New Metrics

It's All Relative:
The Fatal Flaw in Corporate Sustainability Ratings

Corporate sustainability ratings are of growing interest to business leaders, investors, and the general public. The annual releases of popular ratings such as the Dow Jones Sustainability Index and the Global 100 Most Sustainable Corporations in the World Index now attract widespread coverage. A key driver of this interest is the belief that these ratings schemes allow us to distinguish sustainable corporations from those that are not.

By Cory Searcy | Published 8 years ago | About a 5 minute read

Sustainability ratings evaluate companies on the basis of key performance indicators (KPIs) that address different economic, environmental, and social issues. All ratings do this a little bit differently. However, the most popular ratings all share the key limitation of not connecting their KPIs to the broader sustainability context in which companies operate. This is a fatal flaw. Measuring corporate sustainability requires an explicit connection between a company’s performance and the ability of nature and society to indefinitely support its activities.

Most existing KPIs in sustainability ratings use absolute or relative measurements. An example of an absolute KPI is greenhouse gas (GHG) emissions per year. An example of a relative KPI is sales divided by GHG emissions per year. These are useful measures, but they do not explicitly tell us whether the level of performance is sustainable or not. In the case of these examples, what is missing is the link to the ability of the planet to support the company’s emissions of GHGs.

To construct a corporate sustainability rating, we need to:

identify the KPIs used in the index;
link those KPIs to the broader context;
normalize the data to avoid comparing apples and oranges;
determine the relative weights of the KPIs; and
combine the individual KPIs to form the overall rating.

Clearly, all of this needs to be supported by high-quality data. As an illustrative example, let’s consider the development of a sustainability rating with two KPIs, one environmental and one social.

Assume that the environmental KPI is focused on GHG emissions. Research by Jorgen Randers suggests that corporations must reduce their “GHG emissions per unit of value added” by 5 percent per year until 2050 to prevent a rise in global mean temperature above 2oC, relative to pre-industrial levels. The science-based target of 5 percent annual reductions ties the KPI to the broader sustainability context.

For the social KPI, let’s assume it is focused on leadership diversity. This is a measure used in the Global 100, which defines it as: “Female representation on the Board of Directors and Executive Management team.” A target in this case might be drawn from the public policy realm, such as Goal 5 of the new UN Sustainable Development Goals, which is focused on gender equality. The target could thus require that the percentage of women in these roles aligns with the percentage of women in the workforce.

To normalize these KPIs, the score in the rating could be assigned based on the company’s closeness to the targets. Consider that a corporation reaching or exceeding the 5 percent GHG reduction target could receive a full score of, for example, 1.0 on that KPI. For each percentage point by which the target was missed, the score could be reduced. Similar logic could be applied to the diversity KPI. In an illustrative variation, we might award the company a full score of 1.0 if it is within 5 percent of the target with deductions for each percentage point outside of that range.

The above examples would provide normalized KPIs that have been linked to the broader sustainability context. To combine them into the overall rating, the next step would be to assign weights to each KPI. For simplicity, let’s assume that the weights of each KPI are the same (this is the assumption in many existing ratings). In our example, the weights for each of the KPIs would be 0.5. If the score for the GHG KPI is 0.6 and the score for the diversity KPI is 0.8, the overall score would be 0.7.

Correctly interpreting this number is critical. Would a score of 0.7 mean a corporation is sustainable? No, it would not. To be deemed sustainable in this rating, a corporation would require a perfect score of 1.0. Other scores would provide a sense of the corporation’s progress towards sustainability over time, but any score less than 1.0 would indicate the corporation is not meeting all required targets. Meeting all context-based targets is a condition of sustainability.

In the illustrations above, there are some clear limitations. For instance, the GHG target requires that all corporations achieve the same annual decreases starting from their current baselines. It may therefore be easier for laggards to meet the target early on. The diversity target is fundamentally normative in nature. There may be differing views on what such a target should be. Setting appropriate targets is undoubtedly difficult. Additionally, in a complete sustainability rating, debates on what should be measured, how the data should be normalized, and how the KPIs should be weighted are also likely.

That said, the basic principles outlined above provide a compass in developing a sustainability rating. Measuring corporate sustainability requires explicit linkages to the broader sustainability context. This is currently missing in the slew of popular sustainability ratings. Rankings based on absolute or relative measures are laudable efforts that are interesting and useful, but they do not allow us to separate sustainable and unsustainable corporations.

All sustainability ratings are based on choices that involve considerable judgment. The specific KPIs, normalization methods, and weighting schemes are likely to vary considerably from one rating to another. There is no one definitive formulation. The area in greatest need of attention, however, is in identifying science- and public policy-based targets that are suitable to application at the corporate level. It is time to build context into sustainability ratings.