Introduction
Diagnosis
Emissions related to sold services are currently accounted [1]as follows:
Emissions related to financial services (investments) are assessed in category 15 of the GHG Protocol.
Emissions related to other non-financial sold services (digital services, consulting, auditing, any type of intellectual service, etc.) are not evaluated.
Corporate carbon accounting standards include an emissions category dedicated to the use of sold products (for instance, category 3.11 Use of Sold Products of the GHG Protocol).
One might think that this category includes the emissions related to the use of sold services. Indeed, although this category 3.11 of the GHG Protocol is entitled ‘Use of Sold Products’, it is written at the top of the dedicated section of the scope 3 calculation guidance that this category encompasses the emissions of both products and services.
However, surprisingly, there is no explanation in this section on the way emissions of sold services should be accounted: emissions of sold services simply are not accounted.
Why it matters
Whatever the goal of corporate carbon accounting is (climate risks analysis, GHG inventory, etc.), this omission is very problematic.
Indeed, many economies are service economies -namely western economies, and this tertiarization is growing and will likely keep growing in the future as digital firms expand.
It would be misleading to claim that services have no environmental impact.
Claiming that would be equivalent to saying that these sold services are without any impact at all, or in other words are … useless. This obviously does not make sense in money-driven economies: why would a client purchase a service if it does not lead to any impact?
The lack of carbon accounting of sold services is both theoretical (there is no explicit recommendation or guide on such carbon accounting in the standards) and a fortiori practical (one can simply check public carbon reporting of companies such as Google, Facebook, etc.).
Some orders of magnitude
Here are below some insightful examples of the problems linked to the lack of carbon accounting of sold services:
90% of Google[2]/Alphabet's revenues are advertising revenues, which amounts in order of magnitude more than €50 billion in revenues.
There are no CO2 emissions associated with these revenues (and therefore Google's carbon neutrality commitment is without effect on these advertising revenues).
Another example: an artificial intelligence sold by Google and which would serve the oil & gas industry is not penalized whatsoever in the carbon footprint of Google.
The same exact analysis can be made for Facebook or any other social media.
Even Microsoft (who aims at being carbon negative) cloud activities and AI services do not lead to any CO2 emission in their carbon reporting, according to current carbon accounting standards.
These few examples highlight how much accounting for the emissions of sold services would be critical for digital firms.
But on closer inspection, many other industries also do not "see" much of "their" CO2 emissions either. One can mention every form of service firms (consulting, engineering, etc.), communication and marketing, various advisory activities, etc.
For instance, many management consulting firms, sometimes weighing several billion euros in revenue or market capitalization, also claim carbon neutrality objectives even though the carbon accounting used does not take into account the emissions induced by their sold services.
From a climate change mitigation perspective, it seems that there is an urgent need to propose a robust methodology for accounting the emissions of sold services.
Methodology
Reminder
The methodology we propose introduces a distinction between two types of services, depending on whether the sold service directly relates to a finished product or is an intermediate service that is part of the value chain of a product or a company.
This distinction (finished/intermediate service) is aligned with the one introduced in the carbon accounting of emissions linked to the use of sold products.
As a reminder, when a company sells a finished product, whatever the added value of its activity (e.g. even if it is only a purchase for resale of the product), the company must account for all emissions related to the use of this finished product (in addition to the upstream emissions of the product - manufacturing, freight, etc., that will be accounted in other categories).
On the other hand, if a company sells a so-called intermediate product (which will not be used as such by the end user but is part of the value chain of a finished product), then the company must in this case allocate itself a share of the total emissions linked to the use of the finished product, for example in proportion to the monetary weight of the intermediate product sold in the finished product.
Concretely, for a finished product (e.g. car), there is no allocation of emissions (the car seller accounts for all emissions related to the car, regardless of its value added in the value chain of that car); regarding an intermediate product (e.g. wheels, engine, etc.), there is an allocation of emissions from the finished product (the car).
Let us now disclose the methodology for services.
1st case: service at the end of the value chain related to the consumption of a finished product.
In this case, and similarly to what is done in the accounting of sold products, the totality of the emissions linked to the sold services is accounted for.
Let us quote some examples of companies (service sellers) included in this 1st case: a travel agency, a car-pooling platform, a rental service or in general any company in the economy of functionality dealing with finished products.
