get all state-owned natural resource reserves regularly and independently certified would be practical – and backed by empirical data.
Finally, the beauty of the advocacy function of public interest models is that it serves as a perfect foil to the analytical function. To the extent that the models have margins of uncertainty caused by having to rely on the partial and dated information in the public domain, they become proof of the need for more comprehensive and timely provision of all relevant data sets.