How to Read RBI Liquidity and Money-Market Data

Read a liquidity and money-market release in a fixed order: the net position of the system first, then the overnight rate, then that rate's place inside the corridor, then the attribution of any change, then one money measure at a stated width, then credit beside nominal output. The order matters because a rate cannot be interpreted before the position is known.

Four things this sequence needs are already built, and each of the four is treated in full elsewhere. Step one reaches for the first: a net position across the whole system rather than in any one place. The second is the corridorThe band with a floor and a ceiling that an overnight rate is expected to trade inside. What sets it, and why it has ends at all, is covered separately under the policy rate. that step three places a rate inside. The third is the money marketThe place where very short-dated borrowing and lending happens, usually returned within a day or a few days. What trades there and who lends is covered separately. that step two takes a rate from. The fourth is the set of money measures that step five picks a width out of. The sequence adds one thing only: the order those four are taken in, and what each step is required to write down before the next one is allowed to start.

Before any of it, one restriction, and it governs every figure below. The Reserve Bank of India (RBI), its liquidity operations and the Liquidity Adjustment FacilityThe standing arrangement through which a central bank in India conducts its daily liquidity operations. What it is and how it works is covered separately. are named as the place a real reading starts. Every number that follows belongs to the Republic of Sankhya, an invented country, and is labelled so wherever it appears. The sequence is the transferable part; the figures are only something for it to run on.

Why does the order of the reading decide what can be said at the end?

The whole sequence hangs off an everyday version, and that version is worth holding on to. A household electricity bill has gone up by Rs 600/- this month. Somebody asks why. There are exactly two ordinary answers: the tariff went up, or the household used more. A reader who never looks at the tariff will attribute the whole Rs 600/- to the air conditioner, and will be confidently wrong about a household lived in every day. The order in which the two are checked decides whether the sentence said at the end is an attribution or a guess.

A liquidity release has the same shape. The overnight rate moved. There are two ordinary causes: the corridor moved, or the position of the system moved inside it. A reader who takes the rate first has nothing to hold the rate against, and will reach for the cause that sounds most like news. The policy cause is almost always the one that sounds most like news. A reader who takes the position first arrives at the rate already holding the scale that makes the rate interpretable. The order is not a preference and it is not a tidiness rule: it is the difference between producing an attribution and producing an impression.

Every step produces something written down, and the next step uses only what the last one produced. Written outputs are what make the sequence checkable by somebody who was not in the room: a reader can ask for step three's output and see whether it exists. A step that produces nothing has not been taken, whatever the reader believes about it.

What is taken first, and what does it produce?

Step one takes the net position of the whole system and produces two things: one figure, and one word. The figure is an amount in Rs crore. A sign is the easiest thing to read past, so the word is either surplus or shortage, written out in full rather than left to a plus or a minus. On the Sankhya release worked through below, step one produces a shortage of Rs 40,000 crore. One figure and one word are the entire output of step one, and the output is small.

Notice what step one does not do. Step one does not say whether a shortage of Rs 40,000 crore is large, whether the shortage is unusual, or what anybody ought to do about it. Step one records a position and states its direction in words, and everything else about what a position in the system is belongs to the treatment of system liquidityThe net surplus or shortage of funds across the whole system taken together, rather than at any one place inside it. How it arises is covered separately.. A step one output that has started to contain the word tight has stopped being step one.

Why is the overnight rate written down before it is interpreted?

Step two takes the overnight rateThe rate at which very short-dated money is borrowed and returned the next working day. Where it comes from and who trades at it is covered separately. and produces one number to two decimals. On the Sankhya release that number is 6.10 per cent. Then step two stops. Step two does not say higher, does not say tighter, and does not compare 6.10 to anything at all. Stopping there feels wrong to almost every reader, and the discomfort is the point worth sitting with.

The discipline is worth the discomfort for one reason. A rate on its own is a number without a scale, in exactly the way that a temperature of 37.8 degrees is a number without a scale until it is known whether the person has been sitting still or has just climbed four flights of stairs. The stairs are the position. A reader who takes the rate first will supply the missing scale from whatever is handy. Whatever is handy is usually the last policy headline read. The scale that makes a rate interpretable was produced at step one and gets applied at step three, so step two writes the rate down and refuses to interpret it.

