Portfolio Turnover vs Transaction Cost: Act and Price
Portfolio turnover measures how much of a portfolio was replaced over a stated period. Transaction cost is what carrying out that replacement cost. One is an activity ratio and is usually published; the other is money and usually is not. The invented Anantara Multi-Asset Portfolio turned over 34 per cent in its stated year, or Rs 170 crore replaced. Nobody wrote down what the replacing cost.
Two things settled earlier carry this guide. The model portfolio sets down the Anantara Multi-Asset Portfolio, a discretionary mandate of Rs 500 crore run for a charitable endowment, and records that it turned over 34 per cent in one stated twelve month period. The risk and attribution calculator then works the cost of delivery in full: a management fee of Rs 6.25 crore and a performance fee of Rs 3.15 crore, Rs 9.40 crore together and 1.88 per cent of assets. Those fees turn a gross return of 14.2 per cent into a net 12.32 per cent against a benchmark of 12.6 per cent. Rukmini Deshpande chairs the endowment's investment committee. Faiz Ahmad Ansari runs the mandate. None of that is reworked here. One cost of delivery that has been measured is set beside another that has not, and the second of the two is the subject.
What is portfolio turnover, and over what period is it measured?
The pattern shows first on an ordinary street. A vegetable seller stocks a cart worth Rs 8,000/- and by the end of the week most of it has gone and been replaced with new stock. A hardware shop on the same street stocks Rs 8,000/- of screws and hinges and replaces perhaps a tenth of it in the same week. Both carts are worth the same. The activity inside them is nothing like the same, and the only way to say so is to count how much of the stock was swapped out and over what stretch of time.
Portfolio turnover is that count applied to a portfolio. The ratio is the share of a portfolio that was replaced over a stated period, set against the portfolio's own size. A turnover ratio contains no money at all. The number says nothing about whether the swapping was clever, expensive or worth doing. Turnover measures how busy the portfolio was, and only that.
The period is not a detail attached to a turnover figure; it is half the figure, and a turnover number quoted without its window cannot be read at all. Take the Anantara portfolio's Rs 170 crore of replacement and put it inside three different windows. Replaced over six months, that is a rate of 68 per cent a year. Replaced over the stated twelve months, it is 34 per cent a year, the recorded figure. Replaced over two years, it is 17 per cent a year. Identical trading, three answers, and a reader handed the bare number has no way to know which of the three is in front of them.
One more piece of vocabulary before the second half. When a portfolio replaces something, it sells one thing and buys another, and that pair is a round tripA sale and a matching purchase taken together. Costs are usually thought about per round trip because a replacement always involves both halves.. Turnover of 34 per cent therefore describes Rs 170 crore of round trips over the stated year, not Rs 170 crore of single orders, and that distinction becomes the unit the rest of this guide counts in.
What single thing must accompany every turnover figure before it can be read?
What is transaction cost, and what is actually inside it?
Now a different everyday scene. A household buys gold for a wedding. The rate quoted on the board is not what leaves their hands. There is a making charge, and it is printed on the bill so everyone can see it. The same shop, asked to buy the bangle back an hour later, offers less than it just sold for, and nobody itemises that gap anywhere. And if the household walks in wanting far more than the shop keeps in the drawer, the shop must go and find the metal, so it quotes a worse rate for the extra. Three separate costs. One receipt.
Transaction cost in a portfolio has exactly that shape. Transaction cost has three parts, and only the first of the three ever arrives as a bill. The first is brokerageWhat the intermediary charges for carrying out an order. It is invoiced, so it can be added up from documents that already exist.: what the intermediary charges for carrying out the order, and the one part of the three that appears on a contract note. The second is the bid and offerAt any instant a buyer can be filled at one price and a seller at a slightly lower one. The gap between the two is a real cost paid by whoever crosses it. gap: at any instant there is a price a buyer can be filled at and a slightly lower price a seller can be filled at, and whoever crosses that gap has paid it. The third is market impactThe movement in a price caused by the size of the order that is chasing it. A large order can move the price against itself before it is complete., the movement a large order causes in the very price it is chasing. The last part of the order fills worse than the first part did.
The second and third are real money. A holder who pays them is poorer by exactly the amount, in the same way and to the same extent as if it had been invoiced. Neither of the two generates a document. No line anywhere reads "gap crossed" or "price moved because of this order". Both costs have to be reconstructed by comparing what was paid against the price at some chosen instant before the order started, and that instant has already gone by the time anybody goes looking for it.
