From the Paris Agreement to corporate climate commitments: evaluation of seven methods for setting 'science-based' emission targets

We constructed a simplified scenario of global production, composed of two unspecified regions, two unspecified sectors and eight archetypical companies. We defined the eight companies by assigning them all combinations of two arbitrarily chosen values for three variables: (a) region and sector; (b) baseline emission intensity; and (c) projected annual activity growth, as illustrated in the left panel of figure 2. All companies have the same baseline year activity measured in value added.

Zoom In Zoom Out Reset image size

We developed a common global scenario, assuming a sigmoid-shaped emission pathway leading to a 78% net ⁷ reduction of 2020 emissions in 2050 (illustrated in right panel of figure 2). The global emission scenario corresponds to cumulative emissions in the 2020–2050 period of 921 Gton CO₂-eq and is broadly consistent with a typical '1.5 °C high overshoot' or 'lower 2 °C' scenario (Rogelj et al 2018, p 117). Within the scenario, we assumed that region #1 and sector #1 will reduce emissions at a steeper pace than region #2 and sector #2, so that 90% of emissions in 2050 will take place in region #2 and sector #2, versus 50% in 2020. Our emission scenario includes emissions of all GHGs from all activities with an economic output (such as industry and agriculture) and excludes direct emissions from final consumption by households and governments. These excluded emissions are largely related to combustion of fuels by individuals and account for roughly 12% of the current global total ⁸ . Hence, when referring to 'global allowable emissions', we imply that an unspecified pre-allocation to final consumption activities already has taken place. In both regions and sectors and at the global level, we assumed a 3% annual increase in real gross domestic product (GDP) during the 2020–2050 period. The parameter values in our scenarios' baseline year (2020) for emissions (50 Gton CO₂-eq) and GDP (8 × 10⁷ million US$) were inspired by UNEP (2020) and the World Bank (2020). The projected annual increase in real GDP (3%) is generally consistent with the scenarios of the International Energy Agency (2017).

We calculated SBTs specified as percentage reductions in baseline (2020) absolute emissions for each company's scope 1 activities for the period 2020–2050 using each method, except for The 3% Solution due to its restrictive focus on the United States and 2020 target year. In our application of the SDA method, we assumed the companies' projected growth in physical production to be proportional to their projected growth in real value added. Since that method assumes the power sector will reduce absolute emissions at a much steeper rate than other sectors (Krabbe et al 2015), we considered sector #1 and #2 rough approximations of the power sector and a mix of other sectors, respectively. In our application of the BT-CSI and C-FACT methods, we considered region #1 and #2 rough approximations of developed and developing countries, respectively, since these methods assume steeper emission reductions in developed countries (Tuppen 2008, Stewart and Deodhar 2009).

To evaluate the robustness of our results, we developed eight sensitivity analysis scenarios involving combinations of smaller and larger differences between the variable values of the archetypical companies (left panel of figure 2, in red) and the use of linear and exponential global emission pathways (right panel of figure 2, in red). In all cases, the combined company variables are consistent with the global scenario. For example, the sum of company value added is equal to GDP in all years. The linear and exponential global emission pathways also lead to a 78% emission reduction in 2050, but involve slightly different cumulative emissions than the sigmoid shape (943 and 799 vs. 921 Gton CO₂-eq).

For each set of method-specific SBTs, we calculated the yearly (y) emission imbalance (EI):

$$\begin{equation}{\text{E}}{{\text{I}}_{\text{y}}} = {\text{G}}{{\text{E}}_{\text{y}}} - {\text{ }}\sum\limits_{{\text{c}}1}^{{\text{c}}8} {{\text{SB}}{{\text{T}}_{{\text{y}},{\text{c}}}}.} \end{equation} \tag{ 1 }$$

