ETFs in a Portfolio: Structure, Liquidity and Tracking

An exchange traded fund reaches a portfolio as units that trade on an exchange, so the account gets a broad exposure in one line and a price set by whoever will trade with it at that moment. Three things then matter to the portfolio. The first is how the holding is counted against the mandate's limits. The second is the liquidity it depends on, out of the two that exist. The third is how far its result drifts from the rule it follows.

Each of the three questions has a different answer, and one invented portfolio is enough to work all three with a figure standing behind every claim. How the unitsThe tradeable pieces a pooled holding is divided into. An account holds a number of them, and its position is that number multiplied by whatever one of them fetches. of a pooled vehicle come into existence, how the value of what such a vehicle holds is struck, and how the vehicle is run from one day to the next are all covered separately. The account's side of the arrangement is a different subject from the vehicle's insides.

The running example throughout is the Anantara Multi-Asset Portfolio, an invented discretionary mandate of Rs 500 crore run by Faiz Ahmad Ansari for a charitable endowment whose investment committee is chaired by Rukmini Deshpande. Its 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. Its mandate states an equity band of 50 to 70 per cent, a cap of 5 per cent of the portfolio on any single holding, no unlisted holdings, and a minimum credit standing on the fixed income sleeve. Every figure attached to that mandate belongs to one stated twelve month period.

What does an exchange traded fund actually deliver into a portfolio?

Three things arrive together, and it helps to name them separately before they get tangled. The first is a single row in the holdings recordThe list an account receives showing what it holds, at what amount, and at what share of the whole. The holdings record is the document a committee actually reads.. The second is exposure to whatever stated rule the vehicle follows. The exposure is a spread of underlying positions somebody else selected by the rule, not a set of choices the manager made one at a time. The third is a price that keeps moving all the time the exchange is open. An order book sits there all day, with somebody on the other side of it.

The plainest version comes from the street. A household buying vegetables can pick each one, weigh it, argue about it and carry home a bag whose contents it chose item by item. Or it can buy the mixed bag the vendor has already made up to a stated recipe, hand over one amount and write one line in the household book. The second household has vegetables too. The second household cannot say, later, exactly how much brinjal it bought, and cannot refuse the brinjal while keeping everything else. An exchange traded fundA pooled holding whose units are bought and sold on a stock exchange during trading hours, giving the buyer exposure to a stated set of underlying positions in a single line. delivers exactly that trade: one line and one price, in exchange for the item by item visibility and the item by item control.

Notice what has and has not changed for the mandate. The exposure is real: rupees of the Anantara Multi-Asset Portfolio are now sitting behind whatever underlying names the vehicle's rule selects, exactly as they would be if Faiz Ahmad Ansari had bought those names one at a time. The reporting object changed. At line levelReading a report one row at a time and treating each row as a single holding, without asking what sits behind the row., the account holds one thing. Behind the line, it holds many. Every difficulty that follows is a consequence of that one split.

Write both sides of that trade down before going further. A manager who has not written them down tends to remember only the half that suited the decision.

Does the holding count as one line, or as everything behind it?

The question is the most practical one in the whole subject, and it is almost never asked out loud. The arithmetic only means something against a stated base, so the account as it currently stands has to be on the table before the question can be worked. The Anantara Multi-Asset Portfolio holds Rs 300 crore of equity across 28 direct names. The largest of them is Rs 23 crore. Against the Rs 500 crore portfolio that is 4.6 per cent, and against the Rs 300 crore equity sleeve the same holding is 7.7 per cent. The ten largest together are Rs 155 crore, or 31.0 per cent of the portfolio and 51.7 per cent of the sleeve. Neither pairing is wrong, and the two answer different questions. Every weight therefore carries its base in the same sentence as the number.

Now suppose part of that sleeve is held as listed pooled units instead of as direct names, at Rs 24 crore. The constraint testThe check that asks whether a stated limit in a mandate is satisfied. The test cannot be run until the base the limit is measured against is known. can be run two ways and they give answers that are nowhere near each other.

Read as one holding, the position is a single row like any other. The cap is 5 per cent of the Rs 500 crore portfolio, or Rs 25 crore, and Rs 24 crore sits inside it with Rs 1 crore to spare. Read by look-throughReading a pooled position as the underlying holdings behind it rather than as the single line it appears as in a report., the same Rs 24 crore is spread across roughly fifty underlying names, so about Rs 0.48 crore of the account rests in each. Against the Rs 500 crore portfolio each name is 0.096 per cent, nowhere near a 5 per cent cap. The two readings differ by a factor of about fifty on the identical position, on the identical day, under the identical mandate.

