Estimation of company product mixes
This page introduces how company product mixes are estimated in the Upright net impact model.
Upright primarily models the impact of companies via their products and services. The approach can be understood in two steps:
Macroeconomic modelling of the global economy to produce an understanding of the impacts of all products and services traded in the global markets
Combining the results of the macroeconomic modelling with detailed information on a company's products and services to produce an estimate of the impact of a company
The second requires detailed information on a company's products and services, including:
Information on which products and services a company produces (e.g. Electricity produced with coal, Electricity produced with offshore wind power)
Information on the scale of each of the products and services within their business (e.g. 70% of revenue from Electricity produced with coal, 30% from Electricity produced with wind power)
At Upright, we refer to this information as a company's product mix. This document provides an overview of Upright's methodology for producing data on companies' product mixes.
Detail: In addition to modelling net impact, company product mixes are also used to produce data on the EU taxonomy, UN SDGs, and some EU SFDR Principal Adverse Impact Indicators.
Overview
Upright seeks to produce all its impact data based on the best available information.
Most companies report their revenue only with a coarse subdivision (or no subdivision at all) that is insufficiently granular to allow for modelling the net impact of a company.
Upright produces estimated revenue breakdowns by activity by using the revenue subdivisions reported by the company as a starting point and refining it based on market statistics, other market research, and quantitative metrics produced from company publications (websites, annual reports, regulatory filings) that relate to the revenue shares being estimated. Additionally, some companies disclose supporting information directly to Upright.
The accuracy of the product mixes is primarily limited by available information. Upright continuously seeks to improve the accuracy of its indicators by using the best available information and statistical methods for integrating information from different sources.
Upright's methodology for producing data on company product mixes is under continuous development. This document describes Upright's current approach, and details may change in the future. Data provided by Upright may have been produced by different versions of this methodology. Upright is taking care to avoid and correct for any systemic biases due to changes in the methodology over time.
Example
The process is best understood in terms of an example. We will base the example on the real-world company Hawaiian Electric Industries, listed on the New York Stock Exchange.
To maximize clarity, we explain the process here in simple terms, as if it was a manual step-by-step process. In practice, most of the work happens automatically within Upright's algorithms.
Step 1: Collection of company-disclosed information
The first step is to collect relevant information disclosed by the company in its annual reports, sustainability reports, 10-K forms (SEC), and the company website.
In its annual report for 2020, Hawaiian Electric Industries reports its revenue as follows:
| Business segment | Revenue, $ | Revenue, % |
|---|---|---|
Electrical utility | 2,265,320 | 87.8% |
Bank | 313,511 | 12.2% |
Other | 944 | 0.0% |
Total | 2,579,775 | 100.0% |
The information is not granular enough for estimating net impact. Moreover, other company reporting does not include more granular business segment revenue figures.
Therefore, Upright combines the business segment revenue figures with other available information to produce a more granular understanding of the company's activities that will be sufficiently specific for understanding net impact.
Upright will perform the granularisation for all business segments, but we will limit the discussion here to only some subsegments that are most relevant for its impact.
From Hawaiian Electric Industries' latest annual report and website, it can be determined that the Electrical utility segment includes the following subsegments:
Electricity production
Electricity retailing
Electricity distribution
Energy storage
Moreover, from the same sources, it can be determined that the company produces its electricity with biofuels, biomass, geothermal, hydro, solar, wind, coal and crude oil, and that it is retailing its own electricity.
While the company does not disclose revenue shares for the different subsegments of the Electrical utility business segment or for different means of electricity production, it does disclose the amount of electricity produced by each means of electricity production.
Step 2: Estimation of granular revenue shares
Given the discussed information on Hawaiian Electric industries, Upright uses global market sizes to split revenue among the main segments within its Electrical utility business segment, and uses disclosed figures on the amount of electricity generated as a proxy for splitting revenue between different means of electricity production employed by the company.
This results in the following revenue shares for different means of electricity production within the Electricity production subsegment:
| Subsubsegment | Revenue share, of total |
|---|---|
Electricity produced with biofuels | 0.6% |
Electricity produced with biomass | 2.6% |
Electricity produced with geothermal | 0.1% |
Electricity produced with hydro | 0.2% |
Electricity produced with solar | 12.3% |
Electricity produced with wind | 4.2% |
Electricity produced with coal | 8.4% |
Electricity produced with crude oil | 41.7% |
A similar granularization is performed also for other subsegments of the company. Discussion of the granularization of other subsegments is omitted for brevity.
The means of estimating revenue shares that Upright uses depend on two factors:
the line of business
what information is disclosed by the company
In this case, a combination of global market sizes and a proxy based on quantitative company-disclosed data was used. Other used proxies include quantitative metrics produced by NLP from company publications (websites, annual reports, regulatory filings) that relate to the revenue shares being estimated.
Upright generally produces more granular product mappings than the ones listed above. The example has been simplified for maximum clarity.
Last updated