How Investment Committees Govern Portfolio Decisions

An investment committee governs by settling what the manager may do before the period it governs: the policy weights, the constraints, the benchmark, the fee terms, who runs the mandate, and whether the arrangement continues. Under a discretionary mandate it picks nothing. Its hardest work is deciding how little a single year of figures can be made to support.

What does an investment committee actually decide?

An investment committeeThe group inside a holder of money that sets the terms a manager works under and then reviews what came back. does most of its real work before the period it is judging even starts, so by the time the review meeting happens the important decisions are already several months old. The meeting is where the committee finds out whether it decided well, not where it decides.

Take the invented Anantara Multi-Asset Portfolio, a Rs 500 crore mandate run for a single institutional holder. Rukmini Deshpande chairs the endowment's investment committee and Faiz Ahmad Ansari runs the mandate. Everything the committee actually settled is a closed set of six items, all of them fixed in advance: the policy weights of 60, 30 and 10 across equity, fixed income and cash; the four constraints the mandate is written inside; the composite benchmark, described rather than named; the fee terms; who runs the mandate; and whether the arrangement carries on into another period.

All six are decided before the period they govern, and a committee taking any of them afterwards is reacting to a result rather than governing a mandate. A weight chosen after a good quarter is chosen by the quarter. A benchmark chosen after the return is known is chosen to flatter the return. The order in time is what makes the six things decisions at all.

The household version. Two people agree in January what the month's income does, and write it on the back of a cupboard door. In November one of them looks at what was actually spent. If they now change the January writing so that the year reads well, they have not budgeted at all: they have described. Governing is the January act, reviewing is the November one, and the second is worthless if it can quietly rewrite the first.

The 60, 30 and 10 are policy weights, and the actual weights drift between rebalancings, so a weight quoted on a review date is not the weight the portfolio ran on all year.

What does the committee deliberately not decide?

A committee drifts across the line by accident rather than on purpose, so the list of non-decisions has to be as explicit as the list of decisions. Under a discretionary mandateAn arrangement where the holder lets the manager buy and sell inside written limits without asking permission each time. such as the one governing the Anantara Multi-Asset Portfolio, the committee does not choose which securities are held. The committee does not decide when a purchase happens. And it does not overturn any individual decision Faiz Ahmad Ansari has taken inside the four constraints.

The three refusals are not humility. Refusing all three is what makes the year afterwards readable. A committee that starts choosing holdings has taken the delegationThe transfer of a decision from one party to another, along with responsibility for how it turns out, described in writing so both sides can see where the line sits. back without amending the document, and once that happens nobody can evaluate the result. The manager returned 14.2 per cent gross over the stated twelve months, but one of the holdings was insisted on by a member of the committee in March. Whose 14.2 per cent is it? No honest answer exists. The arrangement producing the number is no longer the arrangement written down.

The everyday parallel is a household that hires a plumber and then keeps taking the spanner off him for the interesting joints. When the pipe leaks in August, the plumber is not responsible and the household cannot say whose work failed. The job stopped having a single author, and a job without a single author cannot be reviewed.

What does a review run through, and in what order?

The agenda is most of the discipline. Reading the same eight facts in two orders makes two different meetings, so the order is fixed once, in advance. Rukmini Deshpande's review packThe bundle of papers circulated before a review meeting: what was held, what came back, what it cost, which limits were tested. for the Anantara Multi-Asset Portfolio runs in that sequence, and each step is a question rather than a heading.

A meeting that reaches a conclusion first spends the rest of its time defending one, so putting the cost of delivery ahead of every conclusion is the single feature of the order that does the work. If the room has already agreed that the year was good, the fee schedule arrives as an administrative detail. If the fee schedule arrives before anybody has said the word good, it arrives as a number that has to be reconciled with the other numbers.

A conclusion already in the room makes gaps feel like tidiness rather than like missing evidence. Further down the list, then, the step asking what the record leaves out sits before the step asking what any of it means.

What did the stated year deliver, gross and net?

Now run the meeting. Every figure below belongs to one stated twelve month period for the Anantara Multi-Asset Portfolio, and none of it is annualised or extended. Faiz Ahmad Ansari reports a gross return of 14.2 per cent against a composite benchmark of 12.6 per cent, a gross excess of plus 1.6 percentage points.

Step five then asks what the delivery cost. The mandate's own commercial terms are a management fee of 1.25 per cent of assets and a performance fee of 15 per cent of the return above a 10 per cent hurdle. Both are computed rather than read off a summary sheet.

Gross 14.2 per cent less 1.88 per cent of assets leaves a net 12.32 per cent against a benchmark of 12.6, so plus 1.6 points of gross excess becomes minus 0.28 points of net excess, and the portfolio beat its benchmark while the holder did not. Both figures are correct, both describe the same portfolio and the same twelve months, and neither on its own is the excess return. Every appearance of either therefore carries the word gross or the word net in the same sentence.

What risk was carried to produce it?

