Risk Tolerance vs Risk Capacity: Preference and Limit

Risk tolerance is how much variability an investor is willing to live with. Risk capacity is how much variability the situation can absorb before something breaks: a payment missed, a commitment unmet, a holding sold at the wrong moment. Tolerance is a preference and capacity is a constraint. Willingness cannot pay a bill, so where the two disagree the constraint decides.

Willingness as a formal coefficient is covered under risk aversion, and turning a preference into a score that ranks portfolios is covered under the utility function. The half both hand forward is the quantity that has nothing to do with preference, and what a committee does when the two disagree.

The working example throughout is the Anantara Multi-Asset Portfolio, an invented discretionary mandate of Rs 500 crore run for an invented charitable endowment. Its policy weightsThe weights a portfolio is designed to hold in each asset class. Actual weights drift away from them between one rebalancing and the next. are equity 60.0 per cent, or Rs 300 crore, fixed income 30.0 per cent, or Rs 150 crore, and cash 10.0 per cent, or Rs 50 crore. Rukmini Deshpande chairs the endowment's investment committee and Faiz Ahmad Ansari runs the mandate.

What is risk tolerance, on its own terms?

Risk tolerance is a statement about willingnessBeing prepared to accept something, as distinct from being able to survive it. Willingness reports a preference and nothing about resources or obligations.. Tolerance answers one question: how much variability in the value of a holding is this investor prepared to sit with, given what they expect in return? No amount of money, no date, no obligation, no schedule. Tolerance is a preference, and the preference can differ completely between two people looking at exactly the same portfolio.

Two households deciding how far from home to send a child to college can have identical incomes and identical savings, and still one is comfortable with a city three days' travel away while the other wants the child within a two-hour bus ride. Neither is a calculation. A tolerance statement reports the appetite for a thing, not the resources behind it.

Tolerance is about willingness and nothing else. On its own it can never establish what a portfolio can carry. An investor with an enormous appetite for variability and Rs 2,000/- to invest has a large tolerance and a small everything else. An investor who dislikes any decline at all, sitting on a fortune with no obligations against it, has a small tolerance and enormous room.

Deviations from the average are squared before being added up, so the variability tolerance refers to is symmetric. The squaring is covered separately. The symmetry matters because a tolerance statement is almost always phrased in terms of falls while the number attached to it counts both directions equally.

The Anantara record carries no minute in which Rukmini Deshpande's committee states how much variability it is prepared to sit with.

The record carries instead the variability itself, over one stated twelve month period. The assumptions put equity at a volatility of 18.0 per cent, fixed income at 5.0 per cent and cash at 0.5 per cent; from those and the stated weights the policy portfolio was designed at 11.20 per cent, and the stated year realised 11.8 per cent. The five figures set the scale a tolerance statement would be placed against.

What is risk capacity, on its own terms?

Risk capacity is a statement about the situation. Capacity answers a different question, one that never mentions what anybody wants: how much variability can this situation absorb before something breaks? Capacity is arithmetic sitting on obligations, dates and resources, and it would give the same answer if the holder were replaced tomorrow by somebody with the opposite temperament.

The word doing the work there is breaks. Breaking means one of three concrete things: a commitmentMoney already promised to somebody else, with an amount and a date attached. Not a plan or an intention. that cannot be met, a payment that has to be missed, or a holding that has to be sold at whatever price is available on the day it is needed. Every one of the three is an event with a date, and none of them is a feeling.

Take the food stall outside a single office building, with one customer base and one lease payment. Whether the operator is bold or cautious changes nothing if the office closes for three weeks and the lease still falls due. The lease has a date and an amount, and the capacity to absorb three closed weeks is a subtraction done with the same arithmetic whoever holds the pen.

Being arithmetic, capacity needs inputs, and they are boringly specific. The amount that has to be paid. The date it has to be paid on. The other resources available to pay it with. How far the rest could fall in the meantime. How long the money has to last. Give somebody those five and they can produce a capacity figure; withhold one and they cannot, however confident the answer sounds.

