Mitigation potential in EU agriculture

In my previous post on this blog, I noted that the Commission’s impact assessment (IA) accompanying its presentation of the new Effort Sharing Regulation (ESR) proposal concluded that very little additional agricultural mitigation is expected in the period 2021-2030, over and above what is projected to occur under current policies.

Two possible conclusions might be drawn from this finding. One is that the agricultural sector lobby organisations have used their political clout to ensure that the sector is required to do as little as possible to contribute to the EU’s 2030 climate targets. This reaction was advanced by some NGO activists in response to the post.

The other explanation is that it is difficult (and expensive) to achieve additional agricultural mitigation above and beyond the business as usual scenario. This was the view of the European Council in October 2014 when, in agreeing the EU’s 2030 climate targets, it noted the lower mitigation potential of the agriculture and land use sectors. Thus, when you model the least-cost pathway to achieving the 2030 target (bearing in mind the need to hit the 2050 target in the decades after that), it may turn out that agricultural mitigation options are not included in the least-cost portfolio at least in the coming decade.

This debate turns, in part, on the question of agriculture’s mitigation potential – what is the abatement potential, or the potential reduction in agricultural emissions, that could be achieved at a given carbon ‘price’. I try to answer this question in this post although it will become clear that the answer is far from simple. It does appear that the potential for agricultural mitigation has been deliberately under-played in the Commission’s impact assessment, most likely due to a belief that mitigation would take place as a result of a fall in production rather than due to the uptake of mitigation technologies.

In reading through this rather technical post, it is important to remember that the expected contribution that agriculture is asked to make to the 2030 targets will not be determined by the outcome of economic modelling, but rather by the concrete decisions that member states will make when trying to meet their national targets under the ESR.

Concept of mitigation potential

The concept of mitigation potential is a slippery one. The IPCC 5th Assessment Report drew a distinction between technical, economic and market potential, illustrated in the diagram below.

    • Technical mitigation potential is the full biophysical potential of a mitigation option, taking account of constraints such as land suitability, but without accounting for economic or other constraints.
    • Economic potential is the mitigation that could be realised at a given carbon price over a specific period, but it does not take into consideration any socio-cultural or institutional barriers to practice or technology adoption.
    • Market potential is the mitigation that could be realised under current or forecast market conditions encompassing biophysical, economic, socio-cultural, and institutional barriers to, as well as policy incentives for, technological and / or practice adoption.

Economic mitigation potential may be the most relevant when setting targets, but the behavioural and institutional constraints to achieving the economic mitigation potential at a given carbon price are often considerable.

Mitigation and cost-effectiveness

Related to the concept of economic mitigation potential is the principle of cost-effectiveness. Cost-effectiveness dictates that we should try to achieve climate policy targets and goals in ways which minimise the loss in economic welfare which might result after taking into account the value of any co-benefits of climate policy.

The principle of cost-effectiveness is recognised in the UN Framework Convention on Climate Change (UNFCCC) in Article 3 which notes that “Where there are threats of serious or irreversible damage, lack of full scientific certainty should not be used as a reason for postponing such measures, taking into account that policies and measures to deal with climate change should be cost-effective so as to ensure global benefits at the lowest possible cost”.

The principle of cost-effectiveness is often hard for non-economists to understand. Faced with the urgent need to reduce greenhouse gas emissions, the eager climate activist will want to fire on all cylinders at once – eliminate vehicle emissions, retrofit the housing stock, decarbonise the energy system, and cut agricultural emissions. The principle of cost-effectiveness states that, once targets have been set (and they should be ambitious targets given the risks of destabilising the cimate), then it makes sense to prioritise. Initiatives and actions which reduce emissions more cheaply should be undertaken first. This not only makes sense in terms of an efficient use of scarce resources, but also in terms of maintaining public support for the major transformations which are required.

The most usual way to prioritise the most cost-effective options is to construct a Marginal Abatement Cost (MAC) curve. A MAC curve presents the extra (or ‘marginal’) costs and carbon reduction (or ‘abatement’) potential of different mitigation options, or technologies, relative to a baseline. A typical example of what a MAC curve might look like is shown in the illustrative diagram below.

The different mitigation options are shown as the grey columns in the diagram. The width of each column represents the amount of abatement that could be achieved by that option. The height of each column represents the abatement cost per year, expressed in euro (or dollars, or yen) per tonne of emission avoided. The area of the columns representing the low-cost options represents the economic cost of choosing that particular low-cost pathway.