Put the two answers on a single scale and the size of the disagreement stops being abstract. Against the 5 per cent cap, the line level reading fills almost the whole allowance while the look-through reading is a sliver against the same axis.

So which is right? Neither, on its own. The two readings answer two different questions, and the honest position is to say so. The reading the Anantara mandate uses is not stated in the record for this mandate, and the mandate document settles it rather than any reader of a report. That is not a dodge. The finding is the practical one: a constraint written at 5 per cent of the portfolio is not a complete instruction until it also says what counts as a holding.

The same Rs 24 crore can be quoted on four different bases without any of the four being wrong. The base therefore has to travel in the same sentence as the number every single time it is printed.

There is a second consequence of the look-through reading that a report will never show, and it is the one that turns a tidy constraint into an untested one. Suppose a single underlying name inside the pooled line accounts for 3.0 per cent of what the vehicle's rule holds. The name is then Rs 0.72 crore of the account. If the equity sleeve already holds Rs 23 crore of that same name directly, the combined exposure is Rs 23.72 crore, or 4.74 per cent of the Rs 500 crore portfolio. The direct line printed 4.6 per cent, the pooled line printed 4.8 per cent, and the 4.74 per cent that the account genuinely carries in one name appeared on neither.

Which liquidity does a pooled holding actually depend on?

Two entirely different things share this word, and mixing them produces some of the worst surprises in a portfolio. The first is traded liquidityHow easily a holding can be bought or sold on the exchange, which depends on who else is willing to trade it at that moment., a fact about the units. The second is underlying liquidityHow easily the holdings sitting behind a pooled position could themselves be bought or sold in their own market., a fact about what sits behind them.

Take the street version first. A vegetable vendor sets up outside an office building. On a Tuesday afternoon there is a queue and everything moves. On a public holiday the office is shut, the street is empty, and the vendor cannot sell a thing. Nothing has happened to the vegetables. The farm they came from is unchanged, the wholesale market is unchanged, the produce is exactly as good as it was. Who was standing in front of the stall changed. A quiet week in the units of a listed pooled holding is that empty street, and what it reports on is the street.

Three of the four combinations get assumed away, so all four are worth laying out. Units and underlying can both be easy, the ordinary case nobody notices. Units can be quiet while the underlying is deep, the situation that panics a committee for no good reason. Units can change hands briskly while the holdings behind them would be slow and expensive to shift, the situation nobody worries about and probably should. And both can be difficult at once, the case where the two problems compound.

For most Tuesday sized actions only one of the two is doing any work at all, so the practical version of the distinction is to ask which one governs the action in hand.

What does the portfolio actually receive from the rule it follows?

Here is where the word tracking earns its keep. A listed pooled holding follows a stated rule, and the natural assumption is that the account receives what that rule produced. The account receives something close to it. The account gets the rule's result, less whatever it cost to deliver that result, less however far the vehicle's own holdings sat from the rule over the period. The distance between the rule's result and the account's result is what a portfolio calls a tracking gapThe distance between the result a stated rule produced over a period and the result an account holding that rule ended up with over the same period..

Two things follow immediately, and both matter more than they sound. The first is that a gap is not a criticism. Delivering any rule costs something, and a vehicle holding what the rule says at every instant would be an expensive thing to insist on. The second is that this is a portfolio level fact rather than a comment on anybody's skill. The gap is measured at the account, over a stated window, and the window has to be quoted every time, exactly as a drawdown window is.

Now make it rupees. A decimal that nobody argues about becomes an amount that somebody does. The Anantara equity sleeve is Rs 300 crore. Every 0.10 points of gap over the period is Rs 0.30 crore. A quarter of a point is Rs 0.75 crore, half a point is Rs 1.50 crore, and a full point is Rs 3 crore. The conversion is the entire reason a tracking gap is worth measuring at all. On a sleeve this size, a rounding error in the reporting is a real sum in the account.

And a gap carries a base exactly as a weight does. A full point of gap on the whole Rs 300 crore sleeve is Rs 3.00 crore, or 0.60 per cent of the Rs 500 crore portfolio. The same full point on a Rs 24 crore pooled position is Rs 0.24 crore, or 0.048 per cent of the portfolio. Until a gap is applied to a stated amount it is a rate rather than a sum of money, and quoting the rate without the amount is how a small decimal gets treated as a small consequence.

Why can two accounts pay different prices for identical exposure?