The Anantara record locks the portfolio in detail and the endowment barely at all. The record states the total, the weights, four mandate limits and a year of performance. The same record states no spending rate, no payment schedule, no horizon and no commitment of any size or date.

How is each one actually established?

The two sides are not established by the same kind of activity. One is asked for. The other is worked out.

Tolerance is elicitedDrawn out of somebody by asking. Its quality depends on the question, the moment and the honesty of the answer.. Somebody asks a question, presents two described outcomes and records which one is picked, or looks back at how the holder behaved the last time a sharp fall arrived. All three produce evidence about a preference. There is no rupee amount in the question, so not one of the three produces a rupee amount.

Capacity is computed. Somebody lists what has to be paid, lists when, lists what is available to pay it with, and subtracts. If the list is complete the answer falls out, and if it is incomplete no amount of asking will complete it. One side is asked and the other is worked out, and treating both as outputs of the same questionnaire is the commonest mistake in the whole subject.

A form asking how somebody would feel if this fell by a fifth is eliciting. A form asking how much is needed, and when, collects inputs for a computation. A single score mixing the two describes neither.

Several phrases circulate as though they were interchangeable, and each is closer to one side than the other.

What happens when the two disagree?

They disagree often. A preference and an obligation are separate facts and there is no reason for them to line up. The question is what a committee does on the day they do not, and the answer is not a compromise.

The smaller of the two is the limit. If willingness runs above what the situation can absorb, the situation is the constraintA limit that binds whatever anybody prefers. A written limit does not negotiate with the holder's mood. and the difference is simply unavailable. If willingness runs below, the room above is headroom that costs nothing which can be missed. Where the two disagree the constraint decides, and the reason is that willingness cannot pay a bill.

A portfolio built past the constraint does not fail on the day the preference turns out to be wrong. The portfolio fails on the day the money is needed. The fall arrives, the payment is due, and the holding goes out at whatever price the day offers. A sale made on those terms is a forced saleA sale made because money is needed on a particular date. The seller takes the price available that day., and computing capacity exists to prevent it.

Where does a mandate write capacity down?

A mandate is where capacity stops being a conversation and becomes a document. The Anantara mandate carries four written limits: equity between 50 and 70 per cent, no single holding above 5 per cent of the portfolio, no unlisted holdings, and a minimum credit standing on the fixed income sleeve stated as a policy rather than as a rating symbol. Not one of them is a statement about how anybody feels.

A written limit is capacity made durable: it survives the committee's mood, and it is why a constraint outlives the person who first felt the need for it. Rukmini Deshpande will not chair the committee for ever, and the 5 per cent cap will still be in the mandate when she does not.

The mandate bandA stated range a weight may move inside, with a floor and a ceiling. Nothing outside the two edges is permitted. converts cleanly into money: equity between 50 and 70 per cent of Rs 500 crore is a sleeve between Rs 250 crore and Rs 350 crore, with the policy weight of 60 per cent at Rs 300 crore. The three rupee figures are what a committee agrees to when it agrees a range in per cent.

Only the first of the other three converts into rupees: 5 per cent of Rs 500 crore is Rs 25 crore, so no single holding can put more than that at risk. Each of the three names a way the portfolio could break rather than a level of comfort anybody reported.

Watch the base. The largest holding is Rs 23 crore. The same holding is 4.6 per cent of the Rs 500 crore portfolio and 7.7 per cent of the Rs 300 crore equity sleeve. Both readings are right and they answer different questions; the cap is written against the portfolio, so the portfolio reading is the one that tests the limit. Moving between the two without saying which is which has misled a reader about concentration.

What does the mandate band come to in rupees?

One multiplication turns the band from a range in per cent into something a trustee can argue about.

