Get Complete Project Material File(s) Now! »
Cost-effectiveness analysis in the literature
For cost-effectiveness analysis Vermont and De Cara (2010) identify three broad approaches for the derivation of marginal abatement cost curves (MACCs), the device typically used to evaluate pollution abatement costs and benefits. These are: i) a bottom-up or engineering approach; ii) an economic approach consisting of modeling the economic optimization of a set of (in this case) farm operations; iii) a partial or general equilibrium approach that extends and relaxes some of the assumptions about wider price effects induced by mitigation activity.
The engineering approach focuses on the potential emission reduction of individual measures and observes their cumulated abatement and associated costs. The required data to appraise abatement costs are ideally collected from measures applied on test farms, thereby reducing some uncertainty the estimated cost and mitigation potential for each mitigation measure. It is normally the case that more measures are assessed using the engineering approach relative to the economic approach (MacLeod et al. 2010, Moran et al. 2010, Pellerin et al. 2013).
The economic approach consists of modeling the economic optimization of a set of farm operations located within a given geographical scale. The objective function is typically to maximize profit of these farms under given constraints such as available arable land or even lay fallow land as imposed by agricultural policies. The introduction of a carbon tax as a new constraint, allows the model to reconfigure farm activities to accommodate the necessary GHG emissions reductions. The resulting loss in profit (opportunity cost) and GHG reduction provide the relevant abatement cost information.
Equilibrium models relax some of the cost assumptions made in the economic approach and include a description of the demand for agricultural products thereby allowing a price feedback into the cost of mitigation (Vermont and De Cara, 2014). Their level of spatial disaggregation is generally lower than that of bottom-up models and their geographic scope and coverage are generally wider. This approach has been used to assess abatement cost at the level of the USA (Schneider and McCarl, 2006; Schneider et al., 2007; McCarl and Schneider, 2001).
A noteworthy difference between the approaches is the frequent observation of negative cost options in the engineer approach for some options (Moran et al., 2010; MacKinsey & Company, 2009). These are obviated in any optimization approach and are in any case questioned by some authors. Kesicki and Ekins (2012) for example suggest that they more likely imply a failure to assess some hidden costs (diffusion of the information, administration barriers) than any real opportunity to reduce emissions while increasing farm gross margins. Another observation is that each mitigation measure in the engineering approach is associated with a constant marginal cost – creating a stepwise marginal abatement curve (each step corresponding to an option). This observation suggests that the economic potential per ton CO2 equivalent mitigation is the same for each specific option irrespective of spatial scale or in terms of the overall volume of emission reduction, which would seem unlikely. Indeed, due to regional variability in soils, farm systems, climate and yields, abatement cost would also vary for any individual mitigation measure. Results from studies employing the economic approach are depicted by continuous increasing abatement cost curves, with no negative cost. An advantage of these studies is optimization of fewer mitigation measures over a large number of farm types. For example De Cara and Jayet (2011) modeled around 1300 EU farms optimizing animal feed, a reduction in livestock numbers, a reduction of fertilisation and the conversion of croplands to grasslands or forests.
Legumes have been specifically assessed in a UK study constructing a national MACC for agricultural GHG emissions (Moran et al., 2010). The marginal abatement cost obtained for legume crops appears constant and very high (14280 £/tCO2eq equivalent to 20 849 euros/tCO2eq). This is in stark contrast to Pellerin et al. (2013) estimate of only 19 euros/t CO2eq. To explore some of the reasons for this disparity we adopt a predominantly engineering approach combined with elements of an economic approach to explore the role of farm systems decision-making around the adoption of legumes as a specific measure that can influence farm profitability.
Defining emissions and gross margin
The analysis assesses the abatement potential in 96 French metropolitan geographical areas, each considered as a single farm decision unit. The analysis is confined to the within farm gate effects and does not account for the upstream or downstream impacts; e.g. associated with lower fertilizer production, or the emission mitigation benefit related to enteric fermentation of cattle consuming legumes (McCaughey et al., 1999). In each geographical area, farmland emissions and profits are calculated and decomposed for each crop (Common Wheat, Durum Wheat, Barley, Maize, Sunflower, Rapeseed, Pea, Horse bean and Alfalfa).
We followed the 2006 IPCC Guidelines for National Greenhouse Gas Inventories (IPCC, 2006) to estimate N2O emissions per hectare. Using mineral nitrogen spreading rates and organic spreading rates from the Agricultural Practices survey (Agreste, 2010) we calculate the following kinds of emission sources:
– direct emissions, happening directly on the field.
– indirect emissions, covering emissions from atmospheric redeposition and leaching and runoff.
