How to Read GDP Data: The Order That Makes It Readable

Read an output release in a fixed order: the vintage first, then the base period, then the level, then the growth rate, then the composition. The headline rate comes fourth because it is a ratio of two figures, and a reader who takes it first has taken a conclusion before either input. Each step writes down one output, and the reading stops when all of them exist.

An output release is the sheet on which a statistics office puts out its estimate of gross domestic product (GDP)The money value of everything an economy produces in a period. How that value is built up is covered separately; the release only reports it., and almost nobody reads that sheet in the order it can be read in. The eye goes to the one figure the headline carries: the growth rate. The growth rate is also the one figure on the sheet that cannot stand alone. A growth rate is a division. Two figures went into it, both of them are printed on the same sheet, and both of them get skipped by a reader who starts at the answer. Everything below is one order for taking those figures instead, with the growth rate arriving fourth rather than first.

In what order is an output release read?

The sequence runs to seven steps. Every one of them is a physical action that finishes with something on paper, and none of them calls for interpretation. Step one records the vintage. Step two names the base period. Step three takes the level. Step four takes the growth rate. Step five reads the composition. Step six checks the reading for mixed vintages. Step seven writes down the conditions that would change it. The order is the method: each step uses only what the step before it produced, so the sheet fills from the inputs outward rather than from the conclusion backward. Nothing in the list below calls for a judgement about the numbers. Each step asks for one thing to be written down and handed forward.

A household bill is already read this way. The reading does not start with the sentence at the bottom saying the amount is up by a quarter. The reading starts with which bill it is, then what the last one came to, then what this one comes to, and the increase is worked out afterwards from the two. Read in that order, the increase is a fact that has been built. Read from the bottom, it is a claim somebody has handed over.

What is written down at step one?

Step one writes down the vintage. The vintage is the label saying which version of this figure is in hand: a first print, a numbered revisionA later, restated version of a figure the same office has already put out. Why figures get restated at all is covered separately; step one only records which version is in hand., or a later vintage of the same release. The vintage label then travels beside every figure the rest of the reading produces. The label is written first because it is the only output that cannot be recovered afterwards: a level, a rate and a share can all be looked up again, but a figure copied without its label becomes a number nobody can place. In the invented Republic of Sankhya, one output growth figure exists at three vintages. The first print puts it at 6.20 per cent. The first revision puts it at 6.50. A later vintage puts it at 6.80, and all three of those are in per cent. All three are correct and they disagree, so the label rather than the value is the first thing on the sheet.

The action is small and physical. Write the vintage at the top of the sheet, then write it again beside each figure as it arrives. If the release calls its figure provisionalA label an office attaches to a figure it expects to restate later. The label describes the status of the estimate, not its quality. or gives it any other estimate label, that word is copied too. If the sheet cannot say which vintage a figure came from, the figure is not finished being read, however precise it looks.

What is established at step two, before any rate is believed?

The base period is the period the growth rate is measured against. Step two names it and adds one word about it: ordinary, or unusual. Those two entries are the entire output. Step two asks nothing about what an unusual base does to a rate, and it calls for no adjustment. Why a strange base pulls a growth rate around is a mechanism covered separately, and this step takes nothing from it. Step two exists so that the base is written down as a named period with a flag beside it, before the rate that depends on it appears anywhere on the sheet.

The one word is decided by how the series came into that base. In the invented Sankhya index, a base of 90.00 index points that the series arrived at after standing at 100.00 fell 10.00 per cent into that base, so it gets the word unusual. A base of 130.00 index points reached from 125.00 rose 4.00 per cent into that base, so it gets the word ordinary. A base of 108.00 index points reached from 90.00 jumped 20.00 per cent into that base, so it also gets unusual. The flag is descriptive and it is not a verdict on the economy or on the release.

Why does step three take a level rather than a rate?

Step three takes the level: one figure, with its unit and its vintage beside it, and nothing else. In the invented Sankhya release the level at the first print is 138.06 index pointsThe unit an index is counted in. An index carries no rupees and no tonnes: it is a level set against its own starting point, so a move in it is described in points.. One level with its unit and its vintage is the whole output of step three. No rate, no comparison, no share. Step three is the step readers skip. A rate without a level describes a movement in something whose size was never established, so skipping step three is what leaves step four uninterpretable.

A level looks unhelpful on its own, so notice how much discipline the step needs. An index level of 138.06 says nothing quotable, and is not meant to. The level is one of the two figures the rate at step four is made from, and its job is to be on the sheet when that rate arrives. The other one, the base, is already there from step two. So by the end of step three both inputs of the headline are written down, in units, with vintages, before the headline itself appears anywhere.

Why does the headline growth rate come fourth?

Step four takes the growth rate and writes the base beside it, in the same line, always. In the invented Sankhya release the first print gives a rise of 6.20 per cent on a base of 130.00 index points, and the way that line is written is the point: the rate never appears on the sheet alone. The rate comes fourth because it is a ratio, and both of its inputs were taken at steps two and three. A rate taken first is a conclusion accepted before either of the figures it was built from has been looked at. The order puts the rate after those two figures rather than before them.