The diagram is divided into four clusters. The first ‘double dividend’ cluster contains those options which can be undertaken at negative cost, i.e. they would both reduce GHG emissions and save money. It might be hard to understand why such apparently advantageous options should exist, and why farmers should not already want to embrace those options. There are a number of possible explanations, including market imperfections, lack of information, behavioural motivations and institutional constraints. Indeed, these barriers can exist for any option in the MAC curve – they explain the difference between the economic mitigation potential and the market potential. Each of these explanations suggests a possible set of responses designed to overcome these barriers to adoption.

The second cluster consists of those options which are broadly ‘cost-neutral’. The third cluster groups those options which are ‘cost-effective’, meaning that they would be attractive at carbon prices roughly similar to those prevailing elsewhere in the economy. Finally, there is the cluster of ‘cost-prohibitive’ options. These are options where there is a technical potential to reduce emissions but the cost, given current technologies, is prohibitively high.

Adding up the abatement amounts resulting from the options in the first three clusters gives the total economic mitigation potential for any given carbon price. As the carbon price increases, so does the economic mitigation potential. The market potential will be less because of the barriers to adoption which will be specific to each mitigation option.

Looking under the hood – a short digression on estimating MAC curves

While reading a MAC curve may seem straightforward, putting one together is anything but. MAC curves are not readily observable, particularly in agriculture where mitigation measures have only begun to be implemented. This means that estimates are derived from modelling studies and simulations, and there is a good deal of uncertainty surrounding these estimates.

Two French economists undertook a meta-analysis study of mitigation potential in global agriculture (based mainly on US and European studies) and concluded: “Even when narrowing the analysis down to the sole emissions of methane (CH4) and nitrous oxide (N2O) from agriculture, a quick overview of the available results reveals a wide range of abatement rate estimates in the literature. For a commonly used price of 20€/tCO2eq, the predicted abatement rates may vary by a factor up to 20 from one study to another” (Vermont and De Cara, 2010).

What are the key factors affecting the estimated abatement potential for different carbon prices? Three factors are particularly emphasised by Vermont and De Cara.

The modelling approach used. Different modelling approaches can be used to estimate abatement potential. The three main approaches are engineering models, optimising programming models and equilibrium models. Engineering studies are usually extremely detailed ‘bottom up’ models which, when combined with cost estimates, can give estimates of abatement potential. Programming models model how farmers optimise the resources at their disposal, subject to technical and economic constraints. A MAC curve is generated by observing how production and resource use and thus emissions respond as a ‘price’ is placed on emissions. Finally, equilibrium studies take into account output and input price reactions as emissions are priced to derive a more policy-relevant abatement potential. Each approach has its strengths and weaknesses. Vermont and De Cara found that equilibrium studies tended to show higher abatement potential than estimates from the other two approaches.

Assumptions on the availability and costs of mitigation technologies. The set of mitigation options considered matters. For example, it makes a difference whether only supply-side technological options are considered or whether non-technical mitigation options such as changing behaviour and preferences (for example, changes in diets towards consumption of less meat and dairy products) are also taken into account.

Mitigation can occur either through changes in activity levels (numbers of animals, area of crops) or through changes in technology (embracing both technical and management changes) which lower emissions intensity per animal or per unit area. Abatement potential will depend on whether changes in activity levels are constrained not to exceed certain boundaries or not. Agricultural mitigation through changes in activity levels will lead to the phenomenon of ‘carbon leakage’ if there is no change in demand. Global emission levels will therefore change less, or could even increase, depending on the relative emissions intensity of the substitute sources of supply.

It is not only a question of the options themselves, but also the level of detail with which they are modelled. For example, some studies might account for crop-yield response to nitrogen (N) fertilisers, while in others crop yields and per-hectare N input use are assumed exogenous. Not surprisingly, Vermont and De Cara found that increasing the degree of flexibility with respect to input use in the underlying production function is likely to lead to larger abatement potential estimates.

When estimating abatement potential into the future, the assumed rate of technological development and thus cost reduction of mitigation technologies is also important.

The treatment of no-regret options. The first cluster of options in the illustrative MAC curve above was labelled ‘double-dividend options because they reduced emissions at negative costs. Such no-regret options can be uncovered using the engineering approach. However, they are ruled out by construction in the programming and equilibrium approaches which assume that the base scenario corresponds to an optimum. This suggests that the total cost of achieving a particular amount of abatement might be lower in engineering studies compared to other approaches.

