Special Report ESG: Carbon Risk, How the low-tracking-error green index strategy works
As pioneered by the Swedish pension buffer fund AP4, low-tracking-error green indices work on a remarkably simple principle: weighting the stocks in each sector by carbon intensity (CO2 per unit of sales) and removing the most carbon-intense companies and exposure to stranded asset risk in intensity based on market cap. The latest iteration produces an index that retains sector weightings, reduces carbon reserves intensity by at least 50% with a minimal tracking error (0.7%) after the remaining stocks are rebalanced.
An initial strategy that AP4 initiated in 2012 against the S&P500 excluded the poorest performing 20% of the index constituents in terms of carbon intensity as measured by Trucost, with the constraint that the stocks removed would not exceed a reduction in the GICS sector weight by more than 50%. This reduced carbon intensity by 48% with a tracking error of 0.5% when it was implemented.
Using MSCI ESG research, AP4 then worked on an approach that excludes companies not only based on emissions but also on intensity of carbon reserves. It applied a related methodology to its Pacific ex Japan portfolio that also maintained the original sector and country weights of the index.
Last year, AP4, FRR and Amundi worked with MSCI to develop low-tracking-error indices based on the MSCI Europe and World indices that excludes the most carbon-intense companies in terms of emissions and reserves, with a maximum turnover constraint, a threshold for sector and country-weight modifications and minimal tracking error.
C02 emissions are calculated as a proportion of sales to remove capitalisation bias and the 20% most carbon-intense companies are excluded, with a cumulative sector weighting of less than 30%. The largest owners of carbon reserves (calculated by reserves divided by market cap) are also then removed from the index.
Tracking-error risk was prioritised over absolute carbon reduction; testing against the MSCI Europe index found that a 100% reduction in carbon would have led to an unacceptable tracking error of 3.5%, while the chosen low-carbon strategy against the MSCI Europe has a 0.72% tracking error with a 62% reduction in carbon emissions intensity and an 81% reduction in carbon reserves intensity.
As with any such exercise, the output is reliant on the accuracy of the input. While there has been considerable progress on the disclosure of carbon emissions, the investors recognise that full data accuracy is yet to be achieved.