Step four asks what was carried. Over the same stated twelve months the Anantara Multi-Asset Portfolio ran a volatility of 11.8 per cent against the composite benchmark's 10.4 per cent, a beta of 1.08 against that benchmark, and a tracking error of 3.7 per cent. Without a risk-free rate none of these ratios compares with anything, so the 6.5 per cent for the period is stated beside every one of them.

The worst peak to trough fall inside that window was 9.7 per cent for the portfolio against 8.1 per cent for the benchmark, and the window is quoted every time because a different window returns a different figure. Return over volatility, the measure carrying William F. Sharpe's name, works out at 0.653 for the portfolio and 0.587 for the benchmark once the 6.5 per cent is taken out of both. The information ratio on the gross excess is 1.6 over 3.7, or 0.43.

Every one of these was carried to produce the 14.2 per cent, so reading the return without them is reading half a sentence. The committee does not rebuild them in the room; the pack carries each with its period and basis attached.

How much of the excess could any decision have produced?

Step three asked for the excess and its basis, and this is where a committee either does one piece of arithmetic or skips it. The gross excess is plus 1.6 points, and some of it came from simply carrying more exposure to the same benchmark. Exposure is a setting rather than a skill, so it is not evidence about anybody's judgement.

At a beta of 1.08 against a benchmark that returned 12.6 per cent with a risk-free rate of 6.5 per cent, the return expected from exposure alone is 6.5 plus 1.08 times 6.1, or 13.088 per cent. The exposure part of that gross excess is therefore 13.088 less 12.6, or 0.488 points. The remainder, 14.2 less 13.088, is 1.112 points, and the record carries it as 1.11 points. The remainder is the measure carrying Michael C. Jensen's name.

Just under half a point, which is close to a third of the headline, came from carrying more of the same market, so a committee reading 1.6 points as a report on judgement has credited a setting it chose itself. The 60 per cent policy weight in equity was the committee's own decision, taken before the year began, so part of what the pack presents as the manager's gross excess is the committee reading back its own allocation choice.

The record for this mandate holds a second split of the same gross 1.6 points, an allocation effect and a selection effect, answering a different question on a different base. A committee that sets a term from one beside a term from the other, or adds them, has produced a figure that means nothing. Rukmini Deshpande's committee runs one split, names which one it ran, and leaves the other where it belongs.

How firm is the residual once the beta estimate moves?

The 1.11 point residual was computed at a beta of 1.08, and a beta is an estimate from a sample, so the residual inherits that estimate's sensitivityHow much an answer moves when one input behind it is changed. A result that swings a long way on a small change is reported as a range.. Move the beta and the residual moves with it, in the opposite direction and by a fair amount.

At a beta of 1.00 the whole gross 1.6 points would be residual; at 1.16 the residual falls to 0.624 points. The return, the benchmark and the risk-free rate hold at 14.2, 12.6 and 6.5 per cent across that whole ladder, and only one assumed input moved.

The residual is a range that depends on an estimate, so a committee treating 1.11 points as a hard measurement has reported a decimal place the evidence does not carry. The robust reading is the other one: across every beta on the ladder, the 1.88 per cent of fees is larger than the residual. The comparison against fees is the sentence worth putting in a minute, and the single decimal is not.

What does one stated year actually support?

The hardest item in the room is a discipline rather than a calculation. The information ratio for the stated twelve months is 0.43, struck on the gross excess of 1.6 points against a tracking error of 3.7 per cent, correctly computed and correctly labelled. The question is what it establishes, and the honest answer is almost nothing.

A year is one drawA single observation from a process that could have come out differently. A single draw shows what happened once, not what the process tends to do., one outcome from a set that could have happened, and reading a single draw as a property of the manager is the same move as concluding from one good harvest that the field is reliable. How many periods a ratio needs before it carries anything about persistenceWhether a result keeps recurring across later periods rather than appearing once. is settled in the material on appraising performance rather than here.

A committee's most valuable output is often a sentence saying that the evidence does not reach the question, and writing that sentence is work rather than an evasion of it. The sentence is also unpopular. Everybody in the room has read a number and would like it to say something. The chair's job is to hold the difference between a figure being correct and a figure being sufficient.

What did the delivery cost, set against what it bought?

The committee's sharpest moment consists of two comparisons the pack already contains. Neither needs a new figure. Both are frequently skipped, and the reason is layout rather than laziness. The fee schedule sits several sheets behind the sheet carrying the return.

The first sets the cost of delivery against the residual. Fees for the stated year were Rs 9.40 crore, or 1.88 per cent of the Rs 500 crore of assets. The residual was 1.11 points, and on the same base that is Rs 5.55 crore. The fees exceeded it by 0.77 points, or Rs 3.85 crore.

The second sets the cost against the benchmark result. The gross excess of 1.6 points is Rs 8.00 crore. Fees of Rs 9.40 crore exceed that by Rs 1.40 crore, the same 0.28 points by which the net excess sits below zero. Both comparisons are legitimate, both are the ones the committee should actually run, and both land against the holder for the stated year.

Why does the committee stop before a verdict?

The arithmetic is finished. The temptation now is to convert it into a judgement, and the committee refuses. The refusal is a governance position rather than a hedge.

