- What does “Backward Data Assumption” mean with respect to ESG data? Typically, when S&P DJI creates back-tested index data, we use data from relevant databases, or actual live data. Examples include constituent-level data such as historical price, market capitalization, and corporate action data. As ESG investing is still in the early stages of development, certain data points used to calculate S&P DJI’s ESG indices may not be available for the entire desired period of back-tested history. In such cases, S&P DJI may employ a process called “Backward Data Assumption (or pulling back) of ESG data for the calculation of back-tested historical performance.
Backward Data Assumption” is a process that applies the earliest actual live data point available for an index constituent company to all prior historical instances in the index universe. For example, if an index methodology requires all eligible constituents to have product involvement data, and actual product involvement data is only available for a company from 2015 forward, then S&P DJI will use the 2015 product involvement data for that company for the purposes of calculating back-tested data for the years 2010 through 2014.
- Why is “Backward Data Assumption” for ESG data sometimes necessary? Employing the Backward Data Assumption technique generally provides a more indicative depiction of index characteristics and risk/return profile than would be provided by limiting back-tests to actual live data. The Backward Data Assumption also allows the hypothetical back-test to be extended over more historical years than would be feasible using only actual live data.
Many ESG data providers started with limited coverage and have been increasing their historical coverage over the past few years, so creating back-tests that use only actual historical live data would often lead to unrepresentative index constituent characteristics. Without Backward Data Assumption of ESG data, far fewer companies would be eligible for or selected from the index universe in the back test compared with the same index’s more recent and on-going index universe of eligible and selected constituents.
Therefore, S&P DJI may employ a Backward Data Assumption methodology to provide a longer and more representative back-test period.
- Are any live index rebalances affected by the practices of Backward Data Assumption? Actual live data is used in the rebalance calculation of an index immediately prior to launch and in all rebalances after the launch of the index. Backward Data Assumption may only affect the historical back-test prior to then.
- Which indices have back-tested history that uses Backward Data Assumption? S&P DJI uses Backward Assumption Data with respect to Sustainalytics and Arabesque data, and sometimes uses it with respect to data from Trucost and SAM, both part of S&P Global. Therefore, back-tested history for indices that use data from any of those sources may be affected by the Backward Data Assumption method.
The methodology and factsheets of any index that uses Backward Assumption Data in back-tested history will explicitly state so. The methodology will include a table setting forth the specific data points and relevant time period for which Backward Data Assumption was used.
- When do indices typically have back-tested history that uses Backward Data Assumption of ESG data? For indices launched from 2020 onward, Backward Data Assumption is used in all indices that use exclusionary screens based on Sustainalytics’ product involvement data and Arabesque’s United Nations Global Compact (UNGC) data.
For indices launched prior to 2020, Backward Data Assumption of ESG exclusionary screen data was limited only to Sustainalytics and Arabesque data for historical rebalances prior to 2013.
S&P DJI may also employ Backward Data Assumption to S&P DJI’s ESG Scores and/or Trucost datapoints, if based on historical coverage it is determined that attaining the index objective would be severely restricted otherwise. Historical coverage is assessed year-by-year, both in terms of the number of constituents and weight of those constituents in the underlying universe.