Portfolio Inertia: The Cost of Never Revisiting

Portfolio inertia is a holding that changed shape because nobody looked, not because anybody chose. Seen from outside it is identical to a considered hold: in both cases nothing was bought and nothing was sold. Only one thing separates them: whether an examination happened and left something written behind. The record is what answers that.

Excess trading, the disposition effect and noise trading all measure an act. Somebody bought, somebody sold, somebody switched, and the log carries a line for it. Portfolio inertia is the opposite thing. Because its material is a set of lines that were never written, it is harder to measure. A holding can move a long way while its statement stays completely blank. The shares of a holding are worked out from values, and values move on their own. Nothing has to be done for the shape of a holding to change, so leaving it alone is not the neutral option it looks like.

What is portfolio inertia, and how is it different from choosing to stay?

Start outside money altogether. Somebody opens two shops in the same market on the same morning and puts the same capital into each. For two years she takes nothing out and puts nothing more in. By the end one shop is worth a good deal more than the other, and her wealth is no longer split half and half. Ask her why she is now more heavily in the first shop than the second and the honest answer is that she is not aware of having chosen it. Nobody moved any money. The split moved on its own.

The shopkeeper's split is portfolio inertiaA holding changing shape because nobody looked, rather than because anybody decided. in one picture. Inertia is not laziness, it is not patience, and it is not a strategy. The state it names is a holding whose shape was produced by arithmetic that nobody supervised. Set beside inertia the state it is constantly confused with, a deliberate holdLeaving a holding alone as a recorded decision.: somebody looked at the same shape, formed a view about it, concluded that nothing needed doing, and left it exactly as it was. Both shopkeepers end the two years having taken no action whatsoever. Only one of them made a decision.

Here is why the distinction earns its place rather than being word play. The two states call for completely different responses from anybody trying to understand the holding, and yet they produce identical evidence. A reviewAn examination that produces a record saying what was found. that concludes nothing needs changing is invisible unless it leaves an artefactA written record that survives the moment of looking. behind. Doing nothing is a decision only where somebody made it, and the only thing that can ever establish that somebody made it is something written down at the time. Arithmetic fills the gap between those two states while nobody is watching.

What actually drifts when nobody does anything?

The shareOne position expressed as a proportion of the whole holding. of a position is not a thing anybody sets. The share is a division: the value of that position over the value of everything held. Both the top and the bottom of that division change every day for reasons that have nothing to do with whether anybody is paying attention, so the answer changes too. Movement produced that way is driftMovement in the shares of a holding produced by value changes alone., and drift is the quiet half of the subject. Buying and selling leaves a trail. Drift leaves nothing at all.

Work the smallest possible case. Two positions, Rs 1,00,000/- each, so each is exactly half the holding. One rises to Rs 1,20,000/- and the other falls to Rs 80,000/-. The holding is still worth Rs 2,00,000/-, so the total has not moved by a rupee. The split is now Rs 1,20,000/- over Rs 2,00,000/-, or 60.0 per cent, against Rs 80,000/- over Rs 2,00,000/-, or 40.0 per cent. A twenty point change in the shape of a holding has just happened with no purchase, no sale, no switch and no entry on any statement. Nobody would call that a decision, and yet three rupees now sit in the first position for every two in the second.

One position doing better than the other happens constantly and is unremarkable. The point is that the relative weights moved, and relative weights are what people believe they control. Somebody who says they hold four things in roughly equal measure is describing a decision taken once, and whether that sentence is still true a year later has nothing to do with them.

How far did the four positions move over eight quarters?

Now take a recorded case. The Palash decision log is an invented file kept by Palash Advisory Services Private Limited, and Meera Sundaram is one of the people in it. On 4 January her holding opens at Rs 12,00,000/-, arranged as four positions of Rs 3,00,000/- each: the Vindhya index scheme, the Nilgiri mid-cap scheme, Suvarna Chemicals Limited and Kesari Logistics Limited. Four equal parts of a whole means each is 25.0 per cent of it, and that is the only thing the log ever records about shares. There is no line anywhere saying she wanted them equal, or wanted them to stay equal, or wanted anything about them at all. Four equal amounts is simply what the opening looks like.