Deciding whether this arrangement was worth having requires two things nobody in the room has: what an alternative would have returned over the same twelve months, and what it would have cost. The record contains no alternative at all, so a judgement made today would compare one arrangement with nothing. Comparing an arrangement with nothing is not a comparison, it is a mood.

The correct minute records that the fees exceeded both the gross excess and the residual for the stated year, states that this is a finding rather than a verdict, and names the information that would be needed before the question could be answered at all. Such a minute is not vagueness. It commits to a factual claim and to a next action, and declines only to invent the missing half of a comparison.

The household version is a person who paid a courier more than the parcel was worth, and is then asked whether a different courier would have been better. The sender cannot answer without a quote from another courier, and no amount of staring at the first invoice produces one.

Which constraints were tested, and what did each return?

Step six of the agenda is the least dramatic and the most often waved through. Four constraints govern the Anantara Multi-Asset Portfolio, and each either held or did not during the stated twelve months. The committee asks for the test result rather than reassurance, and the difference is that a test result carries a number and a base.

All four returned inside for the stated period, each as a tested number rather than an assurance. The equity sleeve sat at the 60.0 per cent policy weight inside a band running from 50 to 70 per cent. The largest single holding was Rs 23 crore, or 4.6 per cent of the Rs 500 crore portfolio, against a cap of 5 per cent written on that same base.

A pass on the 5 per cent cap is a statement about single names and nothing else, so a committee reading it as a statement about concentration has read a different sentence from the one written. The ten largest holdings together are Rs 155 crore, or 31.0 per cent of the Rs 500 crore portfolio and 51.7 per cent of the Rs 300 crore equity sleeve. Both are correct, they answer different questions, and the base has to travel with the number every time.

The single holding shows it too: Rs 23 crore is 4.6 per cent of the portfolio and 7.7 per cent of the sleeve. A committee that hears the first in one paper and the second in another, with neither naming its base, will believe two things about one holding.

The stewardship line closes the same step. Over the stated twelve months the mandate voted on 214 resolutions across its holdings and against the board's recommendation on 19 of them, or 8.9 per cent. The 8.9 per cent is a count rather than a verdict, and since the mix of resolutions is nowhere in the record the committee records the figure and refuses to read conduct into it. A high share could mean vigilance or an unusual crop of contested items, and nothing available tells the two apart.

What is missing from the pack, and where does it go?

Step seven asks the one question nobody enjoys. A review pack is a set of things somebody chose to include, and the committee's job includes noticing what was not chosen. For the Anantara Multi-Asset Portfolio the record itself names six gaps, and each becomes a named absenceA gap in the evidence written down and given a name, so somebody has to bring it back rather than quietly assume it away. on the minute.

The portfolio replaced 34 per cent of itself over the year, and no dealing cost figure appears anywhere. Nothing states which side of dealing costs the 14.2 per cent gross return sits on. The other four gaps are the categories behind the 214 resolutions, an engagement record of any kind, the cost of the derivatives overlay placed over part of the equity sleeve, whose pricing and settlement are somebody else's subject entirely, and an alternative arrangement, costed or otherwise.

Writing an absence down converts it into an action for the next meeting, and leaving it unwritten converts it into nothing at all. A gap that is spoken about in the room and not recorded has the same status six months later as a gap nobody noticed.

What makes a decision judgeable a year later?

Everything above ends in one artefact. The minuteThe written record of what a meeting decided. Its usefulness depends on how much of the thinking it captures. is the only part of the meeting that survives, and a future committee will read it without any of the context that was in the room.

A minute worth writing records four things beyond the decision itself: the reasoning, the evidence relied on with its basis attached, what would change the view, and the date along with who was present. The third of those is the one most often left out and the one that does the most work. Naming what would change the view turns a decision into something testable. If the minute says the view would change on a costed alternative showing a materially lower cost of delivery for a comparable result, a year later somebody can check whether that arrived.

A minute recording only the outcome leaves a future committee unable to tell a good decision that turned out badly from a poor decision that happened to work. A decision taken on sound reasoning with the evidence available can still be followed by a bad year, and a careless one can be followed by a good year. Only the reasoning distinguishes them, and only the minute preserves the reasoning.

The minute for this stated year, in the order the agenda ran, says roughly this. The mandate asked for equity between 50 and 70 per cent, no holding above 5 per cent of the portfolio, no unlisted holdings and a minimum credit standing on the fixed income sleeve, against a described composite. The portfolio delivered 14.2 per cent gross and 12.32 per cent net after a cost of delivery of Rs 9.40 crore, being 1.88 per cent of assets. The gross excess was plus 1.6 points; the net excess was minus 0.28 points. Volatility ran at 11.8 against 10.4, beta 1.08, tracking error 3.7 and the risk-free rate 6.5 per cent, with a worst peak to trough fall inside the window of 9.7 against 8.1. Of the gross excess, about 0.49 points is carried exposure and about 1.11 points is residual, on the exposure split rather than the other one. The fees exceeded both the gross excess and the residual. All four constraints returned inside. The voting count is 19 of 214 and is not read further. Six absences are recorded as actions. The alternative has not been costed, so no conclusion is reached on whether the arrangement continues.