The IEA, 27% biofuels and the cost of indirect land use change

A few weeks ago, the IEA published a roadmap for biofuels in transportation. This argues that 27% of transport energy needs worldwide could be met with biofuels by 2050, saving a cumulative 20 billion tonnes (Gt) of carbon dioxide emissions over the next 40 years – all without affecting food security. The headline messaging for the report has been broadly represented (,, as a call for continuation and expansion of existing biofuel mandates. By 2050, the roadmap assumes 32 exajoules of biofuel production, split 50:50 between crop-based and waste-based fuels, and that the crops will require 70 million hectares (Mha) of land to grow.

What about the land?

Now, you just don’t get 70 Mha of land carbon free. 70 Mha is a little more than the area of Germany and Poland together and similar to Texas. There is a lot of carbon stored in plants and soils in an area that large, and the IEA recognises this. Indeed, they comment that, ‘in the most extreme cases the emission savings … could in some cases be more than off-set by the emissions caused by land use change.’ They even have some actually quite strong recommendations for dealing with the problem – they call for the world to ‘Adopt an overall sustainable land-use management system that aims to ensure all agricultural and forestry land is comprehensively managed … to avoid negative indirect land-use change’, and list various ways to reduce risks (more at the bottom of this post).

But after reviewing, discussing and recognizing these issues, the IEA decides that trying to actually work out and account for the land use emissions would be too uncertain, so they haven’t done it.

Conversion emissions

While it is hard to be certain, we can hopefully do a little bit more than just throw our hands up in the air. Helpfully, the European Joint Research Centre (JRC) released a report last year that looks at some similar issues. They suggest that 40 tonnes of carbon release per hectare is a rough but reasonable estimate for the typical carbon cost of land conversion, towards the low end of the IPCC range of land conversion emissions for Europe/USA. Of course, if the land comes primarily at the expense of forests the carbon release would be much higher, etc. etc.

Unfortunately, at an average of 40 tonnes of carbon per hectare, our 70 Mha allotment of land would result in about 10 Gt of emissions – that would cancel out fully half of the cumulative 20 Gt of emissions savings that the IEA hopes to achieve by 2050. That would mean that using biofuels would not cause any reduction in atmospheric carbon dioxide levels until after 2050.

Now, it isn’t quite that simple. There are more ways to make 70 Mha available than just ploughing up new land. Maybe food price rises will reduce demand by making people eat less. Some biofuel crops produce coproducts that reduce demand for animal feed. You can argue that the extra income from biofuels will let farmers improve yields. These factors would reduce the amount of new land that would need to be converted, and less land use change would mean less land use change emissions.

The IEA argues that these questions are, again, too uncertain. But if you can’t answer them, the uncertainty doesn’t go away – it applies directly to the headline carbon savings the IEA is claiming. So let’s think about the possible answers by using some of the results from the JRC report, which details the extent to which improved yields, reduced consumption and use of coproducts are predicted to limit the need for land use change in different models.

Here’s an example. The Global Trade Analysis Project (GTAP), the model used by the California Air Resources Board, models where the land for maize ethanol will come from. In the table below, we can see how several successive factors reduce the net land demand for every hectare of gross land demand:

Gross land requirement in hectares 1 ha (Remaining net land requirement)
Land saving from reduced food consumption 0.36 ha (0.64 ha)
Land saving from using by-products 0.31 ha (0.33 ha)
Land saving from changing yield 0.14 ha (0.19 ha)
Ratio of net land use change to gross land requirement 0.19  

In this case, they find that you would only need one fifth of a hectare of net land use change for every hectare of gross land requirement. That would cut the land use change emissions from the roadmap (assuming an average of 40 tonnes of carbon released per hectare) by 80%, to a total of 2 Gt between now and 2050. However, not all of the cases are so positive. When looking at European oilseed rape biodiesel the same model found only a small reduction: 0.9 hectares of land use change required for every hectare of biofuel production. At that rate, we would be back up to 9Gt of land use change emissions.

So what might be a reasonable number for the net:gross land demand ratio in the IEA roadmap?

  • The IEA argues that the 27% biofuel target can be achieved without damaging food security. If that’s true, then there shouldn’t be much area saved by reduced food consumption.
  • By 2050, the crop-based biofuels in the IEA roadmap come primarily from sugars and cellulosic bioenergy crops, which don’t save much land with coproducts.
  • Yield improvements for these bioenergy crops are already built into the IEA model to give the 70 Mha of additional land demand, so there shouldn’t be a big additional reduction there either. Because these are largely purpose grown bioenergy crops, any technological improvements (except for in sugarcane) won’t apply to food production so we won’t save much land by simultaneously improving yields in other sectors either.