What does placing the rate inside the corridor add?

Step three takes the bare rate from step two and places it inside the corridor, producing three things: the distance to the floor, the distance to the ceiling, and a yes or no on whether the rate is sitting on either end. Sankhya's corridor runs from a floor of 5.75 per cent to a ceiling of 6.25 per cent, with the policy rate at 6.00 per cent in the middle. The rate of 6.10 per cent sits 0.35 above the floor and 0.15 below the ceiling, and it is on neither end. Step three's output is those three items and nothing further.

The third item is the one readers skip, and it changes what the other two mean. A rate resting exactly on the ceiling is no longer reporting where the system settled; it is reporting where the system was stopped. A rate sitting at an end of the corridor is reporting a boundary rather than a position, so the clamp check is written down at step three whether the answer is yes or no. The corridor itself, the reason it has ends, and the forces that hold a rate inside it are covered separately under the policy rate and its corridor. Step three uses the corridor as a ruler and moves on.

How is a change split between the corridor and the position?

Step four takes the change since the last reading and splits it into named parts. There are two named causes. The first is the corridor part: the amount of the change explained by the corridor itself having moved. The second is the position part: the amount explained by the system's net position having moved inside a corridor that stayed where it was. Then comes the third slot, and it is the one this step exists for. Whatever is left over after both named parts have been taken out is the remainder, and the remainder is written down as unattributed.

On the Sankhya reading, the rate is up by 0.10 against the last reading. The policy rate held at 6.00 per cent and the corridor did not move, so the corridor part is nil. The system moved from balanced to a shortage of Rs 40,000 crore, and that move accounts for the whole 0.10. The remainder is nil. A nil corridor part, a position part of 0.10 and a nil remainder make a clean attribution, and the attribution is written as three lines, not one.

Now the case that matters more. Suppose the rate had come in at 6.18 per cent instead. The corridor part is still nil, the position part is still 0.10, and 0.08 is left over with nowhere to go. The pull at that moment is enormous: round the position part up, or call the extra 0.08 a policy effect, and the line balances. A remainder pushed onto a named cause stops being a remainder and starts being a claim about why something happened, so a change that cannot be attributed is reported as unattributed rather than assigned to the likelier cause. The reading is allowed to end with an unexplained 0.08 in it. A reading that never has one is not more rigorous; it is just quieter about its guesses.

Which width of the money measure was used?

Step five reads one money measure and produces three things: the level, the width it was read at, and the growth rate read at that same width. Sankhya's widest measure, M3The widest of the four standard money measures. What each of the four contains, and why they are ordered the way they are, is covered separately., stands at Rs 18,00,000 crore, and that is what this reading uses, so the output reads: the money stock, at the M3 width, Rs 18,00,000 crore. The width is not a footnote to that sentence. The width is part of the number.

The everyday version is a vegetable seller quoting a price of Rs 40/- without saying whether that is per kilo or per bunch. The number is real, the seller is honest, and the price still cannot be used. A money level quoted without its width has the same defect, and the defect is worse here because there are four widths in ordinary use and they differ by a factor of five between the narrowest and the widest. A growth rate taken at one width and reported beside a level taken at another is the specific error step five exists to prevent, so the width is written down beside every number rather than assumed from context.

One honest note about the third output on this particular run. The Sankhya reading worked below carries the level and the width, and the growth rate at that same width is not carried on this reading, so it is written down as not carried. The growth rate is not filled in from a different width and not estimated from the credit series. The blank goes into the reading exactly the way the unattributed remainder at step four does. The contents of each of the four widths, and the reason they are ordered as they are, are covered separately under the money measures; step five names a width and moves on.

What does credit growth need beside it?

Step six reads credit growth beside nominal outputOutput measured at the prices actually paid, without stripping out price changes. How the measure is built is covered separately. growth, never alone, and produces two growth rates and the ratio between the two stocks. On the Sankhya reading, credit grew 10.00 per cent, from Rs 14,00,000 crore to Rs 15,40,000 crore. Nominal output grew 4.00 per cent, from Rs 17,47,200 crore to Rs 18,17,088 crore. Credit went from 80.13 per cent of output to 84.75 per cent, a rise of 4.62 points.

