How to Read an RBI Monetary Policy Statement

A monetary policy statement is the account the Reserve Bank of India (RBI) gives of a rate decision and the thinking behind it. The decision itself is one line, and a document built around one line still rewards a reading order. The vote, the reasoning and the projections around it carry far more, and a reader who stops at the rate has stopped at the thinnest part.

A reading order and monetary policy itself are two different subjects. How a policy rateThe single interest rate a central bank sets and announces, from which other short term rates in the system take their cue. How it gets set is a separate subject. works, why an economy responds to one, and what any published series measures are each set out under their own headings. The rate, the toolkit, guidance, market operations and the journey a change makes are all taken as already familiar. A document that touches all five only becomes worth opening once they are.

What is a monetary policy statement, and who issues it?

A monetary policy statement is issued by the Reserve Bank of India, and the reasoning inside it is the reasoning of its Monetary Policy Committee, the body that takes the decision. Two things about the document hold still, and everything else about it moves. When a statement lands, how often statements land, how many members sit on the committee, which way any of them has ever voted and what any statement has ever said are all capable of changing. Anything written down about them ages quietly while continuing to look perfectly correct, so each one is read off the statement in hand.

Place the document by its job rather than by its calendar. The statement is the account a decision maker gives of a decision it has just taken: a short document that reports the outcome, sets out the case behind it, publishes what the committee expects, and points at the statistics it leaned on. The document argues a case and reports a result in the same voice and inside the same paragraph, and every step of the reading order below follows from that one property. A reader who does not separate the two comes away holding a case as though it were a count.

Why is the rate decision the least informative line in it?

Everything below follows from a single claim. Start with it. The rate decision is the shortest, loudest and least informative line in the entire document. It is one sentence. Every headline has already carried it, well before the file was opened. The line says what was decided, and nothing whatever about how close the decision was, what the committee thought it was responding to, what it expects next, or what it stopped saying since the last time it wrote. All four of those are in the same document, further down, and all four carry more than the line at the top.

Here is the everyday shape of it. A housing society committee votes on whether to raise the monthly maintenance charge, and two things come out of the meeting: a one line resolution, and the minutes. The resolution says the charge is unchanged. The minutes say the vote was four to three, that two members who had been comfortable last quarter were not comfortable this quarter, and that the lift contract is up for renewal. Every resident reads the resolution and a handful read the minutes. The handful are the only ones not surprised at the next meeting. A monetary policy statement bundles both into one document, which is convenient and is also exactly why a reading order is needed: the resolution is at the front and the minutes-like material is behind it.

In what order should a policy statement be read?

There are eight steps, and each leaves behind material the next one picks up. They run in the order printed. Step one finds the decision and puts it down. Step two counts the vote. Step three marks up the reasoning. Step four lists the projections with what each depends on. Step five puts one question to the guidance. Step six holds this statement's reasoning against the last one's. Step seven takes the numbers actually needed back to the series they came out of. Step eight records what the document left alone and stamps the notes with the meeting they belong to. Five of the eight steps consume material an earlier step produced and cannot make for themselves. The order is not a preference.

Two things about the list are easy to miss on the way past. It repays one read through before it is run. The first is that not one step calls for a conclusion about any economy. Every step finds, counts, marks, lists, asks, compares, follows or writes down, and the procedure ends without a verdict anywhere in it. The second is that steps six and eight are the two people skip, and they are the two that make the notes worth anything a month later.

Why does step one read the decision and then put it down?

Step one takes about ten seconds. The sentence that says what was decided is found, read once, written at the top of the notes, and then physically passed over. A line that loud will otherwise sit at the front of the mind and colour everything read after it, so the step exists to get the decision out of the way. That is the whole of the step. Step one produces one line of notes and one cleared head, and the cleared head is the reason it is a step at all rather than something that happens by accident.

There is a reason this needs saying out loud. The decision travels faster than the file does. Almost every reader arrives at the document already knowing it. Opening the statement to confirm something already known is a strange way to start reading, and it leaves the rest of the document being scanned for agreement rather than for information. Writing the line down and closing the question is a small trick with a large effect: it converts the decision from the thing being looked for into a fact already filed. Then step two starts, and step two is where the document begins to earn the time being given to it.

What is the vote doing as a step of its own?

Step two counts. How many members took part, which way each of them went, and whether anybody recorded a dissent. The count takes a minute and goes on the notes as a line of arithmetic and nothing more. The vote is a step of its own because two statements carrying the identical decision can carry completely different votes, and the vote is what shows how settled the reasoning behind that decision is among the people who wrote it. A committee that agreed without exception and a committee that split down the middle have published the same outcome from very different rooms.

