How Legal and Regulatory Constraints Bind a Portfolio

A legal or regulatory constraint is a limit the holder did not choose and cannot waive. The limit sits in the same document as the chosen constraints, and wherever the two cover the same quantity the tighter one binds. Its defining property is that it can change without anybody connected to the portfolio doing anything. A portfolio can go non-compliant with no trade and no price move.

Every other kind of limit in a policy statement was put there by somebody in the room, and the same room can take it away. An imposed limit cannot be taken away by that room, and three consequences follow. A portfolio can go offside overnight. A committee can vote unanimously and change nothing. A limit that has never mattered can become the only limit that matters, with no decision taken anywhere near the portfolio.

The running example is the Anantara Multi-Asset Portfolio, an invented discretionary mandate of Rs 500 crore run by Faiz Ahmad Ansari for an invented charitable endowment whose investment committee is chaired by Rukmini Deshpande. Its stated shape, drawn below, is equity 60.0 per cent, fixed income 30.0 per cent and cash 10.0 per cent, summing to Rs 500 crore exactly. Actual weights drift away from that shape between one rebalancing and the next.

Every threshold, period, rate, category and registration condition belongs to the published texts of the authorities. Every outside limit below is written as x per cent. A specific limit can be moved by the authority that set it, and a text quoting the old figure gives no sign that it has moved. The requirements that apply to a real arrangement are published by the Securities and Exchange Board of India (SEBI) at sebi.gov.in and by the Pension Fund Regulatory and Development Authority (PFRDA) at pfrda.org.in, and are read there on the day they are needed.

What makes a line in a policy statement legal or regulatory?

Not what it restricts. Two lines can restrict the identical quantity, in the identical words, against the identical denominator, and one is a chosen line while the other is not. The dividing property is whether the line came from inside the room or from outside it. A legal constraintA limit on a portfolio that comes from law rather than from any agreement the holder signed, so nobody in the room can set it aside. comes from law; a regulatory constraintA limit set by an authority that supervises the arrangement. The authority changes it, not the holder and not the manager. comes from an authority that supervises the arrangement. Neither was negotiated with the holder, and neither can be negotiated away.

The everyday version: a household renting a flat writes its own house rules, and can rewrite them over dinner. The lease is different. The landlord and the law behind the lease decide what it says, and no agreement across the dinner table moves a clause in it. Both sets govern the same flat and the same people. Only one responds to the people living there.

The trap is that sameness. Read cold, a policy statement shows every line looking the same, and nothing in the typography signals that a committee vote can move one of them and cannot move the next. A document that does not mark which lines are chosen and which are imposed has hidden the property that decides whether any of them can be changed.

Who can move the line, and how fast?

The differences fall out of that single property. A chosen limitA line the holder or the committee wrote for itself. The same body can rewrite it in writing on a stated date. is changed by the body that wrote it, through an act with a form: a resolution, a date, a note of who was in the room. An imposed limitA line in the policy statement that came from outside the room, so no vote inside the room can move it. is reached by no vote at all, and the date it changes is a date somebody else picks.

Both kinds live in the same policy statement, and a statement that does not mark which is which invites a committee to waive something it cannot waive. The document gives them no signal, so the failure happens to careful people. The full case is set out under the error that gets made.

The Anantara mandate carries five stated lines, drawn below, and all five are that committee's own choices: the equity range, the single holding cap, the exclusion of unlisted holdings, a minimum credit standing written as policy rather than as a rating symbol, and the 60 by 30 by 10 shape those weights sit inside. Every one of them was written by Rukmini Deshpande's committee and could be rewritten by it.

Not one of the five is an imposed limit, so the imposed column of the record stands empty. The imposed limit stays a variable called x per cent, and everything below is worked as a function of x. Working across the whole range shows the shape of the arithmetic instead of one convenient point.

How legal and regulatory constraints affect portfolios, and which of the three cases applies?

Take any single quantity a policy statement limits: the share of the portfolio in one holding, in one asset class, in anything at all. Ask what an outside limit does to that quantity. There are exactly three answers, and they exhaust the possibilities.

The first case is that the imposed limit is tighter than the chosen one. The chosen line is then decorative: still written, still read out, still reported on, and never breachable. A position would have to pass the outside line first. The committee's judgement about that quantity has been overtaken and the document does not say so.

The second case is that the imposed limit is looser. The chosen line is then doing all the work: every breach on that quantity is a breach of the committee's own choice, every waiver conversation is one the committee is entitled to have, and the outside limit sits behind it, untouched and untouchable.

The third case is that the imposed limit addresses a quantity the chosen constraints never mentioned. The imposed limit overtakes nothing. It adds a line the portfolio must now satisfy, and the document has to grow a row it did not previously have. A committee has to know which of the three cases it is in for every quantity its document limits, and most have never sorted their document that way.

