Double checking double counting: Quantifying the overlap in input-output-table-based scope 3 emissions

For an introduction to financed emissions based on EEMRIO estimates, refer to part one of this publication series, in which we argue that if financial institutions use EEMRIO tables to calculate financed scope 3 emissions, then both upstream and downstream emissions should be reported. In this publication, we discuss the double counting that results from reporting the financed scope 3 emissions. In particular, we first map out the types of double counting in emissions reports that are acceptable and types that should be avoided according to the Greenhouse Gas Protocol. Next, we show how to quantify undesirable double counting – conceptualized as statistical overlap between sectors – in large portfolios when EEMRIO models are used to estimate supply chain emissions. Finally, in a couple of hypothetical portfolios, we demonstrate with the EXIOBASE EEMRIO database how double-counting percentages vary. In this small simulation, we observe that double counting, as conceptualized here, increases with increasing market share and with stronger trade relations between sectors.

Simply put, double counting means that the reported emissions overlap such that the overall sum is larger than the actual emissions. Emissions estimates can overlap between different company reports, but also within a single report. Whether or not double counting is problematic depends on the context and use cases for the reported emissions. Figure 1 lists five different types of double counting that are acceptable or problematic according to the Greenhouse Gas Protocol. First, the different scopes of a single reporting entity should not overlap – i.e., emissions counted as scope 1 should not be counted again as scope 2 or scope 3 by the same entity. Second, scope 1 and 2 emissions between different reporting entities should be mutually exclusive. This requirement means that two reporting entities should not report the same scope 1 or scope 2 emissions.

Yet, by definition, the scope 1 emissions of a reporting entity are the scope 3 emissions of another reporting entity. Hence, there is an overlap in reported emissions from the macro perspective if the direct emitter reports its scope 1, which are then reported again as the scope 3 emissions of another entity (type 3 in figure 1). Furthermore, the Greenhouse Gas Protocol states that it is natural that separate reporting entities count overlapping scope 3 emissions in their separate reports (type 4). It considers this type of double counting unproblematic because the purpose of scope 3 reporting is to quantify the indirect emissions for which there is joint responsibility – the idea being that both companies are responsible.[1] Finally, there could be literal double counting of the same scope 3 emissions within one reporting entity’s report –i.e., simply counting the same scope 3 emissions twice (type 5).

[1] See also Hertwich, E.G. & Wood, R. (2018). The growing importance of scope 3 greenhouse gas emissions from industry. Environmental Research Letters 13 104013.

The type 5 double counting is of special interest here, as this type of double counting can occur in the financed emissions report of financial institutions. For example, undesired double counting occurs if one financial institution reports the same financed emissions (their scope 3) twice because there are two financed companies whose supply chains overlap.

In this report, we focus on double counting in the financed emissions reported by financial institutions. As detailed in part one of this series, financed emissions (also referred to as scope 3 category 15 emissions) are the scope 1, 2, and 3 emissions of a financial institution’s clients that are attributed to the bank. Financial institutions are the reporting entity for financed emissions, and they typically finance multiple companies within different sectors. Here, we face the special case that the scope 1, 2, and 3 emissions of the financed clients contribute to the financial institution’s scope 3 emissions. Therefore, overlap can occur in the financial institution’s emissions report. This constitutes type 5 double counting as listed in figure 1.

For example, consider a case in which the scope 1 emissions of a food processing company are part of the upstream scope 3 emissions of the supermarket (see figure 2). Both report their emissions separately. However, if a financial institution finances both entities, it will count the emissions of the food processing company as financed scope 1 emissions and as financed scope 3 emissions. Similarly, financed scope 3 emissions could be counted twice if both the farmer and the supermarket are part of the same portfolio (type 5) and both count the emissions of transporting between them (see figure 3). In the context of financed emissions reporting, we refer to this as portfolio-internal double counting.

