Green washing within asset management, the act of appearing to be more ESG-credible than one actually is, comes about because there is as yet no agreed standard for what it means to be a sustainable investor, nor how to demonstrate this. Wrangling over labels aside, debates continue around whether there’s quantifiable alpha to be found through sustainable investing (and the right way to measure it, if so) and how best to capture E, S and G portfolio risks.

Despite this backdrop of flux and uncertainty, the number of ESG fund launches continues on an upward trend, and it is clear that ESG and sustainable investing is set to become a driver for structural shifts within active management.

While active managers continue to see the share of flows into passive trackers increase, the shifting landscape of ESG provides an opportunity for stock pickers to add insight and value. It would seem logical, then, to consider ESG positioning through an active lens

Understanding active ESG positioning may seem like an unrewarding exercise, but it can reveal insights. These insights are of value not only to the end-to-end portfolio construction process, but also for building coherent narratives around positioning – as demanded by clients.

In this note we describe the results of applying a process to decompose an active (but otherwise generic) ESG metric into allocation and selection effects, relative to a sector view. We do this by mathematically reframing the well-knownBrinson et.al performance attribution model in to an exposure context. What follows is an abridged description of select examples complete with commentary.

Clients can find a full description of the model over at theclient portal, as well as tools to perform this analysis on portfolios.

For all examples we consider a hypothetical portfolio benchmarked against an index partitioned into just 4 sectors: IT (60%), Industrials (20%), Financials (18%) and Health Care (2%). Without loss of generality, we can fix the sector aggregate ESG scores for both the portfolio and benchmark, and vary the portfolio (and therefore active) weights to highlight different features.

We start by considering a portfolio with sector weights equal to that of the benchmark. The first metric we compute is the contribution to active ESG, whose sum is equal to the active ESG (6.45-6.378 = 0.072). In the case of a fully invested long-only portfolio for each sector the contribution to Active ESG is given by wp*ESGp-wb*ESGb. We see the largest positive contribution coming from IT at 0.3 (largest sector weight), while Industrials is a detractor to active ESG (-0.24).

Sector	Pf weight %	Pf ESG	Bmk weight %	Bmk ESG	Act ESG	Cont Act ESG
IT	60	6.5	60	6	0.5	0.3
Industrials	20	4.6	20	5.8	-1.2	-0.24
Financials	18	8.5	18	8.1	0.4	0.072
Health Care	2	5	2	8	-3	-0.06
TOTAL	100	6.45	100	6.378	0.072	0.072

In the above, the portfolio sector weights match the benchmark sector weights exactly – therefore the difference in sector ESG scores is due to differences in stock selection: either different names entirely, or the same names with differing weights compared to the benchmark. By applying a modified version of the Brinson model, we can quantitively decompose active ESG into sector allocation and stock selection. And in this case, we would like to see 0 attributed to sector allocation (which indeed we do):

Sector	Pf ESG	Bmk ESG	Act ESG	Cont Act ESG	Selection Eff
IT	6.5	6	0.5	0.3	0.3
Industrials	4.6	5.8	-1.2	-0.24	-0.24
Financials	8.5	8.1	0.4	0.072	0.072
Health Care	5	8	-3	-0.06	-0.06
TOTAL	6.45	6.378	0.072	0.072	0.072

We see above that in the case with 0 active sector weights the selection effect column reduces to the contribution to active ESG metric – as expected. Similarly, for portfolio sector weightings that are simply a re-scaling of benchmark positioning (no active selection), we obtain 0 attributable to selection effects.

If we now allow the sector active weights to vary, we capture the active ESG contribution attributable the sector positioning (underweight or overweight) in the Allocation Effect column, alongside the same for stock selection in the Selection Effect column.

Sector	Pf ESG	Bmk ESG	Act weight %	Act ESG	Cont Act ESG	Allocation Eff	Selection Eff
IT	6.5	6	-7	0.5	-0.155	0.026	0.265
Industrials	4.6	5.8	-7	-1.2	-0.562	0.04	-0.156
Financials	8.5	8.1	-4	0.4	-0.268	-0.069	0.056
Health Care	5	8	18	-3	0.84	0.292	-0.6
TOTAL	6.233	6.378	0	-0.145	-0.145	0.289	-0.435

At this point, it should be pointed out that the allocation and selection effects columns are constructed independently of the standard contribution to active ESG column. In fact, there is an isomorphic relationship

Meaning that the sum of allocation and selection are comparable, in a sense, to the contribution column.

Although this example is somewhat extreme, it serves to highlight some key features of both allocation and selection effects. First, the contribution to active ESG from IT sector is computed as -0.15 and appears to be diminishing total active ESG. At the same time, both the allocation and selection effects are positive.

The allocation effect for IT is positive (0.026) as the portfolio is short this sector, which has a benchmark ESG score of 6 versus the benchmark total of 6.378. This translates as a positive allocation score.

The selection effect for IT is significantly positive (0.265), due to the portfolio achieving a higher sector ESG score (6.5) while underweight.

Similarly, the Health Care sector has significantly positive allocation effect due to the large active overweight in a high ESG sector (8) versus the benchmark total (6.378). Although even with this significant portfolio weighting, the sector ESG score of 5 is significantly less than the benchmark sector total (8), which drives the significantly negative selection effect.

As hinted above, by summing the allocation and selection effects, we recover a metric comparable to the standard contribution to active ESG metric. We call this combined metric, Total Effect and note the sum of total effect over sectors is equal to the total active ESG (-0.145)

The Total Effect metric is comparable to contribution to active ESG, in the sense that both metrics are an arithmetic decomposition of active ESG over sectors. In addition, total effect provides more information as it is decomposed over allocation and selection dimensions. The Brinson-type metrics described here are not a replacement for the standard contribution measures – both can be used together to help reveal sources of ESG exposure and how sector level ESG exposure is obtained. In this regard, the Brinson metrics can inform the portfolio construction process, and help communication with clients by feeding into a narrative around active positioning and name selection, relative to aggregate ESG scores.

Active ESG: asset allocation and stock selection