The Economic Calendar: What Is Released, and When
An economic calendar states what is coming and in what order. The order is the useful part and it is durable: surveys arrive before volumes, volumes before aggregates, and aggregates before their own revisions. The dates are not durable. A date is a lookup, and the body that publishes a figure is the only place that keeps it current.
Two earlier ideas are borrowed here without being taught again. One is that a figure comes with a vintageWhich pass of a figure is in hand: the version published first, or a later version that has taken in more information. The same release can have several vintages, and they can disagree without any of them being an error. attached, and that the vintages of one figure do not agree with each other, so a single release carries more than one correct value. The other is the set of aggregates, the output figures and the price indices, whose meaning is settled elsewhere and simply used here. Neither of those settles the sequence the releases arrive in, or the reason that sequence has the shape it has.
What is an economic calendar actually for?
Most people answer this question wrongly, and the wrong answer sounds sensible. A calendar, they say, states when to look. The answer is true and almost worthless. Anyone who wants to know when something lands can find that out in ten seconds from the body that publishes it. Dates move, so the check has to be made again anyway.
Here is the answer that earns the reading. A calendar names what will be known next, and therefore what today's reading is still missing. Those are two halves of one fact, and the second half is the part nobody uses. Knowing that a counted figure lands after a survey figure establishes something precise about this moment: the survey is the only evidence in the round, the count has not happened yet, and any conclusion drawn today rests on one kind of measurement with nothing to check it against.
A household waiting on a diagnosis is in the same position. The doctor has looked, asked questions and formed an impression, and the blood work has been sent off and has not come back. The impression is real information and it is worth having. But a household that treats the impression as the result has quietly written the pending test off, and the whole reason a hospital lists what is pending is to stop anybody doing that by accident. An economic calendar does the same job. The calendar is the list of what is pending, and that list describes the hole in what is currently known.
So the order matters more than the dates. Knowing the hour a figure lands changes nothing about how today's evidence is weighed. Knowing that a counted figure is still to come changes it completely. The reading becomes an early view and not a settled one. The first is a lookup. The second is a habit of thought, and it is the only part of a calendar that can be taught.
Strip a calendar down to the single thing it is most useful for. What is that thing?
What is the order of releases, and why that order?
The order has four positions and they are always in the same relation to each other. Surveys come first. Counted volumes come next. Aggregates come after the volumes. Revisions come last. The part that makes the whole thing memorable is that none of it has to be memorised at all.
The order follows from how each figure is built, so it is not a convention and it does not need learning as a list. Work down the four and watch what each one needs before it can exist. A survey is built by asking: put a short question to a panel of firms and the answers come back about as quickly as they can be given. Nothing has to be assembled first. A volume figure is built by counting: somebody has to collect what producers actually made, chase the ones who have not sent anything in, and check the result. Collecting always takes longer than asking. An aggregate is built by assembling: it is put together out of volume figures and other components, so it is downstream of them by construction and cannot possibly finish first. A revision is built by waiting: statutory returnsForms and filings that businesses, offices and tax authorities are required to send in. Statistical figures are often built out of them, and they do not all arrive together, so some of the information exists before the compiler has it. that had not reached the compiler by the first print reach it afterwards, and the figure is recomputed to take them in.
Read those four descriptions and the order is simply their build time, sorted. Asking is faster than counting. Counting has to finish before assembling can start. Assembling has to finish before anything can be revised. There is no step where a compiler is making a choice about presentation; each step is waiting on the one above it to hand over something it needs.
The contrast between asking and counting is doing most of the work, so the two ends of the order are worth looking at side by side. A survey and a volume figure are both trying to describe the same thing, and they get there by methods with completely different amounts of gathering in them.
A survey figure on activity is published before a counted volume figure on the same activity. Why is that the order rather than the reverse?
The third step is the one readers underestimate, so it is worth drawing on its own. An aggregate is not a measurement. An aggregate is an arithmetic result computed out of other measurements. Every one of its inputs is a release with its own position in the order.
Revisions sit at the end of the order. What puts them there?