Take a fall of 10 per cent in equity, as an illustration and not a forecast. At the floor the sleeve is Rs 250 crore and the fall is Rs 25 crore; at the policy weight of 60 per cent, Rs 300 crore and Rs 30 crore; at the ceiling, Rs 350 crore and Rs 35 crore. The band is therefore a Rs 10 crore range of exposure to one specified fall, and a band is a capacity statement expressed in weights rather than in rupees.

As a share of the whole portfolio those three are 5.0, 6.0 and 7.0 per cent, so twenty percentage points of equity band is two percentage points of portfolio against this fall.

Hold the weight at the policy 60 per cent and vary the fall instead: on a sleeve of Rs 300 crore, falls of 5, 10, 20 and 30 per cent are Rs 15 crore, Rs 30 crore, Rs 60 crore and Rs 90 crore, or 3.0 to 18.0 per cent of the whole Rs 500 crore.

What did the stated year actually show?

The band is what the mandate permits; what happened is a separate record. The worst drawdownThe fall from the highest point a portfolio reached to the lowest that followed, inside a stated window. was 9.7 per cent, measured peak to trough inside that window. On Rs 500 crore that comes to Rs 48.5 crore. The composite benchmark fell 8.1 per cent over its own worst stretch, or Rs 40.5 crore on the same base, a gap of Rs 8 crore.

Peak to trough inside a different period gives a different answer from the same price history, so a drawdown figure with no window attached is not a figure at all.

The drawdown establishes something narrow. The endowment lived through a Rs 48.5 crore fall and the mandate held, and that is evidence about capacity rather than about tolerance. Nothing was breached, nothing had to be sold to meet a payment, and the band was not tested at either edge. The record is silent on what anybody was willing to sit with while it happened.

Establishing capacity properly needs the endowment's spending commitments and their timing, and the record states none. So no coverage figure can be produced, no spending rate, no number of years of cover, and no substitute for any of the three. A manufactured commitment would make every figure downstream of it uncheckable, so refusing the calculation the reader most wants is the correct answer.

Why does the moment of measurement matter so much?

Because the two sides move at completely different speeds, and a process that ignores that difference will record the calendar and think it has recorded the investor.

A tolerance score taken immediately after a strong run reads high. The same holder, asked three months after a sharp fall, reads low. Between the two askings no commitment was added or removed, no date moved and no resource appeared or vanished. Capacity barely moves over a window in which tolerance swings hard, so a mandate rebuilt on a freshly measured tolerance has recorded the timing rather than the preference.

The damage is not that one reading is wrong; it is that each revision compounds the one before it. Rewritten on each fresh score, the recorded weight goes from the policy 60 per cent to 68 after a strong period, to 52 after a sharp fall, settling at 62 a year later. Against the same 10 per cent equity fall that is Rs 30 crore, Rs 34 crore, Rs 26 crore and Rs 31 crore of exposure, and all four settings are inside the band, so nothing is ever breached.

No limit was crossed in any of the four. The portfolio was simply rebuilt three times on readings that reflected the previous few months, and each rebuild carried its own trading. The record notes turnover of 34 per cent over the stated year, and turnover carries a cost no return figure shows.

Can two households share a view and differ on capacity?

Yes, completely. Put two households side by side. Both have Rs 10,00,000/- saved and both hold the same opinion about where the market is going, so asked the same questions they give the same answers, and their tolerance is identical by any measurement that could be run.

The first household runs on one salary of Rs 60,000/- a month and has promised Rs 8,00,000/- towards a wedding four months from now. The second runs on two salaries, Rs 60,000/- and Rs 55,000/-, with nothing promised to anybody for years. Let the same 30 per cent fall arrive: both savings pots go to Rs 7,00,000/-. The first household is short by Rs 1,00,000/- on a date it cannot move. The second is short of nothing at all.

The two households may have the same tolerance and completely different capacity, and no amount of discussing the market moves either one closer to the other. One salary is one source and one failure point; two salaries with nothing promised is a different structure, and it was there before the fall arrived and will be there after it goes.