Which part of a dealing cost arrives as a bill, and which does not?
How do the two differ, and what does one show about the other?
Both sides are now defined, so they can be set against each other. Turnover is a count of activity. Transaction cost is the price of that activity. One is a ratio with no rupee in it; the other is money and nothing else. The two figures sit on different sides of a multiplication sign, and seen that way the relationship between them stops being vague.
The dragThe amount by which a cost pulls a return down. Here it is the return given up to the act of trading, expressed against the size of the portfolio. that trading places on a portfolio is the turnover multiplied by what one round trip costs. The multiplication is the whole relationship, and it is arithmetic rather than a claim about any market. A turnover figure therefore fixes one term of a product and leaves the other one entirely unknown. Knowing the turnover says nothing whatever about the cost. A portfolio that replaced a third of itself in cheap, easily traded holdings and one that replaced a third of itself in awkward, thinly traded ones report the same turnover and paid very different amounts.
Line the two up against the same criteria and they part company on every one. Parting company on every criterion is the test for whether a comparison has found a real distinction rather than two words for one thing.
| Criterion | Portfolio turnover | Transaction cost |
|---|---|---|
| What it measures | How much was replaced | What replacing it cost |
| Its unit | A ratio, per cent of the portfolio | Money, or money per unit traded |
| What it needs to be readable | Its stated period | The price at the instant of dealing |
| Where it comes from | A record of holdings that already exists | A reconstruction against a price now gone |
| Is it usually published | Yes, as one line | Often not, and rarely in full |
| Where it sits in a return | Nowhere; it is not a cost | Usually already taken out, invisibly |
| What it fails to say | What the activity cost | How much activity there was |
Two portfolios both report turnover of 34 per cent for the same stated year. What can be concluded about their dealing costs?
What is 34 per cent of a Rs 500 crore mandate in rupees?
Rs 170 crore. The working is one line: 0.34 multiplied by Rs 500 crore. The multiplication takes four seconds, and almost nobody does it. Skipping four seconds of arithmetic is why the conversion deserves a section of its own rather than a footnote.
The conversion is the whole difference between a number and a fact, and a committee reading thirty four per cent and a committee reading Rs 170 crore of replacement are not having the same meeting. Thirty four per cent sounds like a third, and a third sounds moderate, and moderate closes the discussion. Rs 170 crore does not sound like anything. Rs 170 crore sounds like an amount, and an amount invites the next question: what did moving it cost? The cost of moving Rs 170 crore is the question worth asking, and a bare percentage never prompts it.
Two honest qualifications sit on that Rs 170 crore. The first is that the base of a turnover ratio is normally the portfolio's average size over the period rather than its size on one chosen day, and the Anantara mandate did not sit at exactly Rs 500 crore on every day of the stated year. So Rs 170 crore is the right order of the thing rather than a figure struck to the rupee, and the word about belongs in front of it. The second is the round trip point from earlier: Rs 170 crore replaced means roughly Rs 170 crore sold and roughly Rs 170 crore bought, so the total value of orders that passed through the market over the year is larger than the replacement figure. Neither qualification changes the argument. Both change how confidently the number can be stated.
Turnover of 34 per cent on a Rs 500 crore mandate. How much replacement is that, and why bother converting it?
Why is one of the two published and the other usually not?
A cynical answer is tempting at this point, and the cynical answer is wrong, or at least is not needed. The asymmetry is structural. The two figures are not equally hard to produce, and the difficulty gap is enormous.
Turnover falls out of a record that somebody is keeping anyway. A holdings recordThe running list of what a portfolio held and when. It has to be maintained for settlement and reporting whether or not anybody computes turnover from it. has to exist for settlement, for reporting and for the holder to know what is held. Compare the list at the start of the period with the list at the end, add up what changed, divide by the size of the portfolio, and the turnover figure appears. Nobody had to measure anything new. The number is a by-product of paperwork that already existed.
A full dealing cost is a different kind of task. Two of its three parts have to be measured against a price that existed for an instant and then stopped existing, and the choice of which instant to measure against is itself a decision that changes the answer. There is no receipt to add up, so the figure has to be constructed, and any constructed figure carries the assumptions of whoever constructed it. The difficulty gap is structural rather than evidence that anybody is hiding anything, and it is exactly why a turnover line so often sits in a report with no cost line anywhere near it.