One thing happens to the arrangement in the whole of the period. On 19 February a television segment names Suvarna Chemicals Limited and she adds Rs 1,00,000/- to it that evening, taking its cost to Rs 4,00,000/- and the total put in to Rs 13,00,000/-. Then eight quarters pass and the valuation is struck on 30 September. On the values themselves the Vindhya index scheme rose 12.0 per cent, the Nilgiri mid-cap scheme fell 15.0 per cent, Suvarna Chemicals Limited rose 15.0 per cent on the larger amount put into it, and Kesari Logistics Limited fell 35.0 per cent, so Rs 13,00,000/- put in was worth Rs 12,46,000/- at the end, down Rs 54,000/- and 4.2 per cent. Between 19 February and 30 September the log carries no instruction of any kind about these four positions, so every movement in their shares across those months was produced by arithmetic rather than by anybody.

This is the number everything turns on, so the division is worth working through. Rs 3,36,000/- over Rs 12,46,000/- is 26.9663 per cent. Rs 2,55,000/- over the same total is 20.4655 per cent, Rs 4,60,000/- is 36.9181 per cent and Rs 1,95,000/- is 15.6501 per cent. All four figures are exact, and they total 100.0000 per cent exactly. Rounding each of them to one decimal place, on its own, gives 27.0, 20.5, 36.9 and 15.7. The four rounded figures total 100.1 rather than 100.0. Nothing has gone wrong. Four numbers rounded independently do not have to add back to the total they came from, and the only mistake available here is printing the four figures and asserting that they make a hundred. The alternative is to show the smallest one as whatever the other three leave. The residual route gives 15.6 and totals exactly 100.0, at the cost of a last figure that is not the rounded value of anything. The route taken has to be stated, or the reader has to guess, and the guess will be wrong about half the time.

How much of that was the one action, and how much was drift?

A fair objection arrives here. Money was added on 19 February, so the ending shape is not purely the work of drift, and quoting 36.9 per cent as though nobody had touched it would overstate the case. The two can be separated by arithmetic that follows from the figures already set out above.

Suvarna Chemicals Limited grew 15.0 per cent on what was put into it. Suppose the Rs 1,00,000/- had never been added, so the position ran on its original Rs 3,00,000/-. The same 15.0 per cent takes that to Rs 3,45,000/-, and the other three positions are untouched by the question. The holding would then be worth Rs 3,36,000/- plus Rs 2,55,000/- plus Rs 3,45,000/- plus Rs 1,95,000/-, a total of Rs 11,31,000/-. The shares work out at 29.7082, 22.5464, 30.5040 and 17.2414 per cent, or 29.7, 22.5, 30.5 and 17.2 at one decimal place. The rounded four total 99.9, for exactly the reason set out above. Drift with no action at all still separates the four positions by 13.3 points, nearly two thirds of the 21.3 points they actually finished apart. The one decision moved the largest share from 30.5 to 36.9 and everything else in the picture arrived on its own. The counterfactual assumes the position would have grown at the same rate on the smaller amount. The assumption isolates the two effects, and the log records nothing either way.

Why is leaving a holding alone not a neutral act?

There is a comfortable feeling attached to not acting, and it deserves examining. Not acting feels like the option with no consequences, the one that can always be taken while the matter is still being thought about, the safe default that keeps every other choice open. In a great many parts of life that feeling is accurate. A house that is not repainted stays the colour it is.

A holding does not behave like the house. The shares are recomputed every day out of numbers that move on their own, so the position held tomorrow is not the position held today, whatever anybody does. There is no setting that means unchanged. Choosing not to act is a choice to hold whatever the drift produces. Drift produces a position with a shape and a size, not the absence of a position. Somebody who leaves Meera Sundaram's four holdings alone from 4 January to 30 September has, in effect, agreed to a holding weighted 36.9 per cent to one company. Somebody may be perfectly content with that weighting. Because the alternative was available every single day and was not taken, the one thing nobody can claim is that nobody chose it.

Put the same point in a household. Two earners each contribute half the income at the start. One is promoted twice in six years and the other stays where they are. Nobody sat down to make the household dependent on one salary, and yet by the sixth year it is. The dependence was not decided. Dependence of that sort arrives on its own, and only somebody who looks afresh will notice that it has arrived.

What does the standing instruction show that the shares do not?

The same holding contains the opposite case. Inertia describes how a shape arose, and it is not a complaint about anybody. Rs 25,000/- goes in every month by standing instructionAn arrangement that acts on a date without anybody taking a fresh decision.. Across the 240 logged decisions in the Palash file, 24 were pauses of a standing instruction, or 10.0 per cent. The arrangements ran as set on the other 216 occasions, or 90.0 per cent. Nobody re-decided the contribution each month. Nobody had to.