Vermont and De Cara in summarising the studies they examined (which were not confined to Europe), found that “The mean abatement rates at 10, 20 and 50€2005/tCO2eq are 8, 10 and 13% of baseline emissions, respectively”. However, they also concluded that: “The range of abatement rates for these three prices (0–57%, 1–57% and 2–69%, respectively) is large, with coefficients of variation close to 1.” This underlines the point made earlier that there can be very large differences between studies in estimates of agricultural mitigation potential even for the same carbon price.

Agricultural abatement potential in the Commission impact assessment

We gain some insights into the potential costs of agricultural mitigation from the LULUCF IA published by the Commission. In its economic assessment of the flexibility options which would allow a certain number of LULUCF credits to be used to offset emissions in the non-Emissions Trading Scheme (NETS) sectors, the Commission estimated the impact of the different flexibility options on both abatement costs and production impacts in the agricultural sector.

It presents results from two different modelling frameworks which examined the impact of reducing agricultural emissions by 20% in 2030 compared to 2005. One estimate of abatement costs comes from the GAINS model housed at the International Institute for Applied Systems Analysis (IIASA). The methodology used in the GAINS model is described in this working paper by the GAINS team at IIASA. The other estimate comes from the EUCLIMIT modelling framework which underlies the Reference Scenario 2016 and the IA of European Council scenarios. The estimates are summarised in the table below.

GAINS model abatement costs. According to the GAINS model, in option F0 (no flexibility) agriculture non-CO2 emissions would have to be reduced by 78MtCO2eq in 2030 compared to 2005 under the assumption of a 20% reduction in emissions. Under flexibility options F1 to F3, an increasing share of this reduction would be delivered by the land use and forestry sector. The options F1 to F3 show a progressive decrease in the mitigation cost in the agriculture non-CO2 sector as the degree of flexibility increases.

Without flexibility, a carbon price of €78.6/tonne CO2eq was modelled to meet the Agriculture non-CO2 sector emission reduction of just over 78MtCO2eq. The annual (direct) costs of this would be €2.071bn in 2030. This estimate is consistent with the Vermont and De Cara (2010) result which projected a 13% fall in emissions for a carbon price of €50/tonne CO2eq. The GAINS model is projecting a 20% fall in emissions would require a carbon price of €78.6/tonne.

As a consequence of allowing increased rates of flexibility (Options F1, F2, F3), the marginal cost of mitigation in the agriculture non-CO2 sector falls to half (€32.5/tonne) under the Low flexibility option F1, 1/10th (€7.3/tonne) for the Medium (F2) and zero for the High (F3) flexibility options. The F1 Low flexibility option implies a reduction of 11% in agricultural emissions at a carbon price of €32.5/tonne, the F2 Medium flexibility option implies a reduction of 6% in agricultural emissions at a carbon price between €7.3 and €31.4/tonne and the F3 High flexibility option shows that a reduction of 2% in agricultural emissions could be achieved at zero cost. Zero-cost mitigation options in the GAINS model include genetic enhancement, anaerobic digestion for larger farm sizes and a ban on burning agricultural waste.

EUCLIMIT modelling abatement costs. The EUCLIMIT abatement costs for broadly the same percentage reductions in emissions are generally higher than the GAINS model. Even a 3% reduction in agricultural emissions would require a carbon price of €10/tonne, while to get the full 20% (on top of the reduction in the Reference Scenario 2016 of -2.4%) would require a carbon price of €120/tonne.

This is partly because the EUCLIMIT models do not take account of negative or zero cost abatement options. Unlike GAINS which is an engineering model, the CAPRI model used in EUCLIMIT combines an optimising with an equilibrium model and thus cannot accommodate ‘no-regrets’ options. In any case, profitable mitigation options would already be taken up in the Reference Scenario so only more costly mitigation options are available for the abatement scenarios, although it appears few mitigation technologies in CAPRI are sufficiently profitable to encourage farmers to adopt them on their own. Also, CAPRI includes costs directly related to the determinants of technology adoption going beyond pure profitability considerations and which are generally unknown.

EcAMPA abatement costs. Further insight into the EUCLIMIT results on the mitigation potential in agriculture comes from the EcAMPA 2 project (Economic Assessment of GHG mitigation policy options for EU agriculture) undertaken by the Joint Research Centre. EcAMPA used the same CAPRI model also used for the EUCLIMIT results. This study looked at the cost of achieving different reduction targets for agricultural emissions with different policy instruments (specifically, whether subsidies were offered to incentive mitigation options or not), as well as at the agricultural production and market effects of these options.

Although not the main focus of the study, it is possible to extract a version of a MAC curve based on the modelling and technology assumptions used in the CAPRI model (using the data in Annex 4 of the report). This MAC curve shows limited but positive mitigation potential on the basis of the 12 mitigation technologies identified in the study. Hardly any of these technologies would be adopted by farmers in the absence of an incentive (either a carbon levy or a subsidy).

