ESG Integration: What a Non-Financial Criterion Does

Environmental, social and governance (ESG) integration is the practice of letting such criteria enter an investment process alongside financial ones. Whatever form the criterion takes, its structural effect is the same: it narrows the set of holdings the manager may choose from. Whether that narrowing helps a portfolio's return or harms it is contested, and the contest rests on evidence quite separate from the narrowing itself.

Three letters start more arguments than almost anything else in investing, and not one of those arguments has been settled. A narrower question can be settled, and once held it is far more useful. Adding a rule to a list of candidates does something to that list, and what happens is arithmetic. Whether the rule was a good idea is a separate question sitting on separate evidence.

The Anantara Multi-Asset Portfolio, an invented discretionary arrangement of Rs 500 crore, carries every figure worked below. Faiz Ahmad Ansari manages it. Rukmini Deshpande chairs the investment committee of the charitable endowment behind it. The policy weights are equity 60.0 per cent at Rs 300 crore, fixed income 30.0 per cent at Rs 150 crore and cash 10.0 per cent at Rs 50 crore, and the equity sleeve is held across 28 names. Four written rules already sit over all of it: equity between 50 and 70 per cent, no single holding above 5 per cent of the portfolio, no unlisted holdings, and a minimum credit standing on the fixed income sleeve.

The Anantara record states no policy of any kind on any environmental or social criterion. Four written rules govern the mandate and nothing beyond them does. Every count of eligible names worked out below is a chosen illustration, picked to make a division visible, and each one is labelled as chosen where it appears.

What does it mean to integrate a non-financial criterion?

Start with the three letters, and treat them as three headings rather than as a scoring system. Environmental considerations cover how a business acts on the physical world around it and how exposed it is to that world changing. Social considerations cover how it treats the people it employs, the people it sells to and the people who live next to it. Governance considerations cover how decisions get taken inside it, who takes them, and who is in a position to check them afterwards.

None of those three is a number. Each is a heading over a whole range of possible rules, and a rule is what actually enters a process. A criterionA stated rule that a candidate holding has to satisfy before it can be considered for the portfolio at all. is a stated rule that a candidate has to satisfy, and the three letters describe where such rules come from rather than what any of them says. How a score or a rating is built is a separate subject.

The word integrationLetting a consideration enter an investment process. The consideration can enter at the list of candidates, inside the analysis of one candidate, or in a report written afterwards. is doing a lot of quiet work, and part of the confusion in this subject comes from it. Integration is used for at least three different operations, one of which changes what may be held, one of which changes an opinion, and one of which changes nothing at all. The three operations are separated below. For now, hold on to the definition that survives all three: something outside the financial case has been allowed to bear on the process.

Does a criterion help a result or hurt it?

The question is genuinely open, and leaving it open is a position rather than an evasion. No study or survey settles it. The findings that exist point in both directions and disagree about what they are even measuring, so a reader who wants a verdict has to weigh that evidence rather than be handed one.

Two things keep a question this obvious unanswered. The first is that a single stated year on a single mandate cannot touch a contested question. The Anantara record for its stated twelve month period shows a gross 14.2 per cent against 12.6 per cent for its unnamed composite benchmark. The record contains no criterion, so neither number can be attributed to one, and reading anything about environmental or social rules out of that pair would be inventing a finding.

The second reason matters more. Understanding what a criterion does to a portfolio does not require the return question to be answered first. The structural question is arithmetic, and arithmetic is settled. A committee has to live with the structural consequence, monitor it and report on it, whichever way the argument about returns eventually goes. So the return argument stays where it is, and the structural part gets worked.

Why is a criterion a constraint rather than a preference?

Here is the whole mechanism, and it takes one picture to see. Before a rule is applied, a manager chooses from a set of candidates. After the rule is applied, the manager chooses from part of that same set. Nothing else about the manager, the money or the market has changed. Narrowing is not a claim about anything at all. Narrowing is simply what applying a rule to a set does, and it would hold equally if the rule were about balance sheets, listing status or the colour of the head office.

The word for what is left over is the eligible setThe candidates a manager is allowed to choose from once every rule in force has been applied. Everything outside it is unavailable regardless of how attractive it looks., and the word for the rule that produced it is a constraintA limit on what a manager may hold, whatever the reason behind it. A constraint is written as a rule and tested against a number.. Notice that the second word carries no opinion about the reason. A rule barring unlisted holdings is a constraint. A rule barring anything the holder finds unacceptable is also a constraint. Both are the same kind of object, and both do the same kind of thing to a set.

Take it out of finance for a moment. The shape is easier to feel when there is no money in it. A shopkeeper is hiring. Fourteen people apply and each is a real possibility. The shopkeeper then adds a requirement: the person has to live close enough to walk in. Five of the fourteen do. Now, was that a sensible requirement? Possibly. Somebody who walks to work is never late because a bus did not come, and the shopkeeper may value that above everything else. Possibly not. The best of the fourteen might live three stops away.

