The following Q&A with Reza Baresh, amee’s Scoring Expert, explains the methodology we use to score businesses. Reza has over 30 years’ experience with statistical analysis and has developed sophisticated risk analytics and scores while working for Equifax and Dun & Bradstreet.
1. What is the ameeScore and how is it calculated?
The ameeScore is a unique number from 1-100 that indicates an organisation’s environmental efficiency compared to its peers, with 100 as the best-possible score. At present we have focused on efficiency of carbon emissions, defined by an organisation’s annual carbon emissions divided by its annual revenue. We then percentile rank the efficiency metric of an organisation against peers in the same industry of similar size.
The score is freely accessible online and enables anyone to compare the environmental performance of organisations operating in the same (SIC 2007) industry sector. For example, a bank with an ameeScore of 80 is not necessarily more efficient than a concrete manufacturer with an ameeScore of 79. However this does indicate that the bank performs better when compared to its peers than the concrete manufacturer does.
2. Can you summarise how the ameeScore was created?
We took the following steps to create the score:
- We gathered business information
- Collected all publicly reported environmental information
- Collected energy usage information from utilities
- Modelled carbon emissions and annual revenue for those companies which did not report data
- Segmented businesses into peer groups based on industry and size
- Calculated the efficiency metrics of CO2 / revenue
- Ranked companies in each industry and size group
- Divided into percentiles for the score
3. What data sources did you use to develop the model for ameeScore?
We sourced environmental data from the UK’s CRC Energy Efficiency Scheme, corporate sustainability reports on company websites, including their disclosure to the Carbon Disclosure Project (CDP), and other disclosure bodies if available.
We sourced business information from our data partners Dun & Bradstreet and Experian and also Companies House, which provides publicly available data.
4. Only a very small percentage of companies currently report their emissions data. How did you resolve this problem?
We obtained aggregate energy spend data from UK utilities covering over 250,000 businesses in the UK. From this we estimated energy consumed by each business per year.
We then used Defra’s emission factors (available on amee Discover) to give us a figure of carbon emissions in tonnes CO2e. Finally we matched these estimated emission values with actual reported emissions data to ensure that our figures were accurate.
5. How does amee model carbon emissions for those businesses for which there is no reported data?
By working with Experian, Dun & Bradstreet and Companies House we were able to gather vast amounts of data on every business and organisation in the UK, which we stored in our database.
The statistical regression-style model that we have developed makes use of 10 data elements in our database, the most crucial of which is the energy spend data. Other data elements include total assets, years in business, number of employees and headquarter location.
We spent many months running the model to ensure that the relationship between the different input variables represented real emissions as accurately as possible.
6. How do you deal with companies that do not report annual revenue?
We developed a model for revenue using historical data from over 1 million companies that we had collected in our database. We take the average actual revenue of those companies in each of the 40 (SIC 2007) industry sectors, as well as number of employees.
In total the model makes use of 11 data elements in the database. Although these are the same variables used in the model for estimating carbon emissions, the two models are independent.
7. How have you made the model as accurate as possible?
Considering that there are 2.1 million organisations and businesses in the UK we needed to make this more manageable by splitting them into four segments:
Top 3000 Companies defined by revenue and all PLCs
Ltd -Corporate with sales greater or equal to £6.5 Million
Ltd-Corporate with sales less than £6.5 Million
Non-Corporate / Non – Commercial (not for profit)
For each segment the input variables used to estimate carbon emissions and revenue have different values. This ensures that the different sorts of organisation and business in each segment are scored as accurately as possible. So there are multiple models, one for each group.
We then spent several months developing each to ensure accuracy. We are continuing to refine the models as we get more and more real company data.
8. Why did you divide carbon emissions by annual revenue?
Annual revenue is a good approximate indication of how large a company is and therefore is useful in comparing companies of different sizes. Dividing emissions by revenue normalises the score so that organisations can be easily compared, particularly for comparing large organisations with small ones.
9. How does amee address data quality issues with reported revenue?
Dealing with poor and inconsistent data has been one of the greatest challenges in terms of developing the model. We have gone to great lengths to aggregate data from multiple data sources and therefore refine our model so it is accurate.
Of course, we also make this data freely available on our website for businesses to update themselves. As individual profiles are updated with data reported by businesses our model becomes more accurate.
10. When and how does amee update this information?
We are constantly updating the data that we hold about individual companies to maintain accuracy. For example, we receive quarterly updates from Experian and Dun & Bradstreet regarding the utility spend data for over 250,000 companies in the UK. The CRC publishes annual updates, the most recent of which was in February 2013.
We also ensure that any CSR and other online data is incorporated into the model as soon as it is published. Our principle is to use publicly available data – we don’t pay individual companies for it.
11. What about water and waste?
As so few businesses and organisations currently report data about their water usage and waste generation we were unable to accurately incorporate these metrics into our model.
However, we are currently working with utilities and our data partners to develop accurate models for these environmental impacts. These metrics will ensure that the ameeScore provides a more holistic impression about an organisation’s environmental impact.
12. How can a business improve its score?
Fundamentally, businesses improve their score by becoming more efficient in the way they use energy. There are a number of fairly basic energy efficiency improvements that can make genuine cost savings in both the short-term and long-term. Organisations such as the Carbon Trust work with businesses across the country to advise and help implement these improvements.
Completing a sustainability report is another good step as this will help you manage and track your progress.
13. How can the score be used?
The ameeScore and company profile available on our website provides a quick and easy way for businesses to assess their performance, compare with others, learn how to improve, and report to key stakeholders such as customers or investors.
The growing importance of sustainability among supply chain managers has been well-documented. This is not only important in terms of corporate image, but also to create a secure and sustainable supply chain. Thus, updating an ameeScore can help businesses retain existing customers and investors as well as promote and differentiate their organisation for future customers and investment.