Feel it at household scale first. Six people in a joint household decide not to buy a second scooter this year. If all six agreed in a minute, the question is closed and nobody will raise it in April. If it went four to two, with the two saying plainly that the commute has changed, then the question is not closed at all: it is parked, and it comes back the moment anything shifts. The decision written on the fridge is identical in both cases. The two decisions differ in how much of the household's thinking each one represents.

Two disciplines keep this step honest. The count and the direction go down as printed, without any added reading of what a member meant. The notes treat a dissentA member of a committee formally recording that they did not agree with the decision the majority took, so that the disagreement becomes part of the published record. as a recorded fact rather than as a signal. Turning a count into an expectation about the next meeting is forecasting, and forecasting is not a reading procedure. The step records the shape of the vote and stops there. The count puts a number on how much agreement sat behind a sentence that would otherwise be treated as a single voice.

Why mark each claim as record or expectation before reading on?

Step three is the reasoning, read with a pen in hand. Every claim gets one of two marks: R for a record of something that has already happened, or E for an expectation about something that has not. Two marks, nothing finer, and no claim is left unmarked. After a few paragraphs the two kinds of sentence have fused into one impression of what the committee thinks, and an impression cannot be pulled back apart into its two halves. The marking is done while reading, before going any further.

Why is this necessary at all? The difficulty is grammatical rather than economic. Both kinds of claim arrive in the same register, share a paragraph and very often share a single full stop, and the signals that separate them are tiny: a past tense set against a verb of expectation, a quantity somebody measured set against a quantity somebody expects. A school circular does the identical thing when it announces that two rooms were added this year and that the fee will therefore need revising next. One half of that could be checked against something on file and the other half is an argument, yet a parent who reads it once carries both away as reported fact. Nobody has deceived anybody. Unbroken prose simply flattens two kinds of claim into one. A mark made on the document while the sentence is still in view keeps the two apart.

What has to be attached to a projection before it can be used?

Step four lists the projections and writes beside each one what it is conditional on. Not what it means, not whether it looks high or low, and not what the reader would have projected. Only the conditions the document itself attaches to it. A projection with its conditions detached is not a weak number, it is an unusable one. No later moment ever arrives at which anybody can say whether it held. Six months on, a conditional projection can be checked against what actually happened to its conditions. Nothing at all can be done with a bare figure that arrived without them.

An everyday version makes the asymmetry obvious. A caterer quoting for a wedding says the cost will be a certain amount, assuming three hundred guests, assuming vegetables at the current market rate, and assuming the venue supplies the tables. Every one of those assumptions is checkable on the day. Strip them off and keep only the amount. When the final bill is higher, nobody can say which assumption broke, and there is no honest conversation left to have. Step four copies the assumptions across, and it is the cheapest step in the whole procedure to do and the most expensive one to skip.

What single question does step five put to the guidance?

Step five reads whatever the document says about what comes next and asks exactly one question of it: which turn of events would let this be set aside without anybody having gone back on their word. Write the answer in a line. Only that question separates a description of intent from a commitment, and readers reliably hear the second when the first was written. Why forward guidanceA central bank saying something about the path it expects to take, so that people can plan against it. Why it works as a tool at all is covered separately. is used as a tool at all, and what it does to expectations, is covered separately and no part of it belongs inside this step.

The step stays small. Step five does not judge the guidance, does not score it, and certainly does not decide whether the committee will stand by it. The step writes down the exit that the wording itself describes, in the words the wording uses. If no exit is described, that is a perfectly good answer and it goes on the notes as one. The step is over in a line. Whether the exit was used will be visible in the next statement, and that is what the line buys.

Why is the change against the last statement the news?

Step six is the step that separates a reader from a headline. Put this statement's reasoning beside the previous statement's, thread by thread, and mark each thread as carried in the same terms, carried in different terms, dropped, or new. Everything that stayed is background that was already available before this document existed, so the threads that moved are the news. A statement is not written from a blank sheet. A statement is written against the last one, and it is read the same way by everybody who reads these documents for a living.

Step six needs two documents rather than one, and finding the second is more effort than a reader with one file open wants to spend. The step therefore comes late, and it is the one most often skipped. The comparison is still the highest yield step in the sequence. Four threads carried in identical terms and one quietly dropped is a different document from five threads all carried, even though every sentence in both is unremarkable on its own. When no previous statement is to hand, the notes say so rather than pretending the step ran. A reading with a known gap in it is a completely different thing from a reading with an unknown one.

Where do steps seven and eight leave the reading?

Step seven takes any number actually intended for use and walks it back out of the document into the published series it came from, copying the figure, the compiler, the release and the date the release was opened. A statement points at statistics; it is not where a statistic is dated, defined or revised, and those three properties are the ones needed when somebody questions the number a month from now. This is done only for the numbers the question turns on. Chasing everything is a way of not finishing.