As explained in part one of this series, financed clients’ scope 3 emissions are often estimated with EEMRIO tables in the absence of better data on emissions attribution throughout supply chains. For large portfolios with substantial market share in different sectors of a supply chain, financed scope 1 and financed scope 3 emissions can overlap, which violates Greenhouse Gas Protocol requirements. Similarly, the financed scope 3 emissions of different clients can overlap. Note, within the framework of an EEMRIO, emissions estimates for companies are based on sector averages, meaning the overlap is not company specific. Hence, we call it statistical overlap.

This report presents a methodology that we developed to quantify the portfolio-internal statistical double counting of financed scope 1, 2, and 3 emissions estimated with an EEMRIO model. This section only presents a qualitative description of the methodology. Readers interested in all mathematical details of our methodology can refer to our technical appendix.

In essence, our methodology subtracts all emissions that are already assigned to the financial institution as scope 3 emissions from all subsequent and preceding supply chain levels. We start with the financed scope 1 emissions, which are then removed from the total emissions that can be allocated as scope 3 emissions in the next step in the supply chain. Next, the scope 3 emissions of the first upstream supply chain step are removed from the first downstream supply chain step. The remainder can then be used to calculate downstream scope 3 emissions. This procedure is repeated at the subsequent supply chain step.

Numerical examples that illustrate how to quantify double counting for a simple hypothetical supply chain are provided in the technical appendix and in figure A.1.

Our double-counting methodology leaves financed scope 1 emissions unmodified. As such, calculating financed scope 1 emissions with a higher-quality data source and financed scope 3 emissions with EEMRIO estimates presents no issues.

We use the EXIOBASE EEMRIO table (v3.8.1, system ixi, year 2022) to showcase how widely portfolio-internal double counting varies, depending on the region, portfolio composition, and market shares of the financed sectors. Financed market share refers to the percentage of outstanding loans over the total value of all companies in that sector. If one bank holds all of the debt of a particular sector that has a debt-over-equity ratio of 1 over 2, then the financed market share is 30%.

Figure 4 shows the percentage of double counting in financed scope 3 emissions as a function of financed market share in the economies of the United States, the Netherlands, and Germany. The financed market share is applied equally to all sectors within the country.

As expected, the percentage of double counting increases with increasing market share. The lower double counting percentages for Germany and especially the Netherlands compared to the United States reflect the fact that international trade comprises a larger share of these economies. Thus, double counting could tend to be higher in portfolios that are concentrated in economies in which international trade has a lower share.

To illustrate the impact of connectivity between the financed sectors of a supply chain, we calculated the double-counting percentage for all the emissions of a hypothetical portfolio with a 10% market share in all 163 Dutch sectors, as well as portfolios with a 10% and 50% market share in the 18 Dutch food processing and agriculture (F&A) sectors. Note that the double counting percentage of 6% in the first scenario – where 10% of all Dutch sectors are financed – is the same as in the third scenario, where 50% of F&A sectors are financed. This means that, despite the fact that the revenue of the latter’s financed market share is smaller by a factor of three, the double counting is the same. This implies that double counting occurs significantly more if the financed activities are concentrated in one strongly connected supply chain, as one would expect.

Our examples based on EXIOBASE imply that the double counting designated as problematic by the Greenhouse Gas Protocol is higher for portfolios with a higher market share or with stronger trade relations between financed sectors. A combination of higher market share and strongly linked sectors, therefore, would likely increase double counting even more.

The method shown here can be used to quantify and report both this type of double counting and scope 3 emissions that were estimated based on EEMRIO models.

The MRIO multipliers for financed scope 3 emissions provided by the Partnership for Carbon Accounting Financials do not give insights into the statistical overlap in attributed emissions. If financial institutions report their financed scope 3 emissions based on PCAF multipliers, this will include portfolio-internal double counting as conceptualized in this article. To guarantee sector-wide alignment, we suggest not correcting for the statistical double-counted emissions. Financial institutions that estimate their financed scope 3 emissions with an EEMRIO table can report the statistical double-counting percentage, as well as the financed scope 3 emissions that include double counting.

Acknowledgements: We express our sincere appreciation to Harry Wilting from PBL Netherlands Environmental Assessment Agency for his thorough review, enlightening discussions, and contribution to this research.

Read the complete appendix section "Appendix: Methodology" as a PDF file.