– emissions from crop residues.
The formula that determines each crop gross margin in each geographical area is summarized as follows (Ecophyto R&D, 2009) : 𝐺𝑀𝑘,𝑖= (𝑝𝑟𝑖𝑐𝑒𝑘,𝑖 ×𝑦𝑖𝑒𝑙𝑑𝑘,𝑖 )−(𝑒𝑥𝑝𝑝ℎ𝑦𝑡𝑜,𝑘,𝑖 +𝑒𝑥𝑝𝑓𝑒𝑟𝑡𝑖,𝑘,𝑖 +𝑒𝑥𝑝𝑠𝑒𝑒𝑑,𝑘,𝑖 ).
Where GM k,i is the gross margin calculation for each crop i in each geographical area k (in euro per ha). Price k,i is the crop price in euros per ton and yield k,i is expressed in tons per hectare. The expenses in phytosanytary products (expphyto,k,i ), in fertilizers spread (expferti,k,i ) and in seed (expseed,k,i) are all measured in euros per hectare.
Appendix 2.9.1 shows the results for the main crops cultivated in France and gives the baseline for overall farmland gross margin (6,4 billion euros) and for emissions (20,4 MtCO2eq). When comparing these emissions with those of the national inventory report, we observe that the amount represents less than half of the category ‘Agricultural Soils’ (46,7 MtCO2eq (CITEPA, 2012)). This category represents all N2O emissions linked to soil fertilisation both from cropland and grassland soils. Hence the baseline emissions assessed here is quite coherent since we only focus here on emissions from croplands which represent less than half of the French Utilized Land Area23.
Introduction of legumes onto croplands
Legume crops have low emissions per hectare and a low gross margin compared with other crops. Consequently, in most geographical areas, as the overall utilized land area remains constant, increasing the share of in N-fixing crops induces a reduction of both profit and emissions. Additional legume crop areas are introduced in each geographical area by 10% increments to the initial legumes area. The loss of profit (dCost) divided by the reduction of emission (dEmissions) linked to these additional areas represents the marginal abatement cost. The marginal cost and marginal emissions also integrate the preceding fixing effect, which induces higher gross margin and lower emission for following year crops that have been preceded by legumes. 𝑀𝑎𝑟𝑔𝑖𝑛𝑎𝑙 𝐴𝑏𝑎𝑡𝑒𝑚𝑒𝑛𝑡 𝐶𝑜𝑠𝑡= 𝑑𝐶𝑜𝑠𝑡𝑑𝐸𝑚𝑖𝑠𝑠𝑖𝑜𝑛𝑠.
Legume substitution continues until a marginal abatement cost of 125 euros/tCO2eq has been exceeded per geographical area. This upper abatement cost threshold has been arbitrarily chosen, considering the relative abatement cost in other sectors (Vermont and De Cara, 2014)24.
Abatement potentials and cost
At the national level and assuming the variable limit of 100%, the maximum technical abatement of 2,5 million tCO2eq/year is possible for an overall cost of 118 million euros/year (see figure 5. c). This corresponds to an increase of 1,6 Mha of legumes and an average abatement cost of 43 euros/tCO2eq.
The overall cost depends on the volume of emissions reduction. Since displaced crops in each geographical area are ordered by their ratio of gross margin per emission, the lower the abatement targets the lower the overall cost. For example, if the target of emission reduction is reduced by 30%, to 1,7 MtCO2eq, the average abatement cost is reduced by 80% to 14 euros/tCO2eq. If the target is lower than 1,4 MtCO2eq, we find a negative abatement cost, implying that legumes are actually now more profitable than the crop that is displaced .
Reducing the variable limit also reduces the overall abatement potential while increasing the abatement cost. Fixing the limit to either 10% or 90% induces a reduction in the maximum abatement potential of 84% and 8% respectively. We thus observe that results are highly sensitive to this variable. But even if the variable is low, we still observe opportunities to reduce emissions while increasing farm gross margins (see figure 5).
Pellerin et al. (2013) suggests that legume introduction could provide an overall abatement potential of 0,9 MtCO2eq, at a cost of 17 million euros. This implies an average mitigation cost of 19 euros/tCO2eq. That study did not consider how cost varies with area and hence the potential for negative costs. By illustrating those results (the blue curve in Figures 5b and 5c) alongside those derived in this study, it is possible to see that defining a variable limit of 50%, which is the average scenario, and the most realistic, for the same amount of emission abated, we obtain the same overall cost and the same average abatement cost (reached for a marginal abatement cost of 80 euros/tCO2eq).
