Open contracts, closed models

How Can Economic Modeling Improve Extractive Sector Governance?

The potential for economic modeling to improve extractive sector governance is beginning to get the attention it deserves. Economic models are already a key tool in “behind the scenes” decision making for companies and, increasingly, for countries as well. Existing modeling efforts however suffer from two serious shortcomings. First, the vast majority of models, even those developed with public sector funds, are confidential. Second, although the underlying logic of most models are comparable, custom-designs inhibit cumulative learning and routine updating. An important exception may be the IMF’s FARI model. OpenOil strongly agrees with the recent call by NRGI that the FARI model should be placed in the public domain. From our perspective, however, publishing FARI is mostly about public accountability – any model used to redesign fiscal regimes in developing countries should be open to public peer review. The complexity of the FARI model, in our view, makes it an unlikely starting point if the objective is to expand the constituency of users able to employ economic models to improve extractive sector governance.

Who is Building Economic Models?

Although economic modeling has not been a significant part of the debate on greater transparency in the extractive sector, these techniques are at the heart of behind-the-scenes decision-making for both companies and governments.

For companies, project economic models are routinely used for formulating negotiating positions and making investment decisions. These economic models (often called discounted cash flow models) allow companies to generate two core metrics such as “net present value” and “internal rate of return” in order to compare the value of one project against another. According to the a 2011 survey by the Society of Petroleum Evaluation Engineers, 89% of respondents used discounted cash flow models as their principal method for valuing projects.

Economic models are also widely used by governments seeking to more effectively manage the extractive sector. A good overview of the various uses for modeling from a government perspective can be found in the very useful new book on Administering Fiscal Regimes for Extractive Industries written for the IMF by UK tax expert Jack Calder (unfortunately not freely available). From this volume and the wider literature on economic models, five common uses can be identified.

1. Fiscal Regime Design and Revision: When designing fiscal regimes and establishing general contract terms, governments are engaged in a balancing act of attracting inward investment while at the same time maximizing revenue generation. Economic models are commonly used to assess the impact of proposed taxes under varying production, price and expense scenarios.

2. Support for Contract Negotiations: A similar but more specialized use of economic models is in direct support for contract negotiations. Companies normally come to the negotiating table with detailed models to bolster their negotiating positions. Increasingly, resource rich developing countries are developing their own economic models in an attempt to level the playing field.

3. Monitoring Incoming Revenues: Governments, particularly Ministries of Finance, develop economic models in order to evaluate incoming revenues in order to ensure that the results, taking into account real production volumes, market prices and actual company expenses, are consistent with the policy expectations when the fiscal terms were set.

4. Risk Assessment for Tax Audits: Revenue inflows are also assessed for the more specialized purpose of aiding risks assessment by tax authorities. Differences between model projects and actual revenue can highlight issues worthy of further investigation including, where appropriate, subsequent audits.

5. Budget Forecasting and Revenue Management: For many resource rich developing countries, extractive sector revenues constitute a major portion of overall revenue. Forward-looking economic models, based on hypothetical scenarios, are used to generate potential revenue forecasts to be used both in longer-term budget planning and also as a way to anticipate revenue management challenges.

The reasons for economic modeling listed above all assume a high degree of technical expertise. They build on extensive knowledge of the sector and the specific fiscal regime. We believe that economic models can also provide a good entry point for understanding the economic implications of contract terms that are in the public domain but not well understood. Thus we suggest an additional purpose:

6. Understand Economic Implications of Contract Terms: Contract monitoring workshops are often a first step in introducing new audiences to contract fiscal terms. This approach generates a bottom up perspective based on an analysis of individual fiscal instruments (i.e. signature bonuses, royalty rates, cost recovery limits, production-sharing splits, corporate income tax, and government participation). Economic models, when combined with a user-friendly dashboard, can provide a valuable top down perspective revealing the interrelationship between various fiscal instruments.

Building Off Existing Efforts?

An industry exists for developing extractive sector economic models. Take the petroleum sector for example. Models are developed by consultants (e.g. Daniel Johnston or Pedro Van Meurs) and larger firms (e.g. Wood Mackenzie or IHS). Other companies including Palantir, Ceasar, and Petrocash sell generic models that clients can develop and adapt to their specific circumstances. Resource rich developing countries often draw directly on this kind of expertise, often with funding through organizations like the World Bank.

One modeling effort of particular interest to resource rich developing countries is the FARI model developed by the Fiscal Affairs Department at the IMF. The model was initially developed to assist in fiscal regime design and allow for comparisons among contracts and countries. It has since been expanded to assess the full economic impact of projects. FARI is unusual in that it was designed from the outset to accommodate both mining and petroleum sectors.