GE is global allowable emissions prescribed by the common global emission scenario, c1 to c8 refer to the eight companies, and SBT refers to the calculated emission target. A positive emission imbalance means there are 'leftover' (or unused) global allowable emissions after allocation of emissions to companies for a given year. A negative emission imbalance means that global allowable emissions are exceeded by the sum of company SBTs for a given year. Further, for each SBT method we calculated a cumulative emission imbalance ratio (CEIR) for the 2020–2050 period:

$$\begin{equation}{\text{CEIR}} = \frac{{\sum\nolimits_{2020}^{2050} {{\text{E}}{{\text{I}}_{\text{y}}}} }}{{\sum\nolimits_{2020}^{2050} {{\text{G}}{{\text{E}}_{\text{y}}}} }}.\end{equation} \tag{ 2 }$$

The ratio shows the importance and sign of the cumulative emission imbalance relative to global allowable emissions for the time-integrated period.

Next, we performed the emission imbalance calculations (equations (1) and (2)) for three hypothetical mixes of methods: (a) selection of the method giving the highest cumulative SBT ('least challenging') for each company; (b) selection of the method giving the lowest cumulative SBT ('most challenging') for each company; and (c) random assignment of methods for each company with ten iterations. All calculations are documented in the supplementary spreadsheet (available online at stacks.iop.org/ERL/16/054019/mmedia).

The ACA method is the simplest, requiring that each company reduces its emissions at the same annual reduction rate required globally to meet an established temperature goal. This reflects the Grandfathering allocation principle, i.e. past emissions being 'grandfathered' into future emission allowances (Knight 2014). The ACA method exists in three versions, reflecting different emission pathways presented by the Intergovernmental Panel on Climate Change (IPCC) and related to different temperature goal classifications ('2 °C', 'Well-below 2 °C' and '1.5 °C'). Application of the ACA method (figure 3(a)) results in a linear emission pathway from baseline year to target year, identical for all eight archetypical companies.

In addition to Grandfathering, the GEVA method relies on the Economic contribution principle, which holds that a company's absolute decrease in emission should be smaller the more it is expected to increase its value added. The method exists in two versions that apply different emission pathways. Application of the GEVA method (figure 3(b)) results in exponential emission pathways from baseline year to target year. The effect of the Economic contribution principle can be observed by the two clusters of graphs that separate the four companies with low (2% per year) and high (4% per year) projected growth rates.

The BT-CSI method is similar to the GEVA method, but uses separate emission pathways for developed and developing countries. Accordingly, companies in developed countries are expected to reduce emissions more than those in developing countries. This appears consistent with the Responsibility, Right to development or Capability principles (EEA/FOEN 2020, van den Berg et al 2020). Application of the BT-CSI method (figure 3(c)) yields exponential SBT pathways, differentiated according to projected growth rates, as with the GEVA method (figure 3(b)), and according to whether companies are located in developed countries (region #1) or developing countries (region #2), which is a stronger differentiator for this method.

The C-FACT method relies on the same sharing principles as the BT-CSI method and exists in a single version, applying an IPCC scenario consistent with the '2 °C' target classification. However, C-FACT assumes that all companies in a region will reduce absolute emissions by the same degree between the baseline year and a fixed future year (2050 per default). Hence, application of C-FACT (figure 3(d)) leads to exponential SBT pathways that are mainly differentiated based on whether companies are located in developed countries (region #1) or developing countries (region #2).

The CSO method relies on the same sharing principles as BT-CSI and C-FACT, but considers every year of a global emission scenario, instead of just the baseline and target years. The three versions of CSO apply different global emission pathways from IPCC, including one that corresponds to below 1.5 °C warming. In the more recent method versions, the emission pathways do not explicitly differentiate between developed and developing countries (as this is not done in the underlying data source, see table 2). Like GEVA and BT-CSI, application of CSO (figure 3(e)) leads to a differentiation of companies with low and high growth projections. The SBT pathways are sigmoid shaped, reflecting the common global emission pathway applied (figure 2, right panel). The CSO tool (CSO 2020) allows for adjustment of SBTs if the company does not follow its original SBT pathway, so that the originally calculated cumulative SBT remains fixed. Note that we did not differentiate between regions in our application of the CSO method, to be consistent with its most recent version.