With a carbon price of €10/tonne CO2eq, around 3.5% of 2030 emissions would be abated (beyond the business as usual level). With a carbon price of €50/tonne, this would rise to 10% (these options are shaded in the diagram below).

At this level of carbon price, around 27% of the mitigation would come from reduced activity, and the balance from uptake of the identified mitigation technologies (extrapolating from the figures in Table 29 in the EcAMPA 2 report). The more stringent the reduction in agricultural emissions, the greater the share that comes from reduced activity in 2030. Conversely, for a carbon price below €50/tonne, the contribution from reduced activity would be lower.

It is hard to make a direct comparison with the cost of abatement in the ETS or other NETS sectors because information on the implied carbon values of meeting the European Council targets in the ETS and NETS sectors is not given. The Reference Scenario 2016 shows a shortfall of 6% in relation to the 2030 target of a 30% reduction in emissions relative to 2005 (that is, NETS emissions are expected to fall by 24% in the Reference Scenario 2016). If it were assumed that this shortfall could be eliminated by raising the price of carbon emitted in the NETS sector to €30-35/tonne, then it would seem that agriculture could make a proportionate contribution to this reduction.


The analytical modelling behind the Commission’s impact assessment supports the view that significant agricultural mitigation is costly and that (for a given carbon price) the agricultural mitigation potential is lower than in other sectors. However, it also suggests that there is a relevant potential for abatement in agriculture which could be taken up with a value on carbon similar to that in place in other sectors of the economy.

The CAPRI modelling in EUCLIMIT/EcAMPA may overestimate this abatement potential. It may underestimate the constraints due to behavioural and institutional barriers to the adoption of mitigation technologies (e.g. the age structure of farming) even though some allowance is made in the modelling for unknown costs which may limit uptake. The question of whether the foreseen emission reductions can be captured in the inventory accounting system is also not addressed – problems of monitoring and verification are particularly difficult in agriculture. On the other hand, the huge variation in production efficiencies (and thus emission intensities) across farms mainly due to differences in management is not well captured in the CAPRI model. This means that the model may underestimate the potential improvement in management practices that could occur if a price were put on carbon emissions.

Given some potential for agricultural mitigation, the modelling strategy in the EUCO27 and EUCO30 scenarios used in the Commission’s impact assessment to assess the implications of the European Council targets is puzzling. In my previous post, I noted that the value put on carbon in the EUCO27 scenario was just €0.05/tonne, with the result that only a small amount of agricultural mitigation is foreseen in this scenario. I overlooked the fact that the EUCO30 scenario includes no policy incentive at all to reduce agricultural emissions until 2030. Not surprisingly, the scenario shows no additional agricultural mitigation (beyond business as usual) taking place in this scenario. The assumed absence of agricultural mitigation in the Commission’s scenarios is a function of the modelling strategy, not because of technical difficulties in reducing agricultural emissions.

The Commission may have left agriculture to one side because of a concern that emissions reduction in this sector can take the form of a reduction in activity levels rather than a change in technology. It quotes the EcAMPA 2 study and other studies as showing “that only a limited level of mitigation of emissions from agriculture is feasible without impacting production” (LULUCF impact assessment, p. 54). However, the EcAMPA 2 study investigated a 20% reduction in agricultural emissions relative to the baseline in 2030 (a level of mitigation inspired by the outcome of the Reference Scenario 2013 used in the impact assessment of the Climate and Energy Framework 2030). The level of additional mitigation sought in the NETS sector in the EUCO27 and EUCO30 scenarios would be much smaller, and thus a much smaller impact on production would be expected as argued above.

In any case, it is not obvious why questioning the existing level of agricultural production or the projected level in 2030 under current policies should be treated as a no-go area. There is no case to maintain production in Europe if the total costs of production (including the social costs of greenhouse gas emissions) exceed the total return to society. There is thus no case to exclude changes in activity levels in agricultural production as a legitimate option for agricultural mitigation.

In conclusion, it has to be said again that the Commission analyses are merely indicative. Once the targets are agreed, it will be up to individual member states to decide how their targets will be met. It is at the national level that the debate will occur regarding the relative contribution of agriculture to the NETS 2030 targets. As noted in my previous post, the emphasis given to reducing agricultural emissions is likely to be very different among the member states.

This post was written by Alan Matthews.

Photo credit: Ploughed moorland © Copyright Jim Barton and licensed for reuse under this Creative Commons Licence.

Print Friendly, PDF & Email