How to Select a Portfolio Benchmark and the Fee Hurdle

Selecting a portfolio benchmark means choosing what the portfolio will be judged against, before the period starts, and writing it into the arrangement. The choice has to match the mandate's own constraints, be measurable without the manager's help, and be stated with the period it covers. Where a fee also carries a hurdle, the hurdle is a second reference point and it is rarely the same number.

The question is familiar from the way a household judges a year. Money is put away for twelve months, and at the end somebody asks whether that went well. Well against what? Against the fixed deposit not taken, the neighbour who chose somewhere else, or the number that was in mind last January? Each is a different verdict on the same year, and the one reached for afterwards is usually the one that flatters the decision actually made.

A benchmark chosen after the period has been chosen with knowledge of the result, so the only benchmark that means anything is the one written down before the period started. The procedure runs in eight steps, worked here on the Anantara Multi-Asset Portfolio, an invented Rs 500 crore discretionary mandate run for a charitable endowment whose investment committee is chaired by Rukmini Deshpande and whose day to day management sits with Faiz Ahmad Ansari.

What is a benchmark actually for, and when does it get chosen?

A benchmarkThe stated thing a result is compared against, agreed before the period it will be used to judge. exists to turn a bare result into a comparison. Fourteen point two per cent for a year is a number and nothing else. Placed beside 12.6 per cent for the same twelve months it becomes a statement: this portfolio finished ahead of the thing it said it would be measured against, by 1.6 percentage points before the cost of delivery is taken out.

Lay that same 14.2 per cent beside the three other numbers the arrangement already contains and the size of the problem shows. Against the composite benchmark of 12.6 per cent the year is ahead by 1.6 percentage points, gross of every fee. Against the fee hurdle of 10 per cent it is ahead by 4.2 points. Against the risk-free rate of 6.5 per cent it is ahead by 7.7 points. Nothing about the portfolio changes across those three sentences, and only one of the three was written down before the year began as the thing the portfolio would be judged against.

The comparison has to be fixed while nobody yet knows which way it will fall. Selection therefore belongs to the drafting of the arrangement and not to the review that follows the year. There is a second reason to fix it early: a manager told afterwards has been handed a target they had no chance to aim at, so this composite was settled at the outset and Faiz Ahmad Ansari knew on day one which comparison the review would run.

Where does the choosing actually start?

Step one. The choosing starts at the mandate. Not at the portfolio, not at what the manager already holds, and not at whatever measure is easiest to look up.

The mandate for the portfolio carries four written constraints, settled long before any benchmark question arose: equity held between 50 and 70 per cent of the portfolio, no single holding above 5 per cent of it, no unlisted holdings, and a minimum credit standing on the fixed income sleeve, stated as a policy rather than as a borrowed rating symbol. The four constraints describe the space the portfolio is allowed to live in, and the benchmark has to describe the same space.

A benchmark chosen to suit a strategy has been chosen to be beaten, so the benchmark follows the mandate and never the other way round. Start instead from what the manager currently holds and look for a measure that resembles it, and the difference between the two is small by construction: the comparison confirms the portfolio rather than testing it.

What four tests does a candidate have to survive?

Step two. Every candidate that reaches the table goes through four requirements in turn, and one failure is a failure whatever it scores on the other three. There is no weighing up and no balance of merits. The four requirements are a gate with four latches.

The first is that it must be measurable without asking the manager. Somebody holding only the portfolio record and the published measure must be able to compute the comparison. A standard that only the manager can produce is a standard the manager controls.

The second is that it must be investableA property of a candidate benchmark: somebody could actually have held the thing it describes, rather than it being a target nobody could have bought.. Somebody could actually have held the thing being described. A return target picked because it sounded right is not investable: no holder could have chosen it instead of the portfolio.

The third is that it must be specified in advanceFixed in writing before the period it will judge begins, so that nobody choosing it can already know how the comparison turns out.. Specification in advance is the timing rule restated as a pass or fail on a candidate.

The fourth is that it must be consistent with the mandate's own constraints. The fourth test takes a reading of the mandate rather than a look at a screen, so the fourth test is the one that gets skipped. A portfolio forbidden from holding unlisted assets cannot honestly be judged against a measure that includes them.

Which candidates actually fall out when the tests are run?

Run the four latches on six candidates for the Anantara Multi-Asset Portfolio. A broad equity measure taken on its own is measurable, investable and can be fixed in advance. The same measure still fails the fourth badly. The mandate holds 30 per cent of the portfolio in fixed income and 10 per cent in cash, Rs 150 crore and Rs 50 crore of the Rs 500 crore portfolio, and an equity measure describes none of it.

The other four rejected candidates fail the fourth latch as well. A measure that includes unlisted holdings fails that one alone, asking the portfolio to match a result it was never permitted to pursue. The number cannot be produced without the manager, so the manager's own valuation of a peer set also fails the first. A fixed 12 per cent target the committee liked the sound of also fails the second. A measure chosen in April, after the year had closed, also fails the third.