The 3% Solution differentiates between the industrial sectors of companies as per the Cost optimization principle (van den Berg et al 2020). This principle holds that the emission reduction burden should be distributed in such a way between sectors that the total cost is minimized. This criterion is reflected in the methods' sector-specific emission pathways (detailed in table 2). These emission pathways are specific to the United States and consistent with the Responsibility principle (or Right to development or Capability, see above). The 3% Solution also relies on the Grandfathering principle and the Physical production principle. The latter is similar to Economic contribution in considering expected activity growth, but it uses sector-specific physical indicators of production instead of value added as activity measure (Bjørn et al 2020). The methods' single version is restricted to the target year 2020 and, hence, no longer operational.

The SDA method builds on The 3% Solution, but is applicable to companies in any country and does not rely on the Responsibility principle (or Right to development or Capability, see above). Instead, SDA relies on the Convergence principle (Meyer 2000) ¹⁰ , as it assumes that all companies in a sector will converge towards a common emission intensity in 2050. SDA is also the only method to calculate targets for scope 1 and scope 2 emissions differently. The approach to calculating scope 2 emission targets reflects both the projected decrease in emission intensity of power production and the projected changes in the sectoral power need per unit of activity. For companies that are 'heterogeneous' (not producing a homogenous output), the SDA method simplifies to pure emission grandfathering of absolute sectoral emissions. The two versions of SDA reflect different sectoral emission pathways, both developed by the International Energy Agency, and are consistent with the '2 °C' and 'Well-below 2 °C' target classification, respectively. Like CSO, the SDA method considers every year of these emission pathways. Application of SDA (figure 3(f)) leads to sigmoid-shaped SBT pathways that diverge to a much greater extent than for the other methods. This is due to the sectoral differentiation and differences in baseline emissions intensities and projected growth rates. Hence, the 2050 SBT for company 1Ig (power sector, high baseline emission intensity and low growth—see figure 2, left part) is a 98% reduction in baseline emissions, versus the 2050 SBT for company 2iG* ('other sectors', low baseline year emissions intensity and high growth) of 14% increase in baseline emissions. Note that despite the SDA method's separate target calculations for scope 1 and 2 emissions and for 'homogenous' and 'heterogeneous' sectors, our calculated SBTs (figure 3(f)) are only for the scope 1 emissions of 'homogenous' companies.

None of the six SBT methods have a perfect balance between the sum of SBTs and global allowable emissions when applied to our simplified global production scenario (figures 3(a)–(f)). GEVA, BT-CSI, C-FACT and SDA have positive imbalances throughout the 2020–2050 period, meaning there are 'leftover' global allowable emissions after allocation of emissions to the eight companies (equation (1)). CSO has a small negative imbalance and ACA has a positive imbalance up until 2030 and a larger negative imbalance in the remaining part. Considering the entire 2020–2050 period, the cumulative emission imbalance ratios (CEIR in equation (2) vary from −2.4% (ACA) to 24.7% (BT-CSI) (figure 3(g)). As briefly explained in table 3, the emission imbalances are all caused by assumptions and approximations within the methods' target equations (further details in SM 1). The two most common sources of imbalance are implicitly assuming that the global emission pathway has a specific shape (linear for ACA and exponential for GEVA, BT-CSI and C-FACT) and approximating the rate by which each company must reduce their emission intensity with the globally required rate (GEVA and BT-CSI).

The emission imbalance ratios show substantial sensitivity to the shape of the global emission pathway for the four methods that implicitly assume a specific shape, with the ratios being closest to zero for ACA using the linear pathway and for GEVA, BT-CSI and C-FACT using the exponential pathway (figure 3(g)) ¹¹ . For these four methods, the linear pathway is generally ¹² associated with highest positive imbalance, followed by the sigmoid and exponential pathways. This is due to differences between their cumulative global emissions (943, 921 and 799 Gton CO₂-eq, respectively). The emission imbalance ratios also show varying degrees of sensitive to the gap between company variables (indicated by 'smaller', 'medium' and 'larger' in figure 3(g)) for all methods, except ACA (not requiring any company variables). A larger gap generally ¹³ leads to a lower positive emission imbalance. The CSO method, followed by the SDA method, has the overall lowest emission imbalance across all scenarios (figure 3(g)).