The broad range of actors engaged in economic modeling of extractive sector projects might suggest that the tool can be easily adapted to serve the wider interest of extractive sector good governance. Unfortunately there are two recurring challenges with existing modeling efforts: they are normally confidential with extremely restricted access, and they are usually custom-built at the level of a country, a sector or even a specific project.

Problem 1 – Restricted Access

For those interested in strengthened transparency in the extractive sector this story will sound remarkably familiar. As was traditionally the case with revenue data or extractive sector contracts, economic models are nearly always confidential.

Companies view their project economics models as proprietary and never make them available for public scrutiny. The same seems to be true for governments. Even among developing countries with strong transparency credentials we know of no examples where revenue projection models have been made available for public scrutiny. And the same approach to transparency seems to characterize international development institutions including the IMF and World Bank. Although the results generated by these models are sometimes included in public reports, the models themselves are closely guarded. The only exceptions to confidentiality seem to come from CSOs, with public models recently published for Timor Leste and Uganda.

Confidentiality undermines the utility of economic modeling far beyond the obvious problem of excluding those outside of government. It also constitutes a major barrier within government circles. Knowledge is power, and power is often not shared. An underappreciated benefit of contract disclosure is ensuring easy access across all relevant government ministries to documents that are often closely guarded. The same challenges of limited access exist for economic models and the input data on which they are based. Confidentiality also stands in the way of peer review undermining the reliability of the results and obscuring the visibility of potentially flawed assumptions.

There should be a presumption that all models developed with public funds (donor or developing country) should be open rather than closed. What would this mean in practice? The principle of transparency in extractive sector good governance is now widely accepted. Our definition of a meaningful implementation, in this context, is that economic models should be accessible, at the very least, to EITI Multi-Stakeholder groups.

As has been the case in the past, commercial sensitivity will be raised as a reason why this cannot be done. But where contracts have been disclosed, the bulk of the necessary information should already be in the public domain. This includes not only fiscal terms and tax laws, but also production volumes and price data (required by the 2013 EITI Standard). The exception may be company-specific cost data.  But this is not a reason to keep models confidential. Specific cost data can be replaced by information already in the public domain or industry averages.

Problem 2 – No Cumulative Learning

The underlying objective of project economic models is the same. A series of inputs (e.g. production volumes, sale price, production costs and fiscal terms) are manipulated in order to generate a series of outputs (e.g. project profits, government take, the company rate of return). And there are clear industry norms on how this should be done (See Upstream Petroleum by Kasriel and Wood and Guidelines for Economic Evaluation of Mineral Projects). But there are many different ways to design spreadsheets, structure input data, perform calculations and convey results. In spite of the common underlying purpose and logic, therefore, custom-built models look and function very differently.

These differences between custom-built models are a major barrier expanding the use of modeling, as there is little if any cumulative learning. The problem is much larger than one approach used in the petroleum sector and another for mining. It is not uncommon for donors to fund multiple models focused on the same project, each built to custom specifications allowing no interoperability. Training on one model provides no insight into the operation of another.

From the perspective of model developers, the custom-build approach makes good sense, and may even generate a competitive advantage. But for resource rich developing countries, the outcome is highly counter-productive. At the outset, a new model looks promising and is accompanied by extensive documentation and training sessions for officials. The early utility declines rapidly however because the model is seldom shared with the right people, because input data is not updated, and because attention is diverted when a newer model is commissioned.

For economic modeling to play a greater role in extractive sector good governance, design should be driven by the needs of analysts in developing countries, inside and outside government. This suggests a move away from custom-built models and towards open-source software combined with openly accessible data.

Is a Generic Public Model Needed?

Can existing modeling efforts overcome the problems of confidentiality and a lack of cumulative learning? Unfortunately for the most part the answer seems to be no. Most efforts by companies and consultants, even when working in the public policy space, have based their business model on confidentiality. Their economic interests are directly in conflict with greater openness and broader utility.

The IMF FARI model may be an exception. Designed to accommodate both petroleum and mining sectors, it addresses at least part of the cumulative learning problem. And although the model currently remains confidential, there have been indications in the last six months that the IMF may be considering making a version of the model, excluding confidential data, more widely available. NRGI recently reiterated the call for the FARI model to be released into the public domain.

Given the importance of the FARI model in fiscal regime design for resources rich developing countries, putting FARI in the public domain is an obvious next step. If the model is made public, it will be possible to assess whether FARI is a good starting point for an open source model. There are reasons however to be skeptical. By the IMF’s own admission, use of the model requires “strong economics and Excel skills.” “Unwieldy” would probably be a fair characterization, particularly when multiple projects and jurisdictions are added.

This overview of the existing landscape of economic modeling in the extractive sector suggests the there would be value in developing an open-source economic model complemented by open-source data. Initial thoughts on how that might be done will be the subject of our next article.

Category: Blogs, OpenOil blogs · Tags:

Comments are closed.