The ratio read on its own hides both of its parts. Both parts moved. Credit rose by Rs 1,40,000 crore and output rose by Rs 69,888 crore, and the ratio went up because the first rose faster, not because the second stood still. A credit ratio that has risen shows the two parts moved at different speeds and shows nothing about which of them did the moving, so step six writes down both growth rates and never the ratio on its own. A shop whose sales grew 10 per cent in a year when every shop on the street grew 4 per cent has a different story from a shop that grew 10 per cent while the street stood still, and the ratio alone cannot say which of the two streets is in question.

What would change this reading before the next one arrives?

Step seven writes down two or three named conditions that would change the reading, and it writes them now, before the next release, rather than afterwards. On the Sankhya reading the conditions are these three. First, the corridor moves and puts a corridor part into the next attribution where there is currently none. Second, the net position crosses from a shortage into a surplus or back the other way, and then the word in step one's output changes, not only the figure. Third, the overnight rate reaches an end of the corridor and stays there. Step three's clamp check would then answer yes, and the two distances would stop meaning what they mean today.

Writing the conditions before the next release is what makes them worth anything. A condition written afterwards is a description of what happened dressed up as a check. Step seven is dated by the reading rather than by the event, and that dating is the only way to tell later whether the reading anticipated the change or merely survived it.

When does the reading stop?

The reading stops when seven outputs have been written down: the net position with its word, the overnight rate, its place inside the corridor with the clamp check, the attribution with its remainder, one money measure at a stated width, the two credit figures with the ratio, and the named conditions. Those seven outputs are the stopping point, and the stopping point is a step rather than a feeling. The list is checked line by line, and when each line has something written on it the reading stops.

The stopping point is not a verdict. The sequence produces no view on whether liquidity is adequate, whether credit growth is too fast, or whether anything ought to be done, and a reading that ends with one of those has stopped somewhere other than where this sequence stops. The refusal is not squeamishness. None of the seven outputs is an input to such a view, so a view arriving at the end of them came from somewhere else and is wearing their authority without having earned it.

What does the whole sequence produce on one Sankhya release?

The sequence runs once, end to end, on the Republic of Sankhya, and nothing in it explains why anything moved. Each entry in the right-hand column is short.

Seven steps produced eight figures, one word, three conditions and no verdict at all. A completed reading looks like that: eight figures, one word, three conditions and no verdict. The absence of a conclusion at the bottom is a feature of the sequence rather than a sign that the reading stopped early.

Can the outputs be held in order without skipping one?

The panel below walks the same sequence one step at a time and refuses to run ahead of itself. A release is picked, the steps are taken in order, and the button that jumps to step four early will not go. The refusal is the point. In real reading nothing enforces the order: the release carries every number at once, and the order is held only by the reader.

What changes when the release is read for one company rather than for the economy?

Somebody reading a liquidity release inside a lending business, or a treasury desk, or an equity team covering lenders, is reading for a narrower purpose than somebody reading it for the economy. The useful thing to know is how little of the sequence changes. The position of the system, the overnight rate, its place in the corridor, the attribution and the width used are properties of the system and not of anybody's book, so steps one through five run identically and produce identical outputs.

Step six is the one that gains something, and what it gains is a question rather than an answer. The credit figure in a release is an aggregate, and an aggregate covers a mixture of borrowers. So the practitioner's step six output carries one extra line: which kind of borrower does this credit figure cover, and is it the kind whose behaviour matters to the thing I am looking at? The borrower question is asked at step six and answered under the composition of credit, a separate subject from the reading order. A household budgeting for a loan and a treasury desk funding overnight are both downstream of the same release, and neither of them gets a different reading order out of it.

The second thing a practitioner does differently is keep the seventh step's conditions somewhere they will be seen again. A condition written down and filed is a check; a condition written down and lost is a sentence. The sequence is only as good as the fact that somebody looks at step seven's output again when the next release arrives.

The Republic of Sankhya, the household electricity bill, the vegetable seller and the shop on the street are invented.
Educational material. Not advice on any investment, tax, budget or market position.