The difference between the two households is a line across the scale at Rs 8,00,000/- with a date attached, drawn long before the fall arrived. The dated line is capacity, and it is the only reason the two outcomes differ.

What does neither of them establish?

Neither figure says whether any portfolio is well built. Neither says what anybody should hold. Neither is a substitute for the other, and no arithmetic converts one into the other in either direction.

A capacity figure is a ceiling: beyond this line, something breaks. The ceiling says nothing about where under the line anything should sit, and a holder far below their ceiling has expressed a preference rather than made an error. A process that collects only one of the two has not simplified the work, it has skipped half of it, and usually the computable half.

High willingness against low capacity is the conflict everybody pictures. The other three combinations turn up more often, and in three of the four cells the interesting question is settled by neither figure.

Route any single statement about a holder by asking whether a date could be attached to it. Statements with dates and amounts are capacity; statements about feelings and preferences are tolerance. The date test is crude and it works often.

Both sides are now defined, and every row of the contrast below was established above rather than asserted here.

How does a practitioner actually use the split?

Faiz Ahmad Ansari, running the Anantara mandate, uses the split as an ordering rule. When the committee asks for a change in the equity weight, the first check is whether it sits inside the written band rather than whether the request is reasonable. The band check is objective and takes a second. Only inside the band does the conversation about preference begin.

Rukmini Deshpande, chairing, uses it the other way round. Her job is to keep the written limits matched to the endowment's obligations, so the question she brings to a meeting is never whether the committee is comfortable but whether anything has been committed since it last looked. A new commitment with a date on it changes the constraint. A strong quarter does not.

An analyst reviewing a mandate from outside uses the split as a two-question filter: where in this document is the capacity written, and what did the process do to establish it? Four crisp limits with no record of the obligations behind them is somebody's comfort called a constraint. The presence of limits proves nothing; what proves something is whether anything was subtracted to arrive at them.

When a mandate is reviewed after a bad period, the split says which part of the review is which. A fresh reading of how the committee now feels is a tolerance revision; a change in what the endowment has promised is a capacity revision. Both can be reasonable, and a review that does not say which it is doing leaves the reader unable to tell whether the mandate was corrected or merely re-timed.

When does the distinction stop mattering?

The two usually diverge. The more useful question is when they do not. In those conditions the second measurement changes nothing anybody decides. There are four such conditions.

The first is capacity so far above anything the mandate permits that tolerance is the only live constraint. The most the Anantara band can expose to a 10 per cent equity fall is Rs 35 crore, at the 70 per cent ceiling. If the endowment's obligations were small enough that Rs 35 crore left every commitment met on its date, nothing inside the band could reach the capacity limit, and everything between 50 and 70 per cent would be preference.

The second is the mirror of it. Where capacity is tight enough that every allocation on offer breaches it, no tolerance can be acted on: the first household, with Rs 8,00,000/- due in four months against Rs 10,00,000/- of savings, arrives at the same place whatever it says on a questionnaire. No room is left for the answer to be spent in.

The third is a horizon short enough that both collapse to the same answer. With the money needed next month a fall has no time to reverse, so the date settles the allocation and there is nothing left for a preference to settle differently. The fourth is a portfolio small against the holder's other resources: the second household, on two salaries with nothing promised, meets the same 30 per cent fall and reaches neither limit.

None of the four announces its own expiry. A capacity that was ample stops being ample the day the liability it was measured against moves, and nothing in the portfolio record marks that day: let the endowment commit Rs 8 crore payable in March and the first condition has ended. Every figure in the mandate still reads as it did the week before. The absence of any marker on that day is why the distinction is worth keeping even where it currently decides nothing.

The Anantara Multi-Asset Portfolio, the charitable endowment that holds it, Rukmini Deshpande, Faiz Ahmad Ansari, the composite benchmark and both households are invented.
Educational material. Not advice on any investment, tax, budget or market position.