Case one costs something easy to miss. The committee has not lost anything it wanted; it has lost the ability to know, from its own document, whether a line it wrote still governs anything. The loss shows up at the next review, when somebody proposes loosening the chosen line and the room debates it without anybody realising the debate cannot change the outcome. A decorative line consumes real attention.

Two separate three-way splits run through this subject, and they are not alternatives. The three cases above sort a quantity by how the outside line relates to the chosen line. The three regions below sort it by where the outside line sits against the chosen cap and the portfolio's own largest holding. A sentence that borrows a term from one and a term from the other has said nothing.

Which line is the manager actually checking?

The lower one, and everything operational follows from that. Where two limits cover the same quantity, the one currently lower is the binding lineWhere two limits cover the same quantity, the one that is currently lower. A position runs into it first.. A position runs into the lower one first. The other limit is an inert limitA limit that is not currently the lower of the two, so nothing can breach it until it becomes the lower one.: not deleted, simply out of reach while the lower line stands in front of it.

Put the Anantara numbers on a scale. The chosen cap is 5 per cent of the portfolio, or Rs 25,00,00,000/- on Rs 500 crore, and the largest holding in the equity sleeve is 4.6 per cent, or Rs 23,00,00,000/-. Both are measured against the portfolio, not against the Rs 300 crore equity sleeve. Stating the base every time is not pedantry: the same holding against the sleeve is 7.7 per cent, and a reader handed 7.7 without its base has been told the portfolio is twice as concentrated as it is.

Now introduce an outside limit on the same quantity and call it x per cent of the portfolio. The letter x is a variable and stands for no requirement anywhere. Slide it up and down and the picture sorts into three regions whose boundaries are 5.0 per cent, where the chosen cap sits, and 4.6 per cent, where this portfolio's own largest holding sits.

The crossover into non-compliance sits at 4.6 per cent, the portfolio's own largest holding. The crossover is a property of the position rather than of any rule. Two mandates with identical documents and different holdings have different crossovers, and the same mandate has a different crossover in March than in January because the largest holding moved. So a compliance question cannot be answered from the document alone, and a committee that reviews its policy statement once a year and its positions once a quarter is looking at the wrong thing on the wrong cycle.

Read the three regions one at a time. Region one is any x above 5.0 per cent: the chosen cap of Rs 25 crore is the lower line, so it binds and the manager checks Rs 25,00,00,000/-. The outside line is written down and reported, and nothing can reach it.

Region two is any x between 4.6 and 5.0 per cent, where the outside line is the lower one, so it binds and the chosen cap goes inert. At 4.8 per cent the binding line is Rs 24,00,00,000/- and the Rs 23 crore holding is inside it, with Rs 1 crore of headroom instead of Rs 2 crore. Nothing has been breached and nothing needs to be traded, yet the committee's cap has stopped governing anything without a word of the document changing.

Region three is any x below 4.6 per cent, where the outside line has passed under the largest holding and the portfolio is offside on the day the change takes effect. No trade was placed, no price moved, and the holding is the same holding it was the day before. Worked at an arbitrary 4.0 per cent, the binding line is Rs 20 crore and the report shows the holding Rs 3 crore over it.

Why does an inert limit still get reported?

Because inert is a temporary condition, and nothing announces the end of it. The chosen cap of Rs 25 crore in region two is standing behind a lower line, and if that lower line moves back above 5 per cent the cap binds again instantly, with nobody in the room deciding anything. The moment the binding line moves, the inert one becomes the constraint, so a limit that is not currently binding is still carried in the report.

Monitoring goes quietly wrong at exactly this point. A pack showing only the active constraint is easier to read, and it throws away what is needed when the outside line moves: what stands behind it, and how much room that leaves. A committee reading a clean pack in March cannot know that a change in April promotes a limit it has not thought about since drafting.

The fix costs nothing. The pack carries every limit that covers the quantity, marks which one binds today, and shows the distance to each. Three columns instead of one, and the report answers the next question before it is asked.

How can a portfolio become non-compliant while sitting perfectly still?

Every other constraint on a portfolio is crossed by something the portfolio did or the market did to it, so the portfolio's own numbers changed and a pack watching them could have seen it coming. An imposed limit can be crossed by an event with no market in it at all. The portfolio's own numbers give no warning, and that missing warning makes it the harder operational problem. The holding did not grow. The total did not move. The line came down and passed underneath a position that had not so much as twitched.

The household version is easy to feel. A rented flat, where the lease sets what may be kept on the balcony. Nothing new went onto it this month, but the clause changed, on a date, in a document the tenant did not draft, and what has been sitting there for two years is not permitted. The tenant did nothing. The tenant is nonetheless in breach, and the first word of it is a letter.

So the operational answer to an imposed limit is never a limit monitor alone. A limit monitor watches the portfolio, and the limit moved instead. The change shows up in one place only, so somebody has to read the published text on a stated cycle and record the date they read it.

What does the return-to-compliance trade actually cost?