Only the composite of 60 per cent a broad equity measure and 40 per cent a broad bond measure clears all four, and it clears them because it was constructed to describe the space the mandate permits rather than found and then justified.

How is a composite built when no single measure fits?

Step three. Most real mandates hold more than one kind of thing, and no single published measure describes that. So the arrangement uses a composite benchmarkA benchmark assembled from two or more measures in stated weights, used where no single published measure describes the mandate.: two or more measures in stated weights, combined into one number.

For the Anantara Multi-Asset Portfolio the composite is 60 per cent a broad equity measure and 40 per cent a broad bond measure. How either is put together belongs to whoever publishes it.

Put the two structures side by side and something uncomfortable appears. The portfolio's policy weights are 60 per cent equity, 30 per cent fixed income and 10 per cent cash. On Rs 500 crore those weights are Rs 300 crore, Rs 150 crore and Rs 50 crore. The composite has two legs, not three. Its 40 per cent bond leg, worth Rs 200 crore on the same base, stands opposite the fixed income sleeve and the cash sleeve together.

The Rs 50 crore cash sleeve is 10 per cent of the Rs 500 crore portfolio, and it has no counterpart of its own in the composite. The mismatch is active riskThe part of a portfolio's difference from its benchmark that comes from not matching it, whether or not anybody chose the difference. nobody decided to take. A difference will show up in the result whether or not a single decision was made during the year, purely because the two structures do not line up.

The mismatch is not a reason to abandon the composite. The mismatch is a reason to write down what the composite is made of, in the detail the portfolio's own weights get: which two measures, in what weights, and how often the weights are put back where they started.

The reset convention catches people. Left unreset, the composite's weights drift with the returns of its two legs. The portfolio's own weights drift between rebalancings for the same reason, which is set out under portfolio rebalancing. A composite whose reset convention is unwritten is a moving target, and a portfolio measured against a moving target is not being measured.

Why does the period have to sit in the same sentence?

Step four. Write the window beside the benchmark, in one sentence, so the two cannot be separated afterwards.

A benchmark with no stated window can be quoted over any window somebody later picks, and there are always windows. Over the stated twelve months this portfolio returned 14.2 per cent gross against the composite's 12.6 per cent, a gross excess of plus 1.6 percentage points. Inside those same twelve months there was a peak to trough fall of 9.7 per cent for the portfolio against 8.1 per cent for the composite, a difference of 1.6 points the other way.

Both statements are true of one year and one portfolio, and they read as opposite verdicts. Nothing about the portfolio changes between them. Only the window changes, and that is exactly why the window belongs in the same sentence as the benchmark.

The pair is easy to over-read. The two 1.6 figures are a coincidence of this invented record, not a relationship between the two quantities.

Every choice in the procedure is made before the period opens, and everything that happens after it closes is arithmetic on choices already made rather than a choice of its own.

What is a fee hurdle, and how is it not a benchmark?

Step five. A second number now enters, and it is not the same kind of thing at all.

A fee hurdleThe return a portfolio has to clear before a performance fee begins to be charged at all. is the return above which a performance feeA charge calculated as a share of the return earned above a stated hurdle, as distinct from a charge on the value of the assets. begins. For this invented arrangement the hurdle is 10 per cent and the manager takes 15 per cent of whatever the year delivers above it. The hurdle and the share above it are this mandate's own commercial arrangement, negotiated between the endowment and the manager. Neither is a market rate, an industry level, or anything a regulator sets.

The benchmark answers whether the portfolio did well. The hurdle answers whether the manager gets paid extra. Two different questions, settled by two different groups of people at two different moments, and nothing anywhere requires them to be the same number.

Think of a shop that pays its manager a bonus on sales above Rs 50 lakh for the year. The people who run the shop judge the year against what the shop next door did. Both measures are reasonable and they are about different things. In a year where the shop clears Rs 50 lakh but the neighbour clears more, a bonus and a disappointment are both correct at once.

What happens in the year the two numbers disagree?

Step six. The arithmetic itself runs in five lines.

The Anantara Multi-Asset Portfolio returned 14.2 per cent gross over the stated twelve months. The management fee is 1.25 per cent of assets. On Rs 500 crore that is Rs 6.25 crore. The hurdle is 10 per cent, so the year cleared it by 4.2 percentage points. On the same Rs 500 crore that is Rs 21 crore, and 15 per cent of Rs 21 crore is Rs 3.15 crore of performance fee. Total fees are Rs 9.40 crore, or 1.88 per cent of the Rs 500 crore portfolio. Gross 14.2 per cent less 1.88 per cent leaves a net 12.32 per cent.

The composite returned 12.6 per cent for the same twelve months. So the gross excess of plus 1.6 percentage points becomes a net excess of minus 0.28 percentage points. The portfolio beat its benchmark and the holder did not.

The third row is the one that does the teaching. After the management fee alone the holder was on 12.95 per cent, 0.35 points ahead of the composite. The performance fee of 0.63 points, triggered entirely by clearing a hurdle of 10 per cent, carried the result across the benchmark and out the other side.