The India Macro Dashboard: Releases, Dates and Official Sources, and where to go for them
Everything above is method, and method has to land somewhere real. In India the figures a reader wants are published by named bodies, in named releases, and the whole discipline is going to those rather than to somebody's account of them. Here is what each of the relevant things is, said plainly, with no reading and no schedule attached to any of them.
The National Statistical Office is the body inside the Ministry of Statistics and Programme Implementation that compiles India's official statistics. The office is a compiling authority rather than a commentator: it collects, checks, computes and publishes, and the figure it publishes is the figure. India's national accountsThe set of statements that add a whole economy's output, income and spending up into a small number of totals. The parts can then be compared with each other and with the whole. releases are its assembled aggregates, computed out of components rather than measured directly. The Index of Industrial Production is a volume measure: it tracks the quantity of industrial output rather than its value, and counted returns from producers have to arrive before it can move. The Purchasing Managers' Index is a survey measure: it is compiled from a questionnaire put to purchasing managers at firms, and it reports the breadth of change across those firms rather than the size of the change in output. Each of those is a different kind of thing, and the kind is what settles where it sits in the order.
A summary is a vintage of somebody's reading rather than a vintage of the figure, so going to the source rather than to a summary is the whole of the discipline. That sentence is worth taking apart. When a figure is revised, the release carries the revision, states which vintage is on view, and usually shows the earlier ones as well. A summary written about the first print does not update itself. The summary sits there, still confident, still readable, describing a number that has since moved, and it never says which pass of the figure it was written about. So where two accounts of the same release disagree, the first thing to check is not which one is right. The question is which vintage each account was reading.
Where the dates for these releases live, and why a copy of them goes wrong
The bodies and the releases named above are real, and each of them settles its own publication dates and states them itself. A date, a reading, a base yearThe year a statistical index is measured against, usually set at a round number such as 100. Later readings are compared with it and with each other. and a method detail all move, and a copy of any of them is wrong the moment the original moves, without ever looking wrong. The site in the table below is where each of them stays current.
Two accounts of the same release disagree with each other. Where does a careful reader go first?
How to Read an Economic Calendar: what to ask of an entry
Consulting a calendar is looking at it. Reading one is asking four questions of every entry, and the four are short enough to become automatic. Take any single upcoming line and put these to it.
First: which aggregate does the entry report? Two different releases can report on the same underlying thing by different methods, so the thing being measured matters more than the name of the release. A reader who tracks release names rather than measured things will treat those two as unrelated. Second: which vintage will this one be, a first print, a revision, or a later pass? The answer settles how much room the figure still has to move. Third, what will it be compared against? A figure on its own is not news; a figure against something named is. Whatever the release itself compares it with, the release states, and that is worth reading rather than assuming.
Fourth, and this is the one that earns the whole exercise: has anything the entry depends on already been published? If the entry is an aggregate assembled out of two volume figures and both of those are already out, then a large part of the entry is already visible, and only the remainder of it is genuinely new. The fourth question catches the commonest error there is in reading data: reacting to a figure whose inputs were already public. A reader who skips it experiences an aggregate as fresh information when most of it was already available in parts, and the sensation of surprise is real even when the information is not.
Of the four questions, one catches the error readers make most often. Which one is it?
Where do the Sankhya vintages sit on this order?
Work the whole thing through on one case. The Republic of Sankhya, an invented state carrying invented figures, has a growth figure with three published vintages: a first print of 6.20 per cent, a first revision of 6.50 per cent, and a later vintage of 6.80 per cent. The three vintages are not three numbers. Each is a position on the order set out above, and a calendar and a set of vintages therefore describe the same thing from two directions.