The everyday version runs like this. A street vendor says takings are up by half. Up by half from what? If last week was the week of a strike and the stall opened for two days, up by half is arithmetic about a bad week. If last week was ordinary, up by half is something else entirely. The sentence is identical in both cases, and the figure underneath it, the denominatorThe lower half of a division. In a growth rate the base period sits there, so it is the figure everything is being measured against., is what separates them. Step four is where the rate does not get written down without that figure beside it. The filed sheet drawn below comes from an earlier stretch of the same invented Sankhya index, where the base stood at 90.00 index points, and it shows what a rate looks like when it is the only line anybody took.

What does step five add that a growth rate cannot?

Step five reads the composition and writes down the largest movers, each with its share of the change. In the invented Sankhya release at the first print, the index moved 8.06 index points, and that change splits into services at 5.20 points, industry at 1.94 points and agriculture at 0.92 points. As shares of the change those are 64.52 per cent, 24.07 per cent and 11.41 per cent, and the three points figures add back to 8.06 exactly. A growth rate can only say that something moved; the composition is the only output on the sheet that says what moved.

Think of a shop whose takings rose. The takings figure on its own is compatible with a hundred different stories: one wholesale order, a festival week, or every counter up a little. Until the takings are split by counter, the owner has a size and no subject. The composition step is that split, and its output format matters: each mover named, each with a share, and the parts adding back to the whole so anybody can check them.

How does step six catch a comparison that mixes vintages?

Step six takes every comparison the reading is about to make and marks it, pair by pair, either like for like or refused. The test is the label from step one: two figures are comparable when they carry the same vintage, and a pair that mixes a first print with a later vintage gets refused rather than adjusted. The invented Sankhya release exists at three vintages, so nine pairs can be formed from them and only three of the nine survive: first print against first print, first revision against first revision, later vintage against later vintage. Six of the nine pairs are refused, and the refusal is itself the output of the step rather than a failure of it.

Refusing feels like doing nothing, so the step has to produce a written mark. If the pair is refused, write refused beside it, and the reading is now on record as having declined a comparison rather than as having quietly made a bad one. Step six is also where a reader catches themselves: the figure in their head from an earlier reading almost never carries a label, and the unlabelled memory gets left out here.

What is written down at step seven?

Step seven writes down two or three named conditions: the specific things that would change this reading, put on the sheet before the next release exists. On the invented Sankhya reading those conditions are a numbered revision to the level of 138.06 index points, a restatement of the base level of 130.00 index points, and a change to the contribution shares that this reading has taken as fixed. A reading that never said what would change it can absorb anything without ever being wrong, so naming the conditions in advance is what makes the reading answerable later.

The everyday version is a wedding budget written with two lines at the bottom: this holds unless the guest count moves past four hundred, or unless the venue changes its rate. Written in advance, those two lines make the budget checkable. Written afterwards they are excuses. Step seven is the same discipline, and it is a step rather than an afterthought precisely because it has to be done before the next figure arrives.

What does the whole reading look like at two vintages?

Here is the reading run end to end on the invented Sankhya release, and then run again on the same release at a later vintage. The second run shows which outputs move and which hold. The base period is the same period in both runs and its level of 130.00 index points is settled, so the whole of step two holds. The level of the period being read is restated from 138.06 to 138.84 index points, so step three moves, and step four moves with it from a rise of 6.20 per cent to a rise of 6.80 per cent. Only step two holds across the vintage, the base period with its level and its one word, and every other step moves. For exactly that reason the vintage is written down first, and then carried beside every figure the reading produces.

Reading the table down the middle column gives one complete reading. Reading across gives the thing the vintage discipline is for: the same release, read the same way twice, producing a rate that differs by 0.60 points between the two runs. Neither run is a mistake. The first print reading was correct at its vintage and the later reading is correct at its own, and a reader who kept only the value and threw away the label now has two irreconcilable numbers and no way to say which is which.

When does the reading stop?

Seven outputs, and the sheet is finished: the vintage, the base period with its one word, the level, the growth rate with its base beside it, the composition as shares, the vintage check pair by pair, and the conditions. Stopping is itself a step here, not a feeling that arrives. The seven steps produce no view on whether growth is strong or weak, so the reading does not stop with one. What it produces is a sheet somebody else can pick up, recompute and disagree with in a specific place.

A finished sheet feels like it is asking for a verdict, and stopping anyway is the hardest part of the method to keep. The sheet is not asking. Every real figure on such a sheet is somebody's estimate at a particular vintage, and the sheet exists to carry those written inputs, so a view formed later and elsewhere rests on top of them rather than standing in place of them.

What changes when the reading is for one sector rather than a whole economy?

The order does not change. An analyst reading a sectoral seriesA series covering one slice of the economy rather than all of it, such as a single industry group. takes the vintage, then the base period, then the level, then the rate, then the composition, then the vintage check, then the conditions, in exactly that sequence. Two things do change. Step five narrows. The composition of one sector splits into its own parts rather than into services, industry and agriculture. A narrow series is also built on thinner underlying data and tends to be restated further than a broad one, so the vintage discipline at steps one and six matters more. The gap between a first print and a later vintage tends to be wider the smaller the thing being measured, so the narrower the series, the more of the reading rests on step one.

The order earns its keep in practice like this. A lender sizing exposure to one industry group, an equity analyst building a demand assumption, a policy team preparing a note: all three are handed a headline and all three need the two figures under it. Reading in this order means the level and the base are already written down when somebody asks the obvious question: what was the rate measured against? Nobody has to go back to the release under time pressure and hope the base was ordinary.