IMF

Ben Sills of BusinessWeek reports today that the International Energy Agency (IEA) estimates global fossil fuel subsidies in 2008 at $557 billion.  The figure is based on an interview the magazine did with Fatih Birol, Chief Economist of the International Energy Agency. 

More than half a trillion dollars per year in subsidies to fossil fuels is, indeed, a large number.  However, based on other sources and other work I am doing, I believe it is too low.  This conclusion is not based on any detailed information I've seen from IEA.  While I attended an expert meeting in February on G20 fossil fuel subsidy reform and provided detailed comments on their draft joint report and estimation methodology, I have not been involved with the G20 subsidy reform process since that time.  My rationale for concluding the numbers are too low include:  

• IMF numbers for oil alone were nearly as high.  In February, the International Monetary Fund released its own analysis (blog post on report; link to report) of global subsidies to petroleum.  They pegged oil subsidies for 2008 at  at least $519 billion, much higher than the $312 billion IEA says went to oil during that same year. 
• IEA value is similar to past "price gap" estimates that form lower-bound for subsidy value.  The figures for both IMF and IEA appear to be primarily "price gap" calculations that examine the difference between world prices and domestic prices for fuels.  Price gap values tend to set the floor for subsidy levels, but miss many other policies that generate subsidies to various energy sources. 
• Earth Track case studies have found substantial data gaps in fossil fuel subsidy information.  For the past year or so, I've been working with the Global Subsidies Initiative on a series of case studies that examine data availability on fossil fuel subsidies in four countries: the US, Germany, China, and Indonesia.  Our review has turned up systematic weaknesses in data reporting and transparency in a number of key subsidy areas.  These include tax subsidies; credit support especially for state-owned energy infrastructure and power plants; and insurance and indemnification subsidies.  It is unlikely that IEA, even with its strong global relationships, was able to solve these substantial data gaps in the few months it has had to prepare its data set.  As a result, it is likely that they have systematically undercounted certain types of subsidies in most of the countries evaluated.  (If you would like to be notified when our case studies are publicly available, you can add your e-mail to our distribution list here). 

Even ignoring the political challenges of reaching consensus, it would be unrealistic to think that IEA or any of the other parties involved with obtaining data for the G20 fossil fuel subsidy reform process (World Bank, OECD, and OPEC; along with many member countries and other organizations such as the IMF that also have input) could assemble a comprehensive data set on global subsidies in such a short time frame. 

What will be important, however, is that IEA be clear about where it has not been able to measure subsidies; and that the Agency state how filling in those gaps would have influenced its reported numbers.   If values from multiple sources exist but conflict, IEA should provide a range if the estimates are plausible.

Finally, where IEA's review on fossil fuel subsidies found that survey countries are simply not collecting needed data at all, the Agency needs to flag the areas for future attention and improvement.  G20 subsidy reform needs to be viewed as a process, not as an event.  However, it will not be a successful process unless rules for full disclosure are put in place and properly enforced.

UPDATE:

Some additional information on the IEA figures can be found in the Financial Times; and on the subsidy reform work program at the OECD web site.  The FT notes that the current numbers are higher than IEA's prior estimates.  (This would be consistent with the sharp increases in energy prices between 2007 and 2008; whether portions are also due to broader coverage of producer subsidies can't be discerned without more detailed information). 

A reader also points out that part of the reason IEA's values are lower than IMF is that IEA's pick up a full year average for 2008, so include the subsequent declines later in the year.  IMF's are based on subsidy rates at mid-year before the sharp price declines.  Price gap subsidies tend to rise when energy prices rise quickly, as policy adjustments to raise prices in countries with consumer subsidies tend to lag.

Two final points.  First, while IEA is the cited party on these estimates in both the FT and BusinessWeek articles, the work is being conducted by four international agencies (IEA, OECD, World Bank, and OPEC), not just one.  And second, while it is not possible to tell at this stage for sure whether any types of subsidies have been systematically excluded, the current plan is to make the detailed basis for the estimates available in the fall.  This would be quite helpful in building more systematic data coverage over time.