Digital application
For example, a carpooling platform should account for the emissions associated with all the trips made via this platform, by multiplying the distances traveled during the trips made via the platform by the emission factor of the vehicles supporting these trips (emission factors in gCO2e/km, typically integrating the emissions amortized from the upstream energy and manufacturing processes of the car, and the emissions related to the use of the vehicle).
If 100 million kilometers of car trips were made via the carpooling platform during the reporting year, in cars emitting on average 200 gCO2e/km (over the life cycle of the car), then the carpooling platform accounts for 20,000 tCO2e as emissions related to the services it sells.
N.B.: it is possible that emissions from such services are already accounted for by current standards, since the methodology mentioned above is quite intuitive - often based on easily identifiable physical flows - and close to what is done for sold products. However, as the standards are not clear on this subject, as was pointed out in the preamble, this section could at least serve as a clarification.
2nd case: intermediate service in the value chain of a product, project, or company.
Compared to the previous case, this time the sold service is an intermediate service: the service is part of the value chain of a product or company but does not directly relate to the consumption of a finished product.
Two intermediate scenarios can be distinguished.
Scenario°1: the final product or project can be identified.
To clarify the type of services belonging to this category, let us take a few examples: it can be an advertisement for a specific product or project, consulting and services for the development of a specific product or project (e.g. from a design office in the automotive sector), etc.
In this case, we suggest evaluating the emissions of sold services by multiplying the carbon intensity of the product (kgCO2e/€ - price of the product sold) by the monetary value of the sold service, which is equivalent to allocating the emissions over the life cycle of the project or product to the company under study, in proportion to its financial weight in the product/project.
At this stage, there are at least two possibilities for assessing the carbon intensity of a sold product: firstly, by referring to the total cost of the product/project (obtained by adding up the various costs - capital, operational, etc. - of the cost structure), or secondly, by referring to the market price of the product.
Referring to the market price is indeed simpler (market data are generally transparent) but exposes carbon accounting to market price volatility, which is sometimes high, particularly for commodities traded on financialized markets.
Finally, other allocation keys could be used (possibly other than monetary), and should, if need be, be explained in a transparent manner and used universally by all the players in the product, project or sector of activity concerned.
Digital Applications
To illustrate this first scenario, let us take several examples.
Let us start with a simple first example, that of advertising revenues.
A social network earns 500k euros in advertising revenue from the BMW X3 model.
How should the social network's emissions be evaluated?
The social network should start by evaluating the carbon intensity of such a car model by dividing the emissions over the life cycle of a BMW X3 (manufacturing, upstream energy, use over the theoretical lifetime) by its average market price.
Then, the emissions allocated to the social network would be obtained by multiplying this carbon intensity by the value of the advertising service sold by this same social network.
A BMW X3 model emits over its life cycle and in real conditions of use around 280 gCO2e/km (approximate value, the details of the calculations are not specified here).
Assuming 10 years of use and 17,000 km traveled per year and considering an average purchase price of 45 k€, we obtain a carbon intensity of 1.05 kgCO2e/€ for a BMW X3. The emissions allocated to the social network linked to the advertising revenues targeted on the BMW X3 models therefore equal 529 tons of CO2e.
N.B.: As a first approximation, if we generalize and consider an average carbon intensity of 0.5 kgCO2e/€ (close to the global carbon intensity obtained by dividing global GDP by global CO2 emissions), then Google's advertising revenues, amounting to close to €50 billion, would lead to emissions of around 25 MtCO2e. Google has reported (CDP 2019) just over 14 MtCO2e in its Scope 3 (of which no emissions related to sold services, and the majority coming from the 3.15 investment category).
The second, more complex example is that of artificial intelligence sold by a digital player and used to exploit an oil field.
How should emissions from the sale of this service (artificial intelligence) be assessed?
By following the proposed methodology, the digital player could therefore proceed in different ways.
In a privileged way, the first possibility would be to start by estimating the lifecycle emissions of the volume of oil about to be exploited during the use phase of the sold service.
This would start by including upstream emissions from energy, transport, etc. (standard factors from ADEME or DEFRA exist for these oil emissions items and could be used), and of course also emissions related to the use of oil extracted from the subsoil.