Percentages hide the size of things. Say the same thing in rupees. The gross excess of 1.6 points on Rs 500 crore is Rs 8.00 crore, total fees are Rs 9.40 crore, and the holder is Rs 1.40 crore short of the composite, the same 0.28 points on that Rs 500 crore base.

The distance that produced that result is the distance between the hurdle and the benchmark, and it is 2.6 percentage points wide.

Both the plus 1.6 and the minus 0.28 are correct, and each is meaningless without the word that says which one it is. The two figures describe the same portfolio, the same twelve months and the same benchmark, and differ only in whether the cost of delivery has been taken out. No excess return in this arrangement is ever quoted without the word gross or the word net standing in the same sentence.

What would a different hurdle have done to the same year?

Hold everything fixed and move one term. The gross year stays at 14.2 per cent, the management fee stays at 1.25 per cent of assets, the manager's share above the hurdle stays at 15 per cent, and the composite stays at 12.6 per cent. Only the hurdle moves.

After the management fee alone the holder is on 12.95 per cent. Every rupee of performance fee comes off that. Since the fee is 15 per cent of the distance between the hurdle and 14.2, the net result is 12.95 less 0.15 times that distance. The net result therefore traces a straight line, and its slope is 0.15 per point of hurdle.

Three points on that line matter. At the arrangement's own hurdle of 10 per cent the fee is Rs 3.15 crore and the net is 12.32 per cent, 0.28 points behind the composite. At a hurdle of 12.6 per cent, set at the benchmark itself, the fee falls to 0.24 points or Rs 1.20 crore and the net rises to 12.71 per cent, 0.11 points ahead. The net exactly equals the composite at a hurdle of 11.87 per cent.

The shortfall of 0.28 points exists only because the hurdle was written 2.6 points below the benchmark, and this counterfactualA recomputation of what a stated record would have shown if one term had been different, used to isolate what that term was responsible for. is arithmetic on invented terms rather than a view about how any hurdle should be set. What the right hurdle is remains a commercial negotiation between a holder and a manager.

The performance fee itself moves along the mirror image of that line: Rs 10.65 crore at a hurdle of nil, or 2.13 per cent of assets, Rs 3.15 crore at the arrangement's 10 per cent, Rs 1.20 crore at 12.6 per cent, and nothing at all once the hurdle reaches the gross 14.2 per cent, where the holder keeps 12.95 per cent.

How do both numbers get written down so neither can move?

Step seven. Write them together, in one clause, with the period and the basisWhich figure a comparison or a charge is struck on, in particular whether costs have been taken out before the number is quoted..

Here is why they usually are not. The benchmark is settled by the holder's investment committee, reading the mandate. The hurdle is settled in the fee negotiation, months later or earlier, by people reading a fee schedule. Neither group is careless; they are looking at different documents on different days, and no single person is holding both numbers at once.

After the year closes the reconciliation stops being a drafting question and becomes an argument, so the reconciliation has to be somebody's job at drafting time.

Every line of that clause earns its place by being a thing that otherwise gets argued about later.

What does a benchmark never say?

Step eight, and it is the step that keeps everything above honest. A benchmark answers exactly one question, and it is a narrow one.

A benchmark shows what the same money, exposed the same way, would have done over the same window, and nothing more. The comparison does not say whether the difference was skill: return attribution splits the 1.6 points of gross excess into about 0.49 points from carrying more market exposure at a beta of 1.08 and about 1.11 points of residualWhat is left of a difference in return once the effect of carrying more or less market exposure has been taken out.. The comparison on its own gives the 1.6 and none of the split.

One twelve month period is one draw, so the comparison does not say whether the window was long enough to mean anything. The comparison knows nothing about what the delivery cost, so it does not say whether the holder finished better off either. A benchmark answers what the same money exposed the same way would have done, and every other question a holder actually has needs something else entirely.

How does a practitioner actually use this pair?

What is never a step in this procedure?

The eight steps say what to do. Four moves sit outside them, and none is a step done badly. Each one takes the comparison apart while leaving it looking intact, so none of them can be part of the procedure at any point in it.

The first is choosing the benchmark once the year's number is in. Somebody holding the result can run 14.2 per cent against every candidate on the table and stop at whichever one it beats: ahead by 1.6 points against the composite, by 4.2 against the hurdle, by 7.7 against the risk-free rate. The standard was picked by the answer, so all three are arithmetically true and none of them is a comparison.

The second is changing the benchmark to explain a result already published. A standard that moves when the number disappoints is not a standard, and the record it leaves behind cannot be read across years.

The third is picking a measure the mandate could never have bought. Picking such a measure fails the fourth latch, and the failure earns being named twice: a portfolio forbidden unlisted holdings, judged against something that holds them, is being asked to close a gap no decision inside its own limits could close.

The fourth is letting one reference point do the other's work, reading the hurdle as a verdict on the year or the benchmark as the figure a fee is struck on. All four are reached for at the same moment, once the result is known. The written clause exists to survive that moment.