Why does a turnover line so often appear with no dealing cost line beside it?
What is actually recorded about the Anantara mandate, and what is not?
The position, stated flat. The record carries the 34 per cent turnover for the stated year and the Rs 500 crore size of the mandate. The same record carries no brokerage figure, no quoted gap between the two sides of a price, and no estimate of what any Anantara order moved. There is no dealing cost in the record at all.
So the honest statement of the trading drag on the Anantara portfolio is a shape rather than a value. The drag is 0.34 of whatever one round trip costs, and the round trip figure is missing. The statement is complete, correct and useful, and it carries no number in rupees or in percentage points because no such number can be produced from a turnover figure alone. A plausible round trip rate supplied at this point would be carried away by a reader and used as though it had been measured.
More survives without a rate than first appears. The relationship is a straight multiplication, so its shape is fully known even though its level is not. Double the turnover and the drag doubles. Halve the turnover and the drag halves. Take the turnover to zero and the drag goes to zero. All three statements are true at any round trip cost whatsoever, and none of them needs a rate that nobody measured.
Turnover doubles from 34 per cent to 68 per cent over the same stated year. What happens to the trading drag?
Move the turnover and watch a known shape meet an unknown price
The control moves the turnover for one stated year from nothing to a complete replacement of the portfolio. The cost of a round trip is not in the record, so the upper bar counts the trading drag in round trips, the only unit available. The lower bar is the fee drag of 1.88 per cent of assets, fixed and measured, drawn on a scale of its own so the two bars are never read as the same units. The default sits at the recorded 34 per cent, giving Rs 170 crore of replacement and a drag of 0.34 round trips.
At a turnover of 34 per cent over the stated year, Rs 170 crore of the Rs 500 crore Anantara Multi-Asset Portfolio is replaced, and the trading drag is 0.34 of one round trip cost.
How does a dealing cost differ from a fee?
The point is about where a cost sits rather than how large it is. A fee is charged beside the return. Somebody works out an amount, an invoice or a debit exists, and the holder can see the subtraction happening. The Anantara mandate's fees for the stated year are Rs 6.25 crore of management fee plus Rs 3.15 crore of performance fee, or Rs 9.40 crore, or 1.88 per cent of Rs 500 crore of assets. Because that subtraction is visible, the risk and attribution calculator could turn a gross of costsA return struck before a particular cost has been taken out. Which cost has been left in has to be stated, or the word carries no information. return of 14.2 per cent into a net 12.32 per cent and set it against a benchmark of 12.6 per cent.
A dealing cost behaves in the opposite way. A dealing cost is normally taken inside the return before anybody computes the return at all. The price paid was simply worse than the price on the screen, so the holding entered the portfolio at a worse number, so every return computed from that point onwards is already lower, and no line anywhere records the difference. One cost has been disclosed as an amount and the other absorbed into a percentage, so the two are not comparable as they are presented. A holder who compares the visible 1.88 per cent with a dealing cost of apparently nothing has not compared two costs. The comparison sets a cost against a silence.
The fees came to 1.88 per cent of assets for the stated year. Can that be compared with the dealing costs?
Is the 14.2 per cent gross return struck before or after dealing costs?
The record does not say. The return basisThe statement of which costs have already been taken out of a return figure. Without it, two returns computed on different bases look comparable when they are not. of the Anantara portfolio's 14.2 per cent for the stated year is simply not recorded, and the two readings that follow are genuinely different.
If the 14.2 per cent was struck after dealing costs, then the trading drag has already been taken. The costs went in when each holding was bought and sold, the 14.2 per cent is what remained, subtracting the 1.88 per cent of fees gives a net 12.32 per cent, and that figure is the result. Nothing further is missing and the sequence's arithmetic stands complete.
If the 14.2 per cent was struck before dealing costs, the picture changes shape. The net 12.32 per cent is then a ceilingA limit that the true figure cannot rise above. A ceiling gives the best case and says nothing about how far below it the answer sits. rather than a result, and the holder's real figure for the stated year is lower than 12.32 per cent by an unknown amount. Saying how much lower would need the round trip cost, and the record does not hold one. The net excess of minus 0.28 points against the benchmark of 12.6 per cent would, on that reading, be the most favourable version of the story rather than the story.