Then, the emissions previously calculated would be divided by the total cost of the operating project (investment costs, variables, etc.) to get the carbon intensity of the project (e.g kgCO2e/€).
This carbon intensity would eventually be multiplied by the monetary value of the sold service.
If this first possibility is not appropriate (for instance because the firm can’t access the project's financial data), the company could then multiply the monetary value of its sold service by the carbon intensity of the oil exploited, which could be calculated in a generic way by making the ratio of the emissions of a given volume of oil (e.g. emissions of a barrel of oil over its life cycle) to the price of a barrel of oil (ideally the production cost rather than the market price, to avoid carbon accounting being affected by oil price volatility on financial markets).
For example, if the cost of the extracted barrel is $50/barrel, knowing that a barrel of oil (159 liters) emits 400 kgCO2e in order of magnitude, then the carbon intensity is 8 kgCO2e/$.
Thus, if the service sold by the digital player is worth $10 million, the emissions of the sold service are estimated at 80,000 tCO2e.
Scenario°2: the final product cannot be identified
In this second scenario, the sold service does not serve a specific project or product but serves in a more general and undifferentiated way the company purchasing the service.
Here are some examples that could be included in this second scenario: organizational/management consulting, generic communication/press relations services for a company, a financial auditing service, etc.
The proposed methodology is as follows: emissions are calculated by multiplying the carbon intensity of the sales of the company purchasing the service by the monetary value of the service sold.
The company's carbon intensity can be assessed like so:
Based on the carbon intensity of a basket of products sold by the company purchasing the service, if these different products can be identified (in which case we can refer to the 1st scenario to estimate the carbon intensity of a product individually; then the different carbon intensities must be weighted appropriately),
If the client purchasing the service is itself a company that sells services, then its own carbon intensity of sold services may be retrieved (there are several possibilities to do so: either the client calculates it following the methodology presented above and then communicates it, or it is recalculated, for example, by doing the ratio of the emissions of category 3.11 Use of products and services sold to the turnover of the client) and is multiplied by the monetary value of the service sold.
If the client sells a mix of products and services, then both methods presented above are applied to obtain a weighted carbon intensity.
Digital Applications
Let us take the example of a service sold by a consulting firm to a luxury actor. This service is worth 100 k€ and is generic (it deals with strategic advisory for example), i.e. it does not relate to a particular product or project.
This luxury player sells a range of leather goods (50% of turnover, cradle-to-gate emission factor of 0.3 kgCO2e/€), ready-to-wear (25% of turnover, cradle-to-gate emission factor of 0.1 kgCO2e/€) and accessories (25% of turnover, cradle-to-gate emission factor of 0.1 kgCO2e/€).
The weighted average carbon intensity of the products sold by the luxury actor is 0.2 kgCO2e/€, the emissions allocated to the service seller are therefore 20 tCO2e.
Let us take a second example, that of a consulting firm that sells a service worth 50 k€ to a digital player (which is therefore another company selling services).
This digital player has evaluated the carbon intensity of its sold services at 1 kgCO2e/€. (following the above methodology). The emissions to be accounted for by the consulting firm are evaluated at 50 tCO2e.
Summary of the methodology
Conclusion
The proposed methodology for accounting emissions of sold services is aligned with the methodology for accounting emissions of sold products.
In particular, the methodological distinction between intermediate products and finished products remains.
The proposed methodology does not in any way revolutionize the carbon accounting of the standards and could therefore be easily plugged into existing accounting frameworks, or even be interpreted as a guide or clarification of the current methodology (which mention the carbon accounting of sold services without specifying how to proceed).
Strengths of the proposed methodology
The climate risks of targeted companies are better identified
The proposed methodology allows a better assessment (and therefore management) of the climate risks of companies involved in the sale of services (digital companies, consulting, service companies, etc.).
This is favorable both for the firms themselves (which would be better equipped to navigate in a carbon-constrained world) and for the stakeholders of these companies (customers, financiers, etc.) who until now were offered a limited understanding of the climate risk exposure of these firms.
The methodology incentivizes climate-friendly actions
The two common leverages for reducing emissions can be activated and have an impact on the accounted emissions of sold services:
Carbon efficiency (or engagement): the service seller can commit with its clients and encourage them to reduce the carbon intensity of their activity.