A standing instruction is not inertia, and calling it inertia would empty the word of meaning. A contribution that runs on a date was decided once, in advance, with a purpose, and the arrangement exists precisely so that it does not depend on somebody remembering. The pauses are the interesting part: 24 of them show that the arrangement was capable of being interrupted, so continuing was available for inspection every month and was allowed to continue. The contributions were decided and then automated. The shape was never decided at all. Both are true of one holding at the same moment.

So the word has to be attached to something specific. Meera Sundaram's holding shows deliberate continuation in what goes in and inertia in how it is distributed once it is in. A reader who says her behaviour is inert has said something false about the contributions. A reader who says she is on top of things has said something false about the shape. The only accurate sentence names which part of the holding it is talking about.

What does a review have to produce to count as one?

One question decides the matter. If somebody looked at the shape on a Tuesday in June, thought about it carefully, concluded that it was fine, and then went and had lunch, did a review happen? In one obvious sense yes, and in the only sense that can ever be established afterwards, no. The examination left the world exactly as it found it, including the record. Six months later nothing distinguishes that Tuesday from any other Tuesday.

So the working answer is narrow and slightly uncomfortable. A review counts when it produces an artefact. An artefact is a written thing that outlives the moment of looking, and that is a different requirement from producing an action. A line saying that the four shares were examined on a stated date, that they stood at such and such, and that the shape was judged acceptable, is a complete review that changed nothing. Such a line is also, on paper, a decision, and it can be argued with later by the person who wrote it or by anybody else who reads the file. Two sentences do the whole job. Nothing has to be bought or sold for the review to have happened.

The distinction is not a small one in practice, and it has been studied. Agnew, Balduzzi and Sunden, writing in the American Economic Review in 2003, examined how often the arrangements inside retirement accounts are altered at all and found the answer to be strikingly seldom, over long periods, for large numbers of people. Their work establishes that the untouched account is the ordinary case rather than the exception. A record showing no changes therefore tells almost nothing about the person who holds it. The absence of a written line is not evidence that nobody looked. The absence simply makes the question unanswerable, and unanswerable leaves the reader in the same practical position as knowing that nobody looked.

What does somebody reading a file for other people do with this?

Devika Rao, the adviser at Palash Advisory Services Private Limited, reads a hundred files that look identical in the one respect that matters here: no transactions. She is not judging the shape when she opens one. She is establishing which of two files she is holding. One has a note in it and one does not. A shape somebody chose can be discussed with them. A shape nobody chose has to be described to them first. Everything she does next turns on that answer.

A person deciding alone, with no adviser and no committee, faces exactly the same question with nobody to ask it. The substitute is the same one that answers every question of this kind: something written at the time. Not a spreadsheet of intentions, and not a plan. One line, dated, saying what the shares were and what was thought about them. The line costs a minute and converts an unanswerable question into an answerable one.

The pattern generalises well past holdings. A lender asks whether the concentration in a set of receivables was chosen or arrived at. An analyst reading a set of accounts asks whether the mix of revenue was managed towards or simply happened. In every one of those readings the numbers already answer what shape is this. The useful question is whether the shape was ever the subject of anybody's attention, and only a written record answers that.

Where is leaving a holding untouched exactly right?

Drift is not by itself a reason to act. A holding that has drifted does not thereby need moving. Excess trading measures what activity costs: the turnover quintiles in the Palash file run from 9 per cent a year to 210 per cent a year, and the net returns run 4.0 points apart while the gross returns sit within 0.3 points of each other. Acting more is not the lesson, and it never was.

Leaving a holding exactly where it is becomes right the moment the drift has been looked at and found acceptable, and that finding has been written down. At that point the untouched holding is a decision with a shape, a date and a reason, and it is a perfectly good decision. Somebody may look at 36.9 per cent in one company and conclude that this is fine, given everything else about their circumstances, and that conclusion is not second best to acting. A recorded conclusion is the same class of thing as acting: a judgement somebody made and can be held to.

The three states below are worth carrying away, and only the third contains any information at all.

Meera Sundaram, Devika Rao, Palash Advisory Services Private Limited, the Palash decision log, the Vindhya index scheme, the Nilgiri mid-cap scheme, Suvarna Chemicals Limited and Kesari Logistics Limited are invented.
Educational material. Not advice on any investment, tax, budget or market position.