The wisdom of the requirement and the count of applicants left are completely independent of one another. Whether the rule was wise is arguable and may never be settled. The shopkeeper now chooses from five people instead of fourteen, and no argument touches that. Only the count has an answer anybody can write down, and investment committees routinely spend an hour on the first question and no time whatsoever on the second.

How ESG Constraints Affect Portfolio Construction, and what follows without any view about returns?

Four things follow, and it is worth saying up front what kind of things they are. None of the four is a prediction, a finding or an argument. Each is what happens when a rule is applied to a set that money is then spread across. None of the four depends on the criterion being any good, so a reader could disagree violently with it and still have to accept all four.

The first has already appeared: the eligible set shrinks. Whatever was excluded is no longer available, however attractive it looks on the financial case, and the manager's choice is made from what remains.

The second is where the arithmetic starts to bite. If the sleeve still carries the same money and there are fewer names to put it in, then each position sizeHow much money sits in one holding, stated either in rupees or as a share of the portfolio it belongs to. is larger on average. A larger average position is a division, not a judgement. The money did not shrink with the list.

The third is about the benchmark. If the composite benchmark holds a name and the mandate is barred from holding it, the portfolio's weight in that name is zero and the benchmark's is not. The difference between the two is an active weightThe difference between the portfolio's weight in a name and the benchmark's weight in that same name. It is negative where the portfolio holds less than the benchmark does., and a portfolio full of differences from its benchmark carries active riskThe risk that comes from being different from the benchmark, whoever created the difference and for whatever reason. It does not care why the difference exists. whether anybody chose those differences for a return reason or not.

The fourth is the one most often missed. A new rule does not queue up politely beside the rules already in the mandate. A new rule interacts with them, and the interaction can be the whole story. Every one of the four follows from applying a rule to a set of candidates, and not one of them needs a view about returns in order to hold.

The second consequence turns into a limit test later, so put numbers on it. The Anantara equity sleeve is Rs 300 crore across 28 names, so the average holding is Rs 10.71 crore. Suppose a rule left 22 of those 28 eligible. The figure of 22 is a chosen illustration. The record holds no criterion, and therefore no count of what any criterion would remove. If the sleeve still carries Rs 300 crore, the average holding becomes Rs 13.64 crore. The money stayed exactly where it was, so the list fell by a fifth or so while the average position rose by a little over 27 per cent.

What does the arithmetic look like on the invented mandate?

A percentage cap is really a rupee limit, so work it properly. The concentration capA limit on how much of a portfolio may sit in one single holding. It is written against a stated base, and the base decides what the test reads. in this mandate is written as 5 per cent of the portfolio. Five per cent of Rs 500 crore is Rs 25 crore, and that rupee figure is what a limit test actually reads. The biggest position in the stated twelve month period stood at 4.6 per cent, or Rs 23 crore. Subtract, and Rs 2 crore separates that position from its limit. The position is already standing at 92 per cent of the ceiling written above it.

The whole subject turns on the base, so state the base every time. The cap is written against the portfolio, not against the equity sleeve. The same Rs 23 crore holding reads 4.6 per cent measured against the Rs 500 crore portfolio and 7.7 per cent measured against the Rs 300 crore sleeve. Neither figure is wrong. The two answer different questions, and the base therefore gets named on every line that carries a weight.

Now bring the two halves together. On the 22 name illustration the average holding is Rs 13.64 crore, comfortably under Rs 25 crore, so nothing is breached and nothing is close to being breached on the average. The arithmetic establishes a direction rather than a breach. The sleeve has moved towards a limit that its largest position was already sitting four tenths of a percentage point away from. A rule that removes some names while leaving the biggest one eligible pushes a portfolio with very little headroom in the only direction it has no room to travel.

A floor is hiding in this, and finding it shows where narrowing eventually stops being a matter of degree. If no holding may exceed Rs 25 crore and the sleeve must carry Rs 300 crore, then the sleeve cannot be held in fewer than twelve names at all: Rs 300 crore divided by Rs 25 crore is exactly 12. At 28 names the average uses 42.9 per cent of the cap. At 22 names it uses 54.5 per cent. At 12 names the average holding equals the cap exactly. Every single position would then have to sit precisely on the limit, and one name below the average forces another above it.

Where in the process can a criterion actually be applied?

Three places, and all three get described with the same sentence. Somebody says a criterion has been integrated into the process, and almost nothing is known until somebody asks where.

The first place is the eligible set itself, before anything is chosen. Removing candidates from the list is exclusionApplying a rule by removing candidates from the list before any choosing happens. An excluded candidate cannot be held, whatever the financial case for it looks like., the operation that everything above describes. Candidates are removed and the manager never gets to weigh them.

The second is inside the analysis of an individual candidate, as one input among many. A governance concern might change the view taken of a business without removing it from consideration at all. A concern used this way changes an opinion rather than a set, and two analysts applying the same input can reach opposite conclusions about the same candidate.