Closing the question would mean inventing the answer. A reader can work around an acknowledged gap and cannot work around a confident error, so an invented answer about whether a cost has already been taken is worse than an open question.
The net figure for the stated year is 12.32 per cent. Is that a result or a ceiling?
The error that gets made, and what it costs
A committee reviews the stated year. The papers show turnover of 34 per cent. Somebody says it looks moderate, nobody disagrees, and the meeting moves on to the fee line, where the numbers are precise and the discussion is lively: Rs 6.25 crore of management fee, Rs 3.15 crore of performance fee, Rs 9.40 crore in total, 1.88 per cent of assets, a gross 14.2 per cent becoming a net 12.32 per cent against a benchmark of 12.6 per cent. The review concludes. Everyone feels it was thorough.
Two things went past that room. The turnover figure was never converted, so nobody in the room ever said the words Rs 170 crore of the portfolio replaced, and that is the sentence that would have prompted the next question. And the next question is the one that matters: what did moving that Rs 170 crore cost to carry out, and is that cost already inside the 14.2 per cent or still to come out of it? Neither answer is anywhere in the papers.
The cost of the error is a review that concluded confidently on the fee line while a second cost of delivery, of unknown size and unknown position in the arithmetic, sat unexamined beside it. The fix is two sentences long. Convert every activity percentage into an amount before discussing it. And ask of every return figure which costs have already been taken out of it before comparing that figure with anything.
How does a practitioner actually use this pair?
Faiz Ahmad Ansari, running the Anantara mandate, uses the distinction defensively. When the committee asks whether the portfolio traded too much, he knows the question has two halves and that only one of them is on the paper in front of them. The turnover half he can answer immediately: 34 per cent over the stated year, or Rs 170 crore of replacement. The cost half he can only answer by naming what would have to be measured, and saying plainly that it has not been. Naming what has not been measured is a better answer than a confident guess, and it is the answer that gets the measurement commissioned.
Rukmini Deshpande, chairing the committee, uses it as a two question filter on any activity figure that reaches her, in the mandate's papers or anywhere else. First question: over what period? Second question: how much money is that, in rupees? A figure that survives both questions can be discussed. A figure that fails either one gets sent back, and the sending back costs nothing while the alternative is a decision taken on a number nobody understood.
An analyst reviewing any arrangement uses the same pair in a different order. The analyst reads the return first and asks what basis it was struck on. Two returns computed on different bases cannot be lined up beside each other, however similar they look. Then they read the turnover, convert it, and hold the resulting amount in mind as the size of the activity whose price is not in the report. Neither step produces a number. Both steps stop a wrong number from being produced.
And the same habit works far away from portfolios. Rent arrives as a demand, so a household that runs a small shop knows exactly what its rent is. Stock bought badly never arrives as anything at all, so the same household has only a vague sense of what that costs. The visible cost gets managed and the invisible one gets tolerated, purely because one of them is easier to see. Naming that pattern is most of the work here; the portfolio version is only the same pattern with larger numbers.
Where the Indian rules sit on this
Regulation, not the arrangement itself, sets what an arrangement must tell its holders about dealing activity and dealing costs, in what form and how often. The Securities and Exchange Board of India publishes the current text at sebi.gov.in, and the Pension Fund Regulatory and Development Authority publishes it at pfrda.org.in where the holder is a retirement mandate rather than an endowment. Trading arrangements and the rules under which orders are executed are published by the exchanges at nseindia.com and bseindia.com.
References
| Source | What it settles | Where |
|---|---|---|
| Securities and Exchange Board of India | What an arrangement must disclose to its holders about dealing activity and dealing costs. | sebi.gov.in |
| Pension Fund Regulatory and Development Authority | The same, where the holder is a retirement mandate rather than an endowment. | pfrda.org.in |
| National Stock Exchange of India | Where trading arrangements and execution rules are published. | nseindia.com |
| Bombay Stock Exchange (BSE) | The same, as the second place those arrangements are published. | bseindia.com |
The Anantara Multi-Asset Portfolio, the charitable endowment that holds it, Rukmini Deshpande and Faiz Ahmad Ansari are invented.
Educational material. Not advice on any investment, tax, budget or market position.