Asset-Liability Management: Repricing, Gaps and Margin

Asset-liability management is a bank deciding the shape of its two sides rather than their size: when cash arrives against when it is owed, and when the rate on an asset resets against when the rate on a liability resets. Liquidity and margin are separate questions. The first decides whether the bank can pay. The second decides what its margin does when rates move.

The Reserve Bank of India sets what a bank must hold in liquid form, what it must fund stably, how the rate exposure on its banking book is measured and reported, and which time buckets it must report its gaps in. The Reserve Bank revises all of that on its own schedule and publishes the current text. Suvarna Commercial Bank Limited, an invented bank, supplies every figure worked through below. Its year is arithmetic that reconciles to the rupee and reaches no further than itself.

What is asset-liability management actually managing?

Start by clearing away the thing it is not. Asset-liability management is not about how big the bank is, how much it has lent, or how many deposits it has taken. Size, lending and deposits are already decided by the time anybody sits down to manage them. The two sides exist. Money has been taken in and money has been put out. The shape of those two sides is what is left to decide, and shape here means dates.

Every single item on a bank's balance sheet carries two dates, and this is the sentence the whole subject grows out of. The first is maturityThe date on which money actually changes hands: the day a deposit must be returned, or the day a loan is repayable., the date money changes hands. The second is the date the rate on that item is allowed to change. A bank that reads only the first has a view on whether it can pay and no view at all on what it earns. A bank that reads only the second has the reverse. Both schedules exist for every rupee on both sides, and the two schedules do not agree with each other.

Here is the everyday version, and it is worth holding on to because the finance version is the same shape. A household pays rent on the first of the month and the salary arrives on the seventh. The income is comfortably larger than the rent. Nothing about that household is unaffordable. The household has a shape problem for six days every month, and the fix is not more income but a different arrangement of dates. Now scale that up to an institution where the dates are in millions of contracts and where being six days short is not a private inconvenience.

Notice the verdict the drawing withholds. The mismatch drawn there is not a mistake. Both items in it are perfectly ordinary and were priced deliberately. Managing the two sides is a very different job from trying to remove the mismatches. The work is choosing which ones to carry and knowing in advance what each one costs if the world moves.

What is a liquidity gap, and what does it not show?

Pick a window of time. The next month, say, or the next three months. Now ask one narrow question: how much cash arrives from the asset side inside that window, and how much cash is owed from the liability side inside the same window. The difference between the two is the liquidity gap for that window. Positive means more arrives than is owed. Negative means the bank has to find the rest somewhere.

A liquidity gap answers exactly one question, whether the bank can pay in that window, and it says nothing whatever about what the bank earns. The narrowness is not a limitation to apologise for. Answering exactly one question is the purpose of a liquidity gap. A household can know with total confidence that it will get through the month and have no idea at all whether it will be better or worse off by the end of the year. Paying through the month and being better off by the year end are different enquiries, and they use different information.

Two things go into it and nothing else: money arriving and money owed, each on its contractual date. An advanceThe reported line for money a bank has lent out. that repays next month puts cash in the window. A deposit that matures next month takes cash out of it. A rate does not move money on a date. A rate only changes the size of an amount that moves anyway. So what either item is charged or paid has no bearing on the count at all.

What is a repricing gap, and why is it a different question?

Take the same window. Ask a completely different question of it: how much of the asset side has its rate reset inside that window, and how much of the liability side does. The difference is the repricing gap. The window is identical, the balance sheet is identical, and the answer describes a different world.

Two words carry the whole distinction. An item on a fixed rateA rate contracted not to move for a stated period, whatever happens to rates elsewhere in the market during that period. does not appear in any repricing window until its stated period ends, however soon it matures. An item on a floating rateA rate contracted to move with something else, so it changes on its reset date whether either party wants it to or not. appears in the window its reset falls in, however far away its maturity is. A deposit contracted for three years can reprice next month, and an advance repayable next month can carry a rate fixed for its whole life, which is why merging the two gaps is the commonest confusion in this whole subject.

What does one percentage point on the funding side cost this bank?

The order matters, so name the absence before the arithmetic rather than after it. Suvarna Commercial Bank Limited's record carries no maturity buckets and no repricing profile for either side, and nobody can therefore say how much of its funding actually reprices inside any window. The arithmetic below rests on a stated scenarioA stated set of assumptions worked through to see what would follow, rather than a measurement of what is actually there., chosen because it isolates one side completely, and it is labelled that way at every single use below.