This leverage can influence the induced emissions and/or the avoided emissions of the company whose emissions are being accounted for.
For example, an advertiser may choose to highlight only the most carbon-efficient products of its clients and commit to supporting its client in its own decarbonization. Other example: a consulting firm could increase its services to support companies in their transition.
Sobriety (or client selection): reorienting the business model towards the least carbon-intensive clients. A firm that sells services is to a certain extent free to choose with whom (and in what proportion) to do business, likewise a financial company might divest from its most carbon-intensive clients to reduce its accounted emissions.
Possible criticisms
Is it too late to change corporate carbon accounting methodologies?
In our opinion, not:
The fight against climate change has barely started: the decarbonization of our economies has not yet really begun - global emissions are still on a rising trend- and will take place at least throughout the whole 21st century. Especially businesses are only at the very beginning of their decarbonization pathway -for a couple of them, they are mostly at the stage of communication and declarations of intent.
The problem (lack of carbon accounting of sold services) is likely to become more acute with the growing emergence of the service economy.
Failure to fill this accounting "gap" now would expose ourselves to even greater backlash and deadlocks soon.
There is a potentially very strong knock-on effect: as part of their climate strategy, companies selling services would be incentivized to engage with their clients (helping them to reduce their carbon footprint and increase their avoided emissions) and/or select their clients to reduce "their" emissions. There is potentially a very strong leverage effect if companies as dominant as Google or Facebook or firms as influential as McKinsey or BCG were to act in this direction.
Some companies have already set targets based on current carbon accounting, is it possible to go backwards?
Indeed, several companies (some of which have already been mentioned in this note) have already set emission reduction targets calculated according to current standards.
A change in these standards because of the integration of the carbon accounting of sold services would in fact lead to an increase in the emissions of some of these companies (e.g. Google, Microsoft, etc.). This upward trend could in fact sometimes be spectacular.
Is this a good enough reason not to change the carbon accounting methodological framework? There is certainly a cost associated with this methodological change, but in order to arbitrate, we need to think not only in terms of costs but rather in terms of cost-benefit balance and judge whether or not the expected benefits outweigh the costs.
In our view, benefits do indeed outweigh costs, for several reasons:
As has already been said, the fight against climate change has only just begun (it will continue for a very long time): from a time perspective, it is certainly not too late to improve the methodology/theory of carbon accounting,
The commitments made by companies on emissions already accounted for would not necessarily be called into question by the suggested methodological evolution. Nevertheless, companies would also have to commit to "its" emissions from their sold services, in addition to its other emissions. On the other hand, such companies would indeed have to review their carbon communication - for example, Google would no longer be able to claim carbon neutrality without somehow addressing the emissions linked to its advertising revenues.
Wouldn't companies that sell climate-friendly services be penalized?
One argument that could be countered against this methodology deals with the allocation of emissions to companies whose activities contribute to the low-carbon transition (consulting firms, engineering firms, etc.). We can provide several pieces of response to this:
First, the same remark could be made regarding the already existing methodology of carbon accounting of sold products. The induced emissions of companies selling products that contribute to the transition are accounted for (e.g. NextEra Energy, Tesla, etc.).
Being a supplier means contributing to the success of its activity - and thus contributing to the emissions of this same company: it does make sense to account some emissions for the firm selling services.
More importantly, the key point is that a company can, in addition to its induced emissions (which are the subject of this note), account for its avoided (or reduced) emissions related to the same sold services, which highlight its contribution to the low-carbon transition. The proposed methodology could be easily duplicated for the calculation of avoided emissions allowed by the sale of services.
Is there a risk of double counting emissions within the same company? What if a company sells both products and services?
The goal here is to account for the emissions induced by the sold service (and not the emissions physically necessary to perform the service - e.g. the electricity needed for an IT service, which are already accounted for under current standards).
There is therefore no double counting in this respect, but rather a broadening of the scope of the assessment reflecting the impact of the service sold in the downstream value chain.
[1] in corporate carbon accounting standards, such as GHG Protocol, ISO 14064, Bilan Carbone®, etc. [2] which also claims to be carbon neutral