It appears to be extremely hard to produce a meaningful ESG score. Standard and Poor’s recently decided that it will no longer rank companies’ environmental, social and governance risks from 1 to 5 . This decision can be understood in the light of how aggregate ESG scores are constructed as discussed in a recent FT publication: “Companies with good ESG scores pollute as much as low-rated rivals.” This article made it clear that the current ESG metrics system has a major flaw which entails that it de facto allows firms to hide bad performance in one metric as it will be compensated by good performance in another metric. The fact that such a compensation is possible at all just illustrates that the current ESG measurement system sets out to squaring the circle.
Even a cursory reading of the economics literature would reveal that creating aggregated ESG scores cannot work, as it is impossible to assign meaningful weights to the individual elements that make up such an aggregate measure, see for example Srikant’s work Datar, Susan Kulp and Richard Lambert and by Paul Milgrom and John Roberts. If you fall short of your target A in measure 1 (say carbon emissions), then it cannot be offset by a top performance in measure 2 (say profit). Such assumptions are made in the current metrics and we therefore see in research that companies that perform very poorly in emissions still receive a favorable overall ESG evaluation.
If carbon emissions performance is important, then any (overall) ESG aggregate measure cannot be positive unless carbon emissions performance is positive. The reason is that unfavorable performance on carbon emissions cannot be offset by positive performance on other metrics. This could indicate that unfavorable performance in any metric included in the “aggregate ESG measure” that captures a key performance area (e.g. carbon emissions) should imply a negative overall performance rating. It’s a bit like safety performance in chemical plants, where factory management doesn’t get a positive performance evaluation if a safety target is missed. Factory management may be able to report exceptionally high profits, but this performance cannot compensate for the safety risk the company faces if safety targets are missed.
Applying this idea means developing targets for each specific metric to decide when aggregate level performance will reach the desired level. Setting such goals is a very difficult task, as on the one hand they have to motivate employees to improve ESG performance, while at the same time the goals have to be consistent with external requirements (e.g. Paris Agreement). To the extent that employees are unable to achieve ESG performance levels in line with externally imposed requirements, it is better from a motivational point of view to set more achievable goals for these employees than set in externally imposed goals. Hundreds of studies show that people only really start working if the assigned goal is actually achievable for them. If the outside world discovers that the (external) target is not being met, the company finds itself in a difficult position of having to explain why targets were set lower than they should be according to the external requirements.
Reality is too complex to capture ESG performance in one overall [ESG] metric. Using an oversimplified overall ESG measure is not going to help us, on the contrary it puts us a farther distance from achieving the ultimate net-zero goal. Hopefully the world’s major investors will see Standard & Poor’s problem and lead the way and replace the current overall benchmark with a (set of) metric(s) that does justice to the complexity of performance measurement!
Professor of accounting University of Amsterdam