The third is after the portfolio is built, as a description of what is in it. Somebody measures the finished holdings against a set of headings and writes it up. The third operation changes neither the set nor the view, and it is routinely described in the same words as the first, so a reader has to ask which one is meant every single time.

How does a criterion meet the constraints already in the mandate?

The Anantara mandate carries four written rules before any non-financial rule is discussed at all. Equity has to sit between 50 and 70 per cent. No single holding may exceed 5 per cent of the portfolio. No unlisted holdings. A minimum credit standing applies to the fixed income sleeve. Add a fifth rule, and it does not take a seat at the end of that list and wait its turn.

The first of the four it meets is the cap, and the reason has already appeared. Narrowing pushes the same rupees into fewer names. Fewer names lift the average position, and a higher average moves the sleeve towards the one limit written per holding. A criterion therefore cannot be assessed on its own. Its effect on a portfolio depends entirely on what constraints were already sitting there when it arrived.

Change the mandate and the same criterion behaves differently. A portfolio with a cap at 10 per cent rather than 5 has twice as much room to absorb the same narrowing. A portfolio whose largest position sits at 2 per cent rather than 4.6 has room to spare. A portfolio holding 120 names rather than 28 barely notices. Same rule, three different consequences, and none of the differences has anything to do with the merits of the rule.

What can be said about the distance from the benchmark, and what cannot?

Here the evidence runs out, and the place where it runs out is worth marking. Suppose the composite benchmark holds a name the mandate is barred from holding. The portfolio's weight in it is zero. The benchmark's weight is something above zero. The difference between the two is therefore not zero, and a difference from the benchmark is active risk regardless of who created it or why. The existence of the difference is fixed by definition and needs no data at all.

Now try to size that difference, and everything stops. Sizing it takes the benchmark's weights holding by holding and a list of the names a criterion would remove. The Anantara record carries neither. So the size of the difference cannot be computed at all, not approximately and not with a range. The sign of the effect is known and the size is not. Saying exactly that is the finding rather than a failure to reach one.

What does this record say about this mandate's own policy?

Nothing, and both halves of that word matter. The first half: the Anantara record states no policy on any environmental or social criterion, so no criterion of that kind can be said to have been applied, refused, or ever put to the committee. Every figure worked above is arithmetic run on the mandate's recorded shape, not a description of its conduct.

The second half is the one a careful reader is already reaching for. The record does hold two vote counts: 214 resolutions voted in the stated twelve month period, and 19 of those cast against the board's recommendation, or 8.9 per cent once computed. Reading a policy out of those two counts is very tempting. A count of votes carries no categories, so nobody can say what any of the 214 resolutions were about. Reading a policy out of them would be manufacturing one rather than finding one.

What makes a criterion testable rather than a label?

Stewardship settles that a duty becomes real when it is written so that a number can check it. A criterion sits in exactly the same position, and there is nothing special about it being non-financial. To be testableWritten so that somebody outside the decision can check afterwards whether it was actually followed, without having to ask the person who made the decision., a criterion has to name four things.

The first is what is excluded or scored. The second is the level at which it is measured. A rule applied to a whole group is a different rule from one applied to a single business. The third is the evidence, and where that evidence comes from. The fourth is who decides, including who reviews the decision afterwards. A criterion that misses any one of those four cannot be monitored by anybody. Such a criterion sits in exactly the position of a duty nobody can check.

Test it on a real-sounding sentence. A mandate that says the manager will avoid poorly governed companies has named none of the four. What counts as poorly governed? Measured at what level? On whose evidence? Decided by whom? A year later nobody, including the manager, can say whether the sentence was followed, and a committee reviewing it has nothing to review.

What does a committee actually write down about this?

Three lines, prepared before the meeting rather than argued out inside it. How many candidates were in the eligible set before the rule and how many after. The average position that follows if the sleeve carries the same money. How much room is left under the concentration cap once that has happened. On the Anantara arithmetic those three read 28 to 22, Rs 10.71 crore to Rs 13.64 crore, and Rs 2 crore of headroom on the largest position before anything moved at all.

An analyst reviewing a mandate from the outside runs the same three and adds a fourth: which of the three operations is actually in force. A report written afterwards changes nothing about what may be held. A lender or a holder assessing a manager's stated policy asks the four testability questions and stops if any one has no answer. None of that requires anybody to hold a view about what a criterion does to a return, and so all of it can be done in a room where nobody agrees about returns.

The household version costs nothing and works the same way. A person decides they will not put savings into anything connected with a particular trade. Fine. Whether that will cost them money is unknowable, and it is not the next question. The next question runs: how many of the options in front of me does the rule remove, does what is left still spread my savings across enough different things, and have I written the rule down clearly enough that in a year I can tell whether I followed it. Same three questions, six zeroes fewer.