Because a listed pooled holding is bought from another holder, not from the vehicle. The account pays what somebody on the other side of the trade will accept at that moment, and there is no rule of nature saying that figure has to equal what the holdings behind one unit are worth. The traded price can sit slightly above that worth and it can sit slightly below. The phrase premium and discountThe traded price of a pooled holding sitting above or below what the holdings behind one unit are worth at the same moment. points at exactly that spread.

The portfolio consequence is the whole of what matters here: identical exposure can cost the account slightly more or slightly less depending on the moment it is bought, so the entry itself becomes part of the result. How that per unit value is computed in the first place is covered separately.

What does the portfolio give up when many lines become one?

Four things, and each is worth naming precisely. A portfolio manager either trades them away deliberately or loses them by accident. The first is the ability to exclude one specific underlying name. The second is the ability to write a constraint at holding level and have it bite. The third is the cost base of each underlying position, gone into a single entry. The fourth is visibility, and it is the one that hurts a monitoring process most: the holder who could compute concentration on either base can no longer do so from the report at all.

Watch it happen on the Anantara figures. Before, the equity sleeve is 28 inspectable rows, and anybody in the room can add up the ten largest, divide by Rs 500 crore or by Rs 300 crore, and state 31.0 per cent or 51.7 per cent with the base attached. After, the same exposure arrives as a line. Rukmini Deshpande's committee can still see how much is held. The committee cannot see what is held, so neither concentration reading is available from the document in front of it.

The same problem lands on the mandate's other two constraints, and this is where it turns from awkward to serious. The Anantara mandate forbids unlisted holdings and requires a minimum credit standing on the fixed income sleeve. Both of those are written about holdings. The wrapper containing a set of positions is not itself a position, and cannot satisfy a condition written about one. A constraint written about holdings therefore has to be tested against what the vehicle holds, never against the vehicle.

Which of these quantities carries no measured figure at all?

Quite a lot, and naming the absences is more useful than filling them. The record for this mandate carries no tracking figure, no cost figure and no price gap for any pooled holding. The conversion arithmetic therefore stands without a rate, on the same footing as the dealing cost, and for the same reason: an invented figure in the shape of a measurement is worse than an honest blank.

The record does hold a measured cost of delivery for this mandate, and the empty fields are worth setting against it. The Anantara arrangement's fees for the stated year were Rs 9.40 crore, or 1.88 per cent of the Rs 500 crore portfolio. For the same year the record also holds alpha of 1.11 percentage points, worth Rs 5.55 crore on Rs 500 crore. One delivery cost is therefore measured and known, and the unmeasured one can be held beside it without anybody having to guess at a number.

The comparison is where the arithmetic ends and the judgement would begin. Whether the Anantara arrangement was worth having depends on what an alternative would have returned and what an alternative would have cost, and the record holds no alternative at all. Going further would mean inventing the very figure that has just been left blank, so stating the comparison and withholding the verdict is the whole of the honest position.

How does a committee actually use any of this on a Tuesday?

By adding one clause to a document and one column to a report. Between them they cost nothing and settle the argument permanently. The clause goes in the mandate and says whether a pooled position is counted as one holding or looked through for the purposes of every stated limit. The column goes in the holdings report and names, on every line, the reading used to produce the percentage printed beside it.

A lender does the same thing when it looks at a borrower who has pledged a bundle of receivables rather than a single named one, and a household does it when it checks whether its emergency savings are in three separate places or in one place with three names on the passbook. In each case the question is not how much is held but what the number in view was measured against, and the fix is always a stated base rather than a cleverer calculation.

What is a listed pooled holding not?

A listed pooled holding is not a way around a constraint. A constraint written about holdings still reaches the holdings. The two liquidities move separately and neither one promises the other, so the holding guarantees liquidity in neither direction. Delivering a rule costs something even where nobody has measured what, so the holding is no free ride on a rule. And the exposure is the same size either way, with only the reporting shorter, so this is no smaller a decision than buying the underlying names directly.

The delivery route also changes very little higher up. If Rs 24 crore of the Rs 300 crore equity sleeve arrives as pooled units, the sleeve is still Rs 300 crore and equity is still 60.0 per cent of the Rs 500 crore portfolio, so the mandate's 50 to 70 per cent equity band is untouched and the policy weights are exactly where they were. The delivery route changed and the asset allocation did not. The trouble therefore belongs to the reporting and constraint questions rather than to the allocation ones.

A listed pooled holding is, said plainly, a delivery choice. The Anantara mandate can reach a spread of underlying names by buying them one at a time, keeping 28 rows and every base computable, or by buying one line and accepting that the bases move behind a wrapper. Neither route is better in the abstract. The choice has to be made knowingly, with the limit question settled in writing before the position is taken rather than after somebody notices the report cannot answer it.