David Coady and his team at the IMF have just released an update on estimated global subsidies to petroleum products. They have done consistently important work in this area, documenting how petrol subsidies have weakened the fiscal health of many countries around the world.  The IMF work has examined "pass-through" metrics for each country to assess how much of a change in world petrol prices shows up in end-user retail prices domestically.  This data has provided quite useful insights into the price transparency of economies in the energy sector.

For the current report, as in previous ones, they rely primarily on the price gap approach.  This examines the gap between domestic prices on oil products and world reference prices (adjusted for such factors as quality, transport, and distribution to ensure the prices are comparable). 

First, a couple of caveats on their numbers.  The price gap does not pick up everything, something Coady et al. fully recognize.  They hope (as do I) that coverage of other types of supports can be improved over time.  In addition, this paper does not incorporate subsidies to other fossil fuels such as natural gas, coal, or related subsidies to carbon capture and sequestration.  Expanding the estimate to include all fossil fuels, and producer subsidies as well as the price gap (that picks up mostly subsidies to consumers) would provide the most accurate estimate of how subsidies work against attempts to address global warming.

Nonetheless, even with the narrowed set of subsidies, the values the IMF found are enormous.  At the peak of the oil price boom in mid-2008, they estimated total world fuel subsidies at more than half of a trillion dollars per year, or 0.7% of global GDP (see page 10).  This dropped to about $140 billion a year later as prices fell sharply post-recession, and as efforts by some of the subsidizing countries to pare back payouts began to take effect.   Rising fuel prices have again begun to boost subsidies, with 2010 estimates of $240 billion (0.3% of global GDP).

The IMF also took a close look at the role of tax policy and how incorporating it into their subsidy calculations would affect totals. This increased their total subsidy estimates substantially:  to as high as $1.2 trillion/year in mid-2008 (1.6% of global GDP), and $742 to $965 billion projected for 2010 (1.0 to 1.3% of global GDP)(p. 10).

The rationale for including taxes is sound.  Like all goods and services, there is no reason that petrol consumption should be exempt from tax.  The IMF's tax-inclusive subsidy values do include an estimate of baseline fiscal charges. 

In addition, however, the measure includes estimates from the economic literature on petrol-related externalities.  Included impacts were global and local pollution, congestion, and accidents.  There is a strong efficiency argument to internalize these costs through pricing, and I wouldn't advocate just ignoring them.  The trouble, however, is one of measurement and politics.  The externality values are generally much less precise than the metrics for fiscal subsidies.  A review John Dernbach and I did on US fossil fuel subsidy estimates that incorporated externalities is instructive (see bottom of Table 2).  The externality calculations were a major source of variance, were defined differently in different studies, and varied from other externality estimates by orders of magnitude.  Cleaning up the estimates can also be a challenge.  To what extent are some congestion costs already internalized through lost time?  How have non-fuel tax fees related to other aspects of transport network usage (property taxes, tolls) been accounted for?  What about adjustment for transport miles reliant on non-petrol fuels?

Ideally, the IMF could work to separate the different types of taxes and fees in their model runs.  The first level are user fees (hypothecated funds in the European vernacular) for petrol-related uses:  financing roadways, cleaning up oil spills or leaking gasoline tanks.  These should not be counted as baseline taxes at all.  The second level, assuming that fees exceed related uses (which they don't in the US) is to set taxes at a fiscal parity with other goods and services.  Subsidy runs using this level of taxation would be helpful.  An additional run with estimates for the externality charges only could also be done, but this approach would segregate the impacts of fiscal subsidies from externalities.  Politically, I think this would be quite useful.

A final point on data accessibility.  There was some mention among IMF staff I spoke with that the baseline data on which the price gap calculations were based would be made public.  They were thinking of adopting an approach similar to the extremely useful practice of data sharing that has been adopted by the US Energy Information Administration with data in downloadable Excel files that are easy for users to integrate into their own work. 

There are now at least three international agencies working on price gap calculations for fossil fuels:  the IEA, the OECD, and the IMF.  The UN has also done work in this area in the past, as has the World Bank.  It would be nice to adopt an open source model for this information so that resources could be more effectively pooled and investments made into the shared data set to expand its coverage and detail.