Step eight closes the reading with two small acts of housekeeping. A silence disappears the instant the file closes, and memory keeps only the sentences that were actually there. So while the file is still open, a line or two goes down on the part of the question the document never went near. Then the notes are stamped with the meeting they belong to, exactly as the document identifies it. An unstamped note ages badly and invisibly: a year later it is still being quoted, nobody can tell which statement produced it, and the only honest thing left to do with it is throw it away and read again.

How do the two cautions change what the document is?

Caution one, and it changes how every step above is weighted. The statement is written by the same body that took the decision, so it is an explanation rather than an assessment. Common authorship is not a criticism, it does not make the document unreliable, and it is not a reason to discount anything in it. It is a plain fact about authorship with one consequence: an independent evaluation of the decision is a different document, written by somebody who did not take it, and this one does not contain it and was never meant to. Expecting it here is like expecting a company's own account of its year to also contain the audit. Both documents can be entirely honest and they still are not the same document.

Caution two is really two habits in one. A projection is not a promise, and a vote is not a forecast of the next vote. Both get read as commitments and neither is one. A projection is what the committee expects on stated conditions. Step four insists on copying those conditions across. A vote is a record of a room on a day. Step two writes the count and stops there. Reading either as a commitment invents an obligation nobody entered into, and the cost of that invention lands on the reader when events go somewhere else.

Notice that both cautions push the same way, toward taking the document for what it is. The statement is a decision maker explaining itself, with a count attached, some conditional expectations, and a set of pointers to statistics. Together those are a great deal of useful material. None of them is a verdict, and no amount of careful reading will turn them into one. The verdict was never in the file.

What does the whole reading look like run end to end?

Here is one reading, run from opening the file to closing the notes, with a question in hand. In the Republic of Sankhya, an invented country, a household carries a home loan of Rs 30,00,000/- over twenty years, and the reader wants to understand where a change of a few basis pointsA basis point is one hundredth of one per cent, used because saying a rate moved by a fraction of a per cent invites confusion about which fraction. in a policy rate would show up in a document of this shape. A real document cannot be asked about an invented republic. So the question carried into the statement is a structural one, and a document of this kind will actually answer a structural question: where inside it would a question shaped like this be addressed, what does each of those places rest on, and which published series sit underneath them.

Now the steps run. Step one, decision located, written as one line, put down. Step two, the count taken exactly as printed, direction and any recorded dissent transcribed and nothing added on top. Step three, the reasoning worked over with a pen, an R or an E against every claim, and how the two marks divide is a property of the file in hand rather than anything a reading order could settle in advance. Step four, each projection listed with the conditions the document itself hung on it. Step five, a single line on what would let the guidance be set down honourably. Step six, the earlier statement open beside this one and each strand marked carried, changed, dropped or new. Step seven, the one figure the question turns on followed out to the compiler's own release, where the figure, the series, the compiler and the date of opening go into the notes and the chase ends. Step eight, two lines on what went untouched, and the meeting stamped across the top.

The shape of what came out is the lesson. Nowhere in those notes does a sentence say how any economy is faring. The notes hold a marked up file, a vote set down as arithmetic, conditional expectations with their conditions still attached, a grid of what shifted since last time, one figure carrying its provenance, a recorded silence and a stamp. On the Sankhya question the biggest single entry turns out to be the silence: a statement of this kind speaks to no individual household's loan at any point in it, and the route from a policy rate to an instalment belongs to a subject settled elsewhere. That is a complete and correct reading, and there is no finding anywhere inside it. A conclusion is later work, done on evidence this reading has put its hands on rather than guessed at. Anyone waiting for one here should stop waiting.

What does an analyst actually pull out of a statement?

What does somebody who reads these for a living actually carry out of the file? Not what a newcomer expects. An analyst goes straight to the vote and to what shifted since the last statement, and treats the rate decision as old news that arrived before the document did. Those two are where a shift shows up as a matter of record rather than as interpretation, and both of them are countable, which is why they can be written into a note that somebody else can check. Everything else the analyst takes is provenance: which series the document leaned on. The numbers can then be pulled from the compiler rather than quoted from the summary.

A lender's credit team needs the pointer to the series it underwrites against, not the reasoning. The team runs a shorter version of the same reading and stops earlier. An investor reading these documents takes the marked up reasoning and the conditional projections, mostly to see which conditions are the ones worth watching. A household reader can run all eight steps at a smaller scale and get real value from three of them: the decision noted and set aside, the projections with their conditions, and the silence written down. Not one of these readers walks away with a quotation. Every one of them walks away knowing where the evidence lives, and none of them knew that beforehand. How any of this reaches a deposit rate or a loan instalment is a separate subject, covered elsewhere.

The Republic of Sankhya and the household inside it are invented.
Educational material. Not advice on any investment, tax, budget or market position.