Severely reduced benefits

From all of this, I think it’s fair to put the expected ‘land conversion ratio’ relatively high for the roadmap – as an example, we could suggest a rough guesstimate that 0.8 hectares of land will be converted for every hectare of gross biofuel demand. At that level, and with 40 tonnes of carbon released per hectare, we would have 8 Gt of extra emissions from land use change between now and 2050.

The conclusion? The IEA, having decided not to try to calculate land use change emissions, argues that biofuels can deliver 20 Gt of carbon savings by 2050. But with some fairly conservative assumptions (remember, we are not assuming that great swathes of forest land will be lost) we find that this is reduced by 40% to only 12 Gt. We might question whether this smaller gain is worthwhile when weighed against the likely increased water stress and loss of biodiversity and other ecosystem services from turning 70 Mha of land over to agriculture.

Methodological problems

With all of these potential land use change emissions ignored, it’s a little misleading of the IEA to emphasise a big headline number for carbon savings. But choosing to ignore land use change emissions also introduces a big flaw in the methodology of the IEA analysis.

The IEA biofuels roadmap is based on their ‘BLUE’ energy scenario for 2050, which is a scenario for the most cost effective way to reduce carbon emissions from energy use by 50% by 2050. However, if you include land use change emissions, that reduces the carbon saving that you can achieve for every $ of biofuel – i.e. the IEA’s abatement cost estimates for biofuels will be artificially low.

So how cost effective are biofuels according to IEA? Well, in the BLUE map, they’re already at the expensive end of carbon abatement options. In the figure below taken from the IEA ‘Energy Technology Perspectives 2008’ report, we see that ‘transport alternative fuels’ fall at the right of the cost curve.

An expensive solution

What if we factor land use change emissions into the cost calculation? If the savings are entirely wiped out by land use emissions, biofuels would have no place on the curve at all; but even a more moderate level of land use could significantly alter the cost profile.

As it turns out, by 2050 the IEA’s scenario relies on relatively land efficient feedstocks with high carbon savings. Even so, using take the example we explored above, with four fifths of the land for biofuels coming from land use change an abatement cost reported by IEA as $300 per tonne would really be more like $380 per tonne. That means that biofuels are less cost effective than IEA claims – and even by the terms of the IEA Roadmap, should have been given a smaller role in the BLUE map than the 27% of transport fuel that the IEA has announced.

It looks even worse for the cost effectiveness of biofuels in the short term. In the figure below, we’ve made a couple of big assumptions; in particular assuming that biofuels have the same cost profile as the other technologies in the same cost range, and that 2050 costs are a reasonable proxy for 2020 costs.

What we see is that for a plausible iLUC scenario to 2020 (where two fifths[1] of biofuel expansion requires new land, 40 tonnes of carbon is released per hectare and land use emissions are amortised over 20 years), the inclusion of iLUC totally changes the cost analysis (if three fifths of biofuel area came from land use change there would actually be no carbon benefit at all, i.e. an infinite cost). This would be still true with a more accurate 2020 cost curve, and demonstrates why land use change emissions should have been included in the IEA calculations if they were to be credible.

In conclusion…

Here are four key points to take away from all of this:

  1. A cost-of-carbon-abatement based analysis like the IEA BLUE map that ignores land use emissions is in danger of giving highly flawed results.
  2. By their own methodology, IEA’s recommendation for the use of biofuels in transport to meet a 2050 carbon reduction target should have been less than the stated 27%.
  3. Rather than 20 Gt of carbon savings accumulating by 2050 from the IEA roadmap, the real value could be much less; we estimated only 12 Gt, and it could even be negative in the worst scenarios (e.g. rainforest and peatland conversion).
  4. The carbon benefits of expanding conventional biofuels production from crops in the short term (e.g. 2020) without avoiding indirect land use change (for instance by implementing more sustainable land planning) are highly questionable.

Finally though, we should take a moment to remember and emphasise some of the less widely reported IEA recommendations (Roadmap page 18), which (if implemented before we ramp up biofuel targets) could actually help manage these risks:

  • focus on wastes and residues as feedstock;
  • maximising land-use efficiency by sustainably increasing productivity and intensity and choosing high-yielding feedstocks;
  • using perennial energy crops, particularly on unproductive or low-carbon soils;
  • maximising the efficiency of feedstock use in the conversion processes;
  • cascade utilisation of biomass, i.e. linking industrial and subsequent energetic use of biomass;
  • co-production of energy and food crops.

[1] We’ve used a lower ratio of land use change to gross land requirement here than when we discussed 2050, as the 2020 scenario is based on conventional biofuels that will result in reduced food consumption etc.



Alternative fuels Life-cycle analyses