The scenario: the entire deposit base reprices upward by 1.00 percentage point for the year, and nothing on the asset side moves at all. Deposits are Rs 1,92,000 crore, so interest expendedThe reported line for what a bank pays out on the money it has taken in, over a stated period. rises by Rs 1,920 crore, from Rs 11,160 crore to Rs 13,080 crore for the year. Interest earned stays exactly where it was, at Rs 18,600 crore for the year. Net interest incomeInterest earned less interest expended over the same period. It is a difference between two reported lines, not a line anybody collects. therefore falls from Rs 7,440 crore to Rs 5,520 crore for the year. The fall is 25.81 per cent.

One percentage point on the funding side took a quarter of the year's net interest income, and the asset side never moved. Nobody defaulted. No loan went bad. Not one borrower was lost and not one deposit left. The bank has the same customers, the same book and the same everything, and a quarter of the difference it lives on has gone.

A margin without its base is not a number anybody can use. Read the margin the same way, and name the base every time. Net interest income of Rs 7,440 crore for the year on earning assetsAdvances plus investments: the part of the balance sheet that is put out to earn. It is the base a bank's margin is struck on, and it is smaller than total assets. of Rs 2,04,000 crore is 3.65 per cent for the year. After the stated reprice, Rs 5,520 crore on the same Rs 2,04,000 crore of earning assets is 2.71 per cent for the year. Same base, same period, both readings printed, and the whole distance between them was created on the funding side.

At plus 2.00 points the fall in net interest income for the year is 51.61 per cent, so slightly more than half the difference the bank lives on has gone for a move that a market can deliver inside a single cycle. Nobody can say in advance which of those four settings a market delivers, so the table shows the shape of the relationship rather than where this bank would land on it.

Why does so small a move land so large?

The reason so small a move lands so large is arithmetic rather than banking. Net interest income is a small difference between two large numbers. Interest earned is Rs 18,600 crore for the year and interest expended is Rs 11,160 crore for the year, and what the bank lives on is the Rs 7,440 crore between them. Change either of the two large numbers by a small proportion and the small difference changes by a large proportion, every time, with no exceptions and no judgement involved.

Put the two bases beside each other and it becomes obvious. Deposits of Rs 1,92,000 crore are 94.12 per cent of the Rs 2,04,000 crore of earning assets that the margin is struck on. Net interest income of Rs 7,440 crore is 3.65 per cent of that same Rs 2,04,000 crore for the year. The funding side is nearly as large as the base the margin is struck on. The margin itself is a sliver of that base, so a point on the funding side is very nearly a point off the margin.

The street version takes ten seconds and it is the same arithmetic. A stall buys a crate at Rs 90/- and sells it at Rs 100/-, so the day is worth Rs 10/- a crate. The buying price now rises by Rs 5/-. The rise is about five and a half per cent in what the stall pays, and it has just removed half of what the stall makes. Nobody stopped buying. Nothing was mispriced. The margin was thin, and thin margins magnify everything that touches either side of them.

Where does the Rs 1,920 crore actually go?

Reading the scenario as destruction is tempting, as though Rs 1,920 crore had evaporated somewhere between the two bars. Nothing evaporated. Interest expended by a bank is interest received by whoever funded it, to the rupee and on the same date. The bank's cost is somebody's income, and the bank's side is the side printed here because the subject is a bank.

Every figure printed here is one side of a pair, and each pair nets to zero. The pairing is what makes the directional question later so hard to answer in general. A rate rise is not a loss to the system. A rise is a transfer inside it, and which institution is on which side of the transfer depends on the shape of its two schedules rather than on whether it is a lender at all.

Does this bank gain or lose when rates rise?

The question cannot be settled for this bank, and that limit is itself the teaching. Whether a lender gains or loses when rates rise is decided entirely by which side reprices first and by how much of each side reprices at all. The scenario worked above moved one side because moving one side is how a mechanism is made visible. A real market move touches both, on their own schedules, and the net effect is the difference between two repricing patterns that this record does not carry.

There are no maturity buckets and no repricing profile anywhere in this record, so the directional question cannot be answered for this bank and no amount of care with the figures that do exist will produce it. A reader should instead find the repricing profile before forming any view on what a rate move does to a lender, and if it cannot be found, treat that absence as the answer rather than as an obstacle to one.