| Step | What is released | Why it sits here | What the growth figure reads |
|---|---|---|---|
| 1 | The survey figure | Built by asking, so it needs least gathering | No growth figure exists yet |
| 2 | The volume figure | Built by counting, which is slower than asking | No growth figure exists yet |
| 3 | The aggregate, first print | Assembled out of step two, so it waits for it | 6.20 per cent |
| 4 | The first revision | Waits for returns that missed the first print | 6.50 per cent |
| 5 | The later vintage | Waits for the slowest returns and a fuller method | 6.80 per cent |
Read the last column downwards and three separate facts fall out of it. The first is that no growth figure has been published at all for the first two steps of the order, so anybody reading Sankhya at step one or step two has only a survey to read. The second is that the first print of 6.20 per cent is the least settled version of itself that will ever be published. The print sits at the first step where the figure appears, with two later steps still to come. The third is arithmetic: 6.80 per cent less 6.20 per cent is 0.60 percentage pointsThe unit a distance between two percentage figures is stated in. Growth going from 6.20 to 6.80 has moved 0.60 points, and calling that 0.60 per cent would name a far smaller change., made up of 0.30 points at step four and 0.30 points at step five, and every one of those components is set out above so the total can be added back by hand.
The four reading questions now go to a single upcoming entry, and the fourth one does its work. Step three, the aggregate first print, is taken as a line still to land.
| The question | The answer for this entry |
|---|---|
| Which aggregate does it report? | The Sankhya growth figure, assembled rather than measured directly |
| Which vintage will it be? | A first print, so it has two later steps still ahead of it |
| What will it be compared against? | Whatever the release itself names, which for Sankhya is a consensusA figure gathered before a release from the people who forecast it, used as the benchmark the published print is then measured against. A consensus is a collected number, not an official one. of 6.40 per cent |
| Has anything it depends on already been published? | Yes, two of its inputs: the survey figure at step one and the volume figure at step two |
The last row is the whole point of asking. Two of the entry's three inputs are already public, so a reader who has read them already holds most of the entry. The 6.20 per cent print against the consensus of 6.40 per cent is a shortfall of 0.20 points, and it is worth noticing how small that is beside the 0.60 points the figure went on to move across steps four and five. The figure travelled three times as far as the gap anybody reacted to, and it did so without the consensus of 6.40 per cent ever being changed.
Stand at a point in the order, and read what is not yet known.
The slider sets a position in the Sankhya release order. Two things redraw underneath it. The ladder shows all five releases with the ones already out filled in and the ones still to come left pale, and a bright marker moves to the chosen position. The dark band at the foot then names what is not yet known at that position, the calendar's actual answer and the output nobody reads. The controls open where the worked example above stands: after the survey figure and before the volume figure, with no growth figure published at all.
A written account of a calendar is better off carrying no dates at all. Why?
Why is a date list the wrong thing to write down?
An account of a calendar that leaves out every date looks at first like an omission, and it is the opposite of one.
The order is the teaching and the dates are a lookup, and the two belong in different places because they age at completely different rates. The order does not move. The order is a consequence of how the figures are built, and a figure built by asking will always be quicker to produce than a figure built by counting and checking. Nothing about that changes when a body reorganises its publication practice. The dates move for all sorts of ordinary reasons, and every one of them is invisible to anything written earlier. A printed date list would now be wrong, and it would look exactly as confident as it did on the day it was written. The failure is a quiet one rather than an obvious one.
There is a second reason, and it is about maintenance rather than truth. A date list is a record somebody has to keep. Records that need keeping decay when the keeping stops, and they decay without any signal. The sensible response is not to keep them better; it is to leave the date where the publishing body maintains it, and to carry only what does not need maintaining. So the order is set out here and the dates sit at the source, in the table below. The principle is the same as reading the release rather than a summary of it, applied one level up.
What does a calendar not tell?
A calendar is confidently silent about three separate things, and the silence is the reason it can be trusted about the fourth. A calendar does not say what any release will show, and it does not say whether the release will turn out to matter. Nor does it say what anything else does after the release lands. A calendar is a list of questions that will be answered, not a preview of the answers.
The confusion between those two is easy to fall into because a calendar entry reads like a headline waiting to happen, and the mind fills the gap. But notice what would have to be true for a calendar to contain any of the three things above. Somebody would have to know a figure before it was compiled, and that is a contradiction. Or somebody would have to know what other people will make of it, and nobody does. A calendar avoids all of that by containing only facts about the entry rather than facts about the content: which thing is reported, then which vintage it will be, and then what it will be set against. Every one of those is knowable in advance precisely because none of them is the figure.