More than the report says, and the direction of the error is what almost everybody gets wrong. Stay in region three at the arbitrary x of 4.0 per cent. The holding is Rs 23 crore, the line is Rs 20 crore, so the pack prints a gap of Rs 3 crore and the obvious move is to sell Rs 3 crore. Return to complianceThe set of trades that brings a position back inside a limit it is currently outside. looks like a subtraction.

The proceeds do not leave, so the correction is not a subtraction. Sell Rs 3 crore inside a discretionary mandate and the cash lands in the same portfolio: the holding falls to Rs 20 crore and the total falls to Rs 497 crore, and 20 divided by 497 is 4.024 per cent. The denominator did not stay where the report left it, so selling the gap leaves the portfolio still offside.

Solve it properly instead. Let s be the amount sold. The holding after the sale must equal the limit share of the total after the sale, so Rs 23 crore less s equals 4 per cent of Rs 500 crore less s. The equation gives 23 less s equals 20 less 0.04s, so 3 equals 0.96 times s, so s is Rs 3.125 crore, being Rs 3,12,50,000/-. The holding lands at Rs 19.875 crore on a total of Rs 496.875 crore, and 19.875 divided by 496.875 is exactly 4.0 per cent.

A check is available. Sell the gap of Rs 3 crore, land at 4.024 per cent, and solve again on the new numbers: the second sale is 20 less 19.88, divided by 0.96, or Rs 12,50,000/-, and the two trades add to Rs 3,12,50,000/- exactly. The arithmetic is not punishing the first mistake, it is charging two sets of costs to reach a point one trade would have reached.

The shape applies to any limit written as a share of a total the sale itself reduces: the sale is the gap divided by one less the limit share. Divide by 0.96 at a limit share of 4 per cent and the sale runs about four per cent above the gap. The higher the limit share, the further above the printed gap the true trade sits.

Which of the mandate's own lines could an outside limit even touch?

Run the three-region reading across every line the Anantara mandate carries and they are not all equally exposed. Some limit a quantity an outside limit could plausibly address, so the whole binding-line apparatus applies to them. Others are of a shape no outside limit can make decorative.

Start with the equity range. The mandate permits equity between 50 and 70 per cent, or Rs 250 crore to Rs 350 crore on Rs 500 crore, with the policy weight at Rs 300 crore inside it. An outside ceiling on equity sorts into the same three regions: inert above 70 per cent, binding between the current equity holding and 70 per cent, and offside below the current equity holding with nothing traded.

Now the line that behaves differently. The endowment's committee decided to hold nothing unlisted. Suppose an outside limit turned up covering unlisted holdings: whatever it says, it permits some amount, and permitting is not requiring. An outside line that permits something obliges nobody to hold it, so a chosen exclusion survives whatever the outside line allows, and no imposed limit can make a zero decorative.

One more thing decides whether any of this arithmetic is comparable: the denominator. Every figure here is measured against the Rs 500 crore portfolio because the chosen cap is written against the portfolio. An outside line written against a different base is not a tighter or looser version of the same line; it is a line about a different quantity, and the two cannot be compared until both are in rupees.

How does a committee use this in a room, on a Tuesday?

Three artefacts, none of them exotic. The first is a marked document: every line in the policy statement carries a chosen or imposed mark, put there when the line was drafted rather than reconstructed afterwards. Reconstruction is where the errors come from. Whoever reconstructs is guessing at the intent of a drafter who left the organisation two years ago.

The second is a constraint sheet carrying both limits for every quantity the document limits, in rupees, with the distance to each and a mark saying which one binds today. A breach is a fact about the whole and cannot be found by reading positions one at a time, so Rukmini Deshpande's committee reads that sheet before the holdings list. Faiz Ahmad Ansari works from the same sheet. The binding line governs whether a trade may be placed at all.

The third is a reading log: who read the published text, on what date, and what they found. The change a log catches never appears in any portfolio number, so the log is the only defence against the waiver failure. A monitoring process that watches only the portfolio is blind to exactly the kind of constraint that can breach it without warning.

A household with a home loan needs the same mark in miniature: its own budget rules on one list, the loan conditions it did not write on another. The budget rules can be argued about on a Sunday; the loan conditions cannot, and finding out what they currently say means reading the lender's document rather than remembering what somebody said at signing.

Where does a reader find what actually applies to their own mandate?

In the published text, on the day it is read, and nowhere else. A requirement is a moving object: it has a version, an effective date and a publisher, and the publisher is the only party who can say what it currently says. Everything downstream of the publisher is a copy of a copy with a timestamp somebody forgot to write down.

Four publishers matter here, two supervising authorities and two market bodies. A limit quoted from anywhere other than the current published text may already have moved, and nothing in the quotation reveals that it has.

Client type changes the reading without changing the mechanism. An institutional holder and a private one are not covered by the same body of text, and which body applies is settled in the published texts themselves. The binding-line arithmetic, the three regions, the crossover at the largest holding and the return-to-compliance solve are arithmetic, so they are identical in both settings. Only the text to be read is different.