The two tables that would settle it are the maturity schedule and the repricing schedule, bucket by bucket, on both sides. The maturity schedule and the repricing schedule are the tables this entire subject runs on. Suvarna Commercial Bank Limited's record holds neither, so the forms below are drawn empty, with the address of the place they come from inside the blank.

Which Indian rules decide the values in the empty rows?

What can be reshaped, and what can never be?

A bank is not helpless in front of its own schedules. Anybody who thinks the shape is fixed will misread every treasury decision a bank ever makes, so the four honest levers are worth naming. A bank can lengthen its funding, and then less of what it owes can leave or reset soon. A bank can shorten its assets, and then more cash returns sooner. A bank can move items between fixed and floating terms as they are written, changing the repricing schedule without touching the maturity schedule at all. And a bank can hold more of its assets in a form that can be turned into cash quickly.

Funding long assets with short money is the service rather than the defect, so no bank can remove the mismatch. A bank that matched every date on both sides perfectly would have stopped doing the one thing an economy needs it for. Savers want their money back soon and borrowers want it for years, and standing between those two wishes is the job description. The levers change how much mismatch is carried and where it sits. None of them makes it go away, and a bank claiming otherwise would have become a different institution.

What does the published balance sheet leave out?

A balance sheet publishes the sizes and none of the dates, and the dates are exactly what decides the answer. The published sheet shows Suvarna Commercial Bank Limited with deposits of Rs 1,92,000 crore, advances of Rs 1,44,000 crore and investments of Rs 60,000 crore, and not one line says when any of it reprices or when any of it comes back.

Which produces the uncomfortable fact underneath the whole subject. Two banks could publish identical assets, identical deposits and identical margins on identical bases for the same year, sit on opposite sides of a rate move, and look indistinguishable to anybody reading only what they published. One of them is about to have a good year and the other a bad one, for reasons that were entirely knowable in advance by anybody holding the repricing tables and entirely invisible to everybody else.

What is missing here, and how much of it can still be bounded?

An absence is not one thing. Some missing figures can be squeezed from both sides until only a narrow strip is left, some can only be capped, and some cannot be constrained at all by anything in the record. Sorting them is worth doing. A reader who treats every gap as equally unknowable gives up information they actually have, and a reader who treats every gap as fillable invents.

Every absence encountered here, each one classified and each one with its proof beside it. The distance between two endpoints is a summary and the endpoints themselves are the fact, so each row prints both endpoints rather than the distance.

How does anybody use this once they are out of a classroom?

Somebody reading a lender before a rate cycle opens the repricing table first and the profit and loss second. The reversal of the usual order is deliberate. Two numbers decide most of the story: what share of the asset side reprices within a year and what share of the funding side does. If the second is much larger than the first, a rise squeezes the margin before it helps it. If the first is much larger, the rise arrives as good news first. Everything else is detail on top of that one comparison.

Inside the bank the same table is a work list rather than a view. A treasury team looks at the window where the gap is widest and reaches for one of the four levers, and each lever has a price attached: funding that cannot leave soon costs more than funding that can, and assets that return cash sooner usually earn less than assets that do not. Managing the two sides is a series of paid choices about shape, not a search for a shape with no cost.

And the same arithmetic reaches a household directly. Most readers do not expect that part. A household carrying a home loan on a floating rate while holding a deposit at a fixed rate for three years has a repricing gap of its own, and it is pointing the wrong way. Rates rise, the outgoing moves within months, the incoming does not move for years, and the squeeze arrives in the household budget by the same arithmetic that ran through the bars above. The same levers are available too: fixing the loan lengthens the repricing on what is owed, and shortening the deposit shortens it on what is held.

What is worth keeping about the two schedules?

Four things, and they survive the arithmetic being forgotten.

1. Every item on both sides carries two dates: when money changes hands, and when the rate may change. Reading one gives half a position.
2. A liquidity gap answers whether the bank can pay in a window. A repricing gap answers what happens to the margin when rates move. Neither answers the other.
3. Net interest income is a small difference between two large numbers, so a move measured in fractions of a point on either large number arrives magnified in the difference.
4. Whether a rate rise helps or hurts a lender is decided by which side reprices first. A balance sheet does not publish that. Where it cannot be found, the direction is unknown and saying so is the honest answer.