Does a calendar say whether an upcoming release will turn out to matter?
How does an analyst use a calendar in practice?
A working analyst uses a calendar backwards from the way most people use it. The interesting column is not what is coming, but what has not come yet.
Before writing anything about a period, the analyst lists which releases covering that period are already out and which are still to come. The list decides how the note is written. If only a survey is out, the note says so and treats the reading as one method with nothing to check it against. If the counted volume figures are out and the assembled aggregate is not, the note can say a good deal about the parts and must stay quiet about the total. If the first print of the aggregate is out but no later vintage is, the note carries the print and states that it is a first print. A reader who is not told that will treat a figure with two more steps ahead of it as settled.
A calendar answers precisely one question precisely well: what is not yet known. The answer decides how much weight today's evidence can carry. Everything else an analyst needs comes from the releases themselves. The discipline transfers straight to a household for the same reason. Anyone deciding something on the strength of one figure benefits from asking the same question a professional asks: what else about this is due, and would it change what I think? A household deciding on a large purchase because prices seemed to be easing is doing the same arithmetic as an analyst reading a first print, and the same list of what is pending protects both of them.
One more habit worth borrowing. When an analyst reads a survey figure, the note names it as a survey rather than as activity. The two are different objects. And when a figure has been seasonally adjustedA method that takes the regular, repeating swings out of a series. Whatever movement is left can then be compared from one reading to the next without the predictable pattern getting in the way., the note says which version it is reading. Neither habit is difficult. Both come from the same place as the four questions above: describe the figure precisely, and most of the ways to misread it close by themselves.
Where does a calendar get read wrongly?
A survey reading gets weighted as though it were a volume figure
Here is the mistake, and it is the most common one on this whole subject. A survey figure is published at step one. The survey figure reads clearly, it is easy to write about, and no counted figure has appeared to contradict it. No counted figure has appeared at all. A reader concludes from it that output rose, and the confidence comes from a mistake about the order rather than from a mistake about the number.
Work it through on Sankhya. A survey covering 100 firms finds 60 of them reporting an improvement, and the diffusion readingA measure built by counting how many of the units surveyed report an improvement, without any regard to how large each of them is. A diffusion reading reports breadth of change rather than size of change. is 60.0 against a boundary of 50.0. The reading looks like a clean answer. But those 60 improving firms hold only 30.00 per cent of Sankhya's output, and weighted by output the change across all 100 firms was minus 1.50 per cent. So more firms improved than worsened and output still fell, and neither the count nor the weighting is wrong: they measure different things. The figure that would have shown the fall is the counted volume figure at step two, and at the moment the reader drew the conclusion, that figure had not been published at all.
The fix is a single question asked in the right order: what has not been released is settled first, and only then is weight put on what has. Arriving first is a property of how a figure is built, not evidence that it is right. A survey arrives first because asking is quick, and quickness carries no information whatever about whether the answer will hold up against a count.
A reader takes a diffusion reading of 60.0 as evidence that Sankhya's output rose. Which release had not been published when that conclusion was drawn?
Where the releases and the dates actually live
| Source | What to look up there | Site | Date checked |
|---|---|---|---|
| National Statistical Office, Ministry of Statistics and Programme Implementation | The advance release calendar and the press note listings for India's statistical releases | mospi.gov.in | 19 August 2026 |
| Reserve Bank of India | The Database on Indian Economy, and the listing of its own publications | rbi.org.in | 19 August 2026 |
| Ministry of Statistics and Programme Implementation | The Index of Industrial Production, with its methodology and base year notes | mospi.gov.in | 19 August 2026 |
| Press Information Bureau, Government of India | The archive of press releases that carry statistical figures | pib.gov.in | 19 August 2026 |
| International Monetary Fund | The Special Data Dissemination Standard, and its guidance on advance release calendars | imf.org | 19 August 2026 |
The Republic of Sankhya and every figure attached to it are invented.
Educational material. Not advice on any investment, tax, budget or market position.