How accurate are EIA subsidy numbers, and what should global subsidy reform look like?

Earth Track Blog Post

When people look for official information on US energy policy and trends, the Energy Information Administration (EIA), part of the US Department of Energy, is one of the first places they look.  I look there too: EIA staff do great work in many areas; they are willing to talk about their work and assumptions; and they have developed a model of data sharing that I wish many more government departments would adopt.

But EIA's numbers on energy subsidies have always irked me, widely cited though they may be.  The subsidy totals have been very low.  The patterns across fuels have been extremely sensitive to estimate shifts in even a single program.  The inclusion rules have often seemed arbitrary.  For example, EIA has sometimes excluded programs based on arguments that applied equally to subsidies the Administration did include.  Other times (e.g., with public power), EIA's reports have carefully developed a theoretical framework on how to evaluate the subsidies, quantified how big they are, and then failed to include the results of all this work in its subsidy tallies. 

The point here is not to cast stones.  There are resource constraints, narrow research mandates from Congressional requestors (with a bit of a political angle, I'm told), and a variety of other factors that have led to the outcomes we see.  But as the role of energy subsidy reform is increasingly recognized as a central tenet in a rational and cost-efficient response to climate change; and as fossil fuel subsidy phaseouts hit the realm of the G20, incomplete or inaccurate subsidy data becomes a big problem.

Today, I've am happy to release a detailed review of EIA's subsidy numbers.  EIA Energy Subsidy Estimates: A Review of Assumptions and Omissions explores EIA's numbers and a variety of the reasons they are so low (see table below).  I have no doubt EIA will continue to be tasked with tracking government energy subsidies; my hope is that this analysis will help them do it better in the future.

I see this Review as a companion piece to a paper I did last summer for the International Institute for Sustainable Development on the strengths and weakness of the "price gap" metric, the most common approach used to measure energy subsidies globally.

There are some important cross-cutting themes: 

First, accurate subsidy measurement is important, but it is also hard.  Political pressure makes that measurement even more difficult.  Fixing this problem will require a signficant commitment by governments, and likely by many other parties as well. 

Second, as was the case with corporate financial reporting, improvements to subsidy data need to be viewed as a process rather than an event.  We will not have comprehensive data on fossil fuel subsidies by the G20 deadline; and likely not two or even five years from now either.  But it also took a very long time to figure out a reasonable way to account for environmental or pension liabilities in corporate financial reports.

Even with the data challenges, and perhaps partly because of them, countries that are serious about reforming subsidies to environmentally harmful activities need to focus much more on building a system and process for doing so.  They need to tackle the problem in a structured way, but in phases. 

What I've seen in international efforts over the past 20 years has mostly been trying to re-task existing institutions to deal with the challenge of subsidy measurement and data collection.  This approach has not be very successful.  Often there is historical baggage, competing demands, or operating rules that impede the task at hand.  There may be an existing skill base that addresses some -- but not all -- of the needs for the new task; but limited will, budget, or expertise to build out the needed areas.  While using existing capacity makes sense in the short-term, I believe over the longer term a specialized and independent structure will need to be set up. 

I think we can gain many insights on what this independent structure should look like by studying relevant institutional structures that have worked in other areas.  Some of the ones I think would offer valuable lessons include: 

  • The initiation of systematic collection of macroeconomic statistics throughout the world after the global Depression in the 1930s.
  • The accounting standards boards (FASB, GASB, IASB) that iteratively, and substantially independently of government, have evolved standardized rules for corporate reporting.
  • The International Standards Organisation that has developed many challenging standards, such as production quality, that have been compelling enough for private firms to go to great lengths to meet iteratively over time.

Finding a handful of failures would also be quite useful in charting the path forward and what might go wrong.

It is inevitable that member governments of the G20 will waste innumerable hours fighting over how to define a "subsidy" and work to extend the number of years they can keep an existing subsidy and still meet the "medium term" phaseout pledge.  But if this is all that happens, we will come up empty.  Less political entities should begin thinking about institutional structure now.  Just as accurate and verifiable information on corporations was a pre-requisite for the development of mature capital markets, the benefits of accurate and verifiable information on subsidies will also be enormous -- not only from fiscal savings, but also through less expensive and large scale gains in environmental quality and poverty reduction.

Related documents on EIA subsidy values and subsidy reform:

Complete Review
Executive Summary only
Extracted table comparing EIA subsidy estimates to other subsidy research
Review of "price gap" approach commonly used internationally to measure energy subsidies
Earlier reviews of EIA subsidy estimates:  20011993.

Expected Bias Resulting from EIA Subsidy Definition and Valuation Conventions


Scale of impact/year

Issue understates subsidies to:

Use of point rather than range estimates

$5.3 billion for subset of tax expenditures alone

Oil, gas, nuclear, coal, efficiency

Use of revenue-loss rather than outlay-equivalent metric for tax subsidies


Oil, gas, wind, biofuels

No marginal analysis of new and expanded subsidies


Clean coal, nuclear

Use of current account rather than actuarial balance on trust funds to assess subsidy level


Nuclear, fossil (to a lesser extent)

Omission of subsidies related to insurance and publicly provided market oversight


Nuclear, coal, hydroelectricity

Omission of minimum purchase requirements such as Renewable Fuel Standard


Liquid biofuels; renewable electricity if federal RPS enacted

Omission of support to bulk fuel transport infrastructure

~1–2 billion

Oil, coal, and, to a lesser extent, ethanol and liquefied natural gas

Omission of support to energy security

>$10 billion

Primarily oil, with some benefits as well to nuclear and natural gas

Omission of subsidized credit through export credit agencies and multilateral development banks


Oil, gas, coal, renewables, new nuclear

Omission of use of tax-avoiding corporate forms


Oil, gas, coal

Omission of lease-related subsidies

>$1 billion

Oil and gas, synfuels

Inadequate reflection of subsidies to public power

>$1 billion

Coal, natural gas, nuclear, hydroelectricity

Omission of most accelerated depreciation to energy


Oil, coal, natural gas, wind, biofuels, new nuclear

Omission of most energy-related tax-exempt bonds


Coal, natural gas, wind, biofuels