Heterogeneity of abatement cost between French geographical areas
The spatial allocation of the abatement potential between different geographical areas can be represented for the same marginal abatement cost. Figure 6 shows the departmental shares for the same marginal carbon reduction cost threshold (80 euros/tCO2eq) and a 50% limit to achieve the same reduction estimated by Pellerin et al. (2013). The results show considerable geographical variability, with some accounting for a small amount of the 0,9 MtCO2eq national abatement. These geographical areas are mainly located in the south and eastern parts of France, and represent each less than 1% of these overall reduced emissions. Departments with the highest potential are located in the north-west, where the majority of the geographical areas represent each more than 1% of the national abatement. Note that two regions, Orne and Manche, can each contribute more than 10% of the national abatement.
An alternative representation of the cost heterogeneity is presented in figure 7 for three geographical areas: Orne, Haute-Vienne and Côtes d’Armor. Introducing legumes in Orne is more profitable than in Haute-Vienne or in Côtes d’Armor. In the latter two regions, even for low levels of mitigation the marginal abatement cost is high (respectively 80 euros/tCO2eq and 110 euros/tCO2eq). This cost heterogeneity demonstrates the challenge of setting a uniform nationwide target. If, for example the objective of reducing 50 000 tCO2eq GHG emissions were assigned for the three previously mentioned geographical areas, the overall cost would be high relative to the case of one region (Orne), mitigating 130 000 tCO2eq on its own. As a result, this simulation demonstrates the advantages of policy instruments that account for the cost heterogeneity between regions.
Table of contents :
1.1 Contexte : l’importance des émissions de GES liées à la fertilisation
1.1.1. La fertilisation : une pratique fondamentale pour la sécurité alimentaire mondiale associée à des impacts environnementaux conséquents
1.1.2. Une source importante d’émissions en Europe et en France
1.1.3. Des réductions d’émissions à poursuivre pour atteindre les objectifs d’attenuation en matière de changement climatique
1.2. Des instruments réglementaires à l’idée de l’inclusion d’un prix sur les émissions de GES .
1.3. Revue de la littérature sur l’estimation des coûts d’abattement en agriculture.
1.3.1. Les différentes approaches pour estimer le coût d’atténuation
1.3.2. Le traitement de la fertilisation dans les différentes approches
1.4. Apports de la thèse
2.3. Cost-effectiveness analysis in the literature
2.4.1. Defining emissions and gross margin
2.4.3. Introduction of legumes onto croplands
2.5.1. Abatement potentials and cost
2.5.2. Heterogeneity of abatement cost between French geographical areas
2.5.3. Sensitivity analysis
2.9.1. Area, emissions and gross margin for the main crops in France at the national level in the baseline situation
2.9.2. Impact on legume introduction on other cereals area (for a carbon price of 80 euros/tCO2eq with a limit of 50%)
3.2.1. Data description
3.2.2. Legume preceding crop effect
3.2.3. Rotation Data
3.2.4. Baseline rotations
3.2.5. Building Abatement Cost Curve
3.2.6. Aggregating the 5 regions results following abatement cost efficiency ranking
3.2.7. Economically efficient abatement
3.7.1. Site class descriptions
3.7.2. GM and GHG calculations
3.7.3. Pre-crop effect
3.7.4. Rotation Selection according to the mitigation cost efficiency (in the B N-Decrease Scenario)
3.7.5. Aggregated crop areas in the 5 NUTS 2 regions in “Yield” scenarios
3.7.6. GM and GHG emissions for each rotation in each site class (“Fertilisation” scenario).
3.7.7. Abbreviation of crops
3.7.8. Comparison of the results with results on legumes from European MACC studies
3.7.9. Crop-frequency in the rotations (for crop abbreviations see Appendix 3.7.7)
4.2. Literature review on risk and fertilisation
4.3. Modelling fertilisation application and risk
4.3.1. Economic decision model
4.3.2. Policy instruments
4.4. Empirical Application
4.4.2. Estimation methodology and functional forms
4.4.3. Estimation methodology for risk aversion
4.5.3. Impact of a price on N2O emissions
4.6. Discussion and conclusion
4.8.1. CARA (Constant absolute risk aversion)
4.8.2. CRRA (Constant relative risk aversion)
4.8.3. Illustration of rish aversion impact on fertilizer spreading
4.8.4. Implicit function theorem
4.8.5. Illustration of the impact of insurance program on nitrogen amounts for one crop
4.8.6. Data base treatment
4.8.7. Accuracy of risk aversion scenario with observed emissions.
4.8.8. Impact of risk aversion on crops associated to convex variance functions
4.8.9. Insurance impact on emissions in the CRRA 4 scenario.