The SBT method that leads to the least challenging target differs across the eight archetypical companies (see legend in figure 4(a)). If each company was to choose the least challenging SBT, the resulting emissions reductions would fall substantially short of global allowable emissions between 2025 and 2050 (figure 4(a)), yielding a cumulative emission imbalance ratio of −16.3% (figure 4(d)). BT-CSI, C-FACT and GEVA are associated with the most challenging targets across the eight companies (see legend in figure 4(b)). Selection of these challenging targets would lead to a 'leftover' of global allowable emissions, yielding a cumulative emission imbalance ratio of 32.2% (figure 4(d)). In the first iteration, the random application of methods lead to a positive imbalance up until 2033, followed by a slightly smaller negative imbalance in the remaining period (figure 4(c)). The ten iterations of random method applications spanned a cumulative emission imbalance ratio of −1.7% to 25.0%, averaging 10.6% (figure 4(d)). The frequent positive emission imbalance associated with random method applications reflects that most methods have a positive emission imbalance (see figure 3).

Zoom In Zoom Out Reset image size

As was the case for most individual methods (figure 3(g)), the emission imbalance ratios of the three method mixes are generally sensitive to the shape of the global emission pathway and the gap between company variables (figure 4(d)). However, the sign of the emission imbalance ratio is consistent across the sensitivity scenarios, except for the random method mix.

Currently, the SBTi presents the ACA and SDA methods separately, but groups the GEVA, BT-CSI, C-FACT and CSO methods together under terms like 'economic intensity target-setting methods' (SBTi 2020c). Given our findings of non-trivial differences between the characteristics of these four methods (table 2) and the resulting SBT pathways (figure 3), they should be presented separately. In the future, the SBTi could use our framework (figure 1) to characterize new SBT methods. Currently, the initiative makes incomplete references to allocation principles. For example, the SBTi correctly identifies the SDA method as being based on Convergence, but omits that it is also based on Grandfathering, Physical production and Cost-optimization (see table 2). We encourage the initiative to be explicit about the allocation principles involved in each method, as these represent a central normative aspect of target-setting.

The SBTi currently recommends ACA and SDA over the other methods (SBTi 2020c). As mentioned (section 1), the reasoning behind this recommendation is not entirely clear, but emission imbalance appears to play a role. However, our results indicate that concerns over emission imbalance should favour the CSO and SDA methods, rather than ACA and SDA (figure 3(g)). Of course, there may be other valid reasons favouring the SDA and ACA methods over CSO. For example, the SBTi cautions against using measures of economic emission intensity (which CSO relies on) in sectors characterized by volatile prices (SBTi 2020c). On the other hand, the SDA method has more application constraints than ACA and CSO (table 2), especially considering that the SBTi no longer recommends applying SDA on heterogeneous companies (SBTi 2020c). Strictly speaking, the SDA method is also currently not applicable with 'the latest climate science', since it relies on sectoral emission scenarios predating the IPCC's Special Report on Global Warming of 1.5 °C (IPCC 2018) ¹⁵ . Given these pros and cons of individual methods, the SBTi should be transparent about the reasons underlying its method recommendations.

We welcome SBTi's recommendation that 'a company should screen several of the methods and choose the method and target that best drives emissions reductions' (SBTi 2020c). This would reduce the risk of a global emission overshoot, such as observed in our 'least challenging targets' scenario (figure 4(a)). We also agree with the SBTi that companies should describe 'the assumptions and methods used to set the target' (SBTi 2020c). However, most companies currently only present their SBT and not how it was calculated ¹⁶ . This lack of transparency makes references to a 'scientific basis' appear somewhat undeserved and effectively hides underlying value judgements from stakeholders. We therefore encourage the SBTi to make target documentation a requirement. Reasons for withholding information, e.g. if a company considers activity projections sensitive, should be stated.