In the picture above, you see 43 computers from 1950’s.
Before the computer machine, there existed a profession of the same name. A human Computer was a person who carried out calculations according to a certain set of rules. Computers took care of the computing in a company.
In this article I will describe:
A.) The process of a business CO2-footprint calculation
B.) Why the current method is so tedious and un-inspiring.
Emission tracking is crucial and will become even more vital in the coming years. Yet the current method of calculating emissions resembles the profession of a human-computer unmistakably. There is a lot of challenges with the existing method but also a firm footing to build development on.
The main challenge is the extent of repetitive, non-knowledge-intensive tasks. A possible solution to decrease these tasks could be a centralised categorisation of all produced products and services into emission categories.
The process of a simple CO2-footprint calculation for a business
I work weekly with impact assessments and CO2 calculations from various industries. Still I had never calculated one myself until a while ago. As an easily outsourcable service most companies contract the calculation rather than do it in-house. This goes for consultancies and end-clients alike.
A while ago I realized that I had very little knowledge of what the process actually holds. I decided to find a CO2-calculation veteran to mentor me and set out to do my first impact assessment for a client.
Seeing the process as a whole, was like hearing that most berries sold in stores are individually hand-picked by humans. It is a logical way of doing it but didn’t feel like the most efficient one.
To outline the process, the phases for a fairly simple business are:
Why is it a pain?
I spent most of my time doing the calculation on repetitive tasks. Approximately 70 % of the working hours involved were tasks that required little or no domain knowledge such as:
• Moving data from one file to another
• Transforming it to a machine-readable format
• Categorising it
About 20 %, however were very knowledge-intensive and required a lot of experience to reliably carry out. This part of the work requires expertise and would be extremely complex to automate. They were:
• Defining the scope of the calculation
• Assigning emission factors to product categories
• And approximating emissions on which we had little concrete data.
The remaining 10 % were somewhere in between.
Why is this a problem?
Impact assessments are instrumental in understanding and comparing emission sources between companies. Of the S&P 500 companies, 354 were tracking their emissions going into 2020. Emission tracking to guide concrete action is essential. It will become next to necessary that all companies track their emissions.
With a labor-intensive model such as the current one, it is hard to see how the cost involved would meet the resources available.
To bring the cost down, we need both automation and reflection on what the necessary accuracy is. Most often the assessments function as guides to prioritization i.e. the distribution is more relevant than the exact numbers.
What needs to happen?
A big part of the tasks that reminded me of the profession of a human computer went into the following part of the calculation*:
Purchased goods and services
Extraction, production, and transportation of goods and services purchased or acquired by the reporting company in the reporting year.
*We used the commonly used GHG-protocol, this is under Scope 3. More info.
This means cataloguing all products, services and their transportation over the span of a year. All products and all services.
In my case, already the whole sale bought products made up a spreadsheet of 3500 rows consisting of products from 16 different suppliers. This was then row-by-row categorized to 47 different categories and an emission factor (tCO2e/kg) was assigned to each to best represent the category as a whole.
3500 rows are a lot, but I’ve heard horror tales of projects where the length of similar summaries have been 50 000 rows long.
All together gathering the info, combining it to a consistent spreadsheet and categorizing the products to emission classes took over one full work week to complete. All tasks that could to some extent be easily automated.
How could this be solved?
In order to cut a big part of the work involved, a standardised and centralised categorisation of products and services into emission categories is needed.
Individual companies are already doing this in their own operations. One of the two large grocery chains in Finland, K-Ruoka, is already categorising their products under emission factors. The factors collected from research papers are assigned for approximately 30 separate product-categories. As a result, K-ruoka is able to outline the distribution and the general ballpark of the client’s emissions.
Implementing a similar approach to whole industries might not be utopia. Along with the time saved, this would also give producers a greater incentive to report their emissions to ensure accurate categorization. The producers who would not report their emissions would be assigned to a category high-medium as is the convention in CO2-calculations when no reliable data is available. In Finland LUKE for example already provides some data on general product categories that could be used here. EU also has initiatives in order to harmonize the used product-category emissions.
A centralized entity to review and categorize the CO2-assesments could also result in more transparency and unity in reporting.
In terms of total cost, the centralised approach would bring down for individual companies. Upholding and developing such a registry would no question be expensive and require a lot of work to properly set up. It would however direct the costs from individual companies toward a more unified distribution. This might also be a way for governments and other institutions to support and incentivise emission tracking without strict regulation.