GDP vs GVA: Why the Same Year Gives Two Growth Rates

Gross value added (GVA) measures what producers actually added, valued at what producers receive. Gross domestic product (GDP) measures the same output at the prices buyers paid, so it carries taxes on products and leaves out subsidies on them. The two usually move together, and they come apart whenever tax or subsidy settings change, even in a year when nothing at all was produced differently.

Two things have to be in place before that answer means anything. The first is that output can be counted by adding up what each producer contributed rather than what each producer sold. Adding up contributions is the only way to avoid counting the same cotton three times over. The second is that a price is not one number: what a buyer hands over and what a seller keeps are separated by whatever the state adds or takes away in between. Taken together, those two facts leave two defensible ways to size an economy, both correct, answering different questions.

The worked case here is the Republic of Sankhya, the invented country of about 20 crore people that this sequence keeps returning to. Value added is built first from a production chain, then the relationship between the two measures is stated and run in either direction, then the choice between them is set out, and then a year is pulled apart in which one measure grew 1.76 points faster than the other while the country produced exactly what it produced.

What does gross value added actually measure?

Gross value addedThe total of what every producer in an economy contributed, where one producer's contribution is its output minus whatever it had to purchase elsewhere in order to make that output. asks one question of every producer in the economy: what did it add? Not what it sold, and not what passed through its hands. Only what it added. A producer answers by taking what its output is worth and knocking off whatever it had to purchase from others in order to produce it. The remainder is that producer's own contribution, and stacking those contributions across every producer in the country gives the value added of the whole economy.

A concrete chain is easier to hold than an aggregate. A cotton grower sells a bale for Rs 200/-, having bought in nothing to grow it, so the grower added Rs 200/-. A spinning mill turns that bale into yarn it sells for Rs 500/-, and since it bought in the bale for Rs 200/-, the mill added Rs 300/-. A shirt maker buys the yarn for Rs 500/- and sells the finished shirt for Rs 900/-, so the shirt maker added Rs 400/-. The three contributions add to Rs 900/-, exactly the value of the shirt at the end of the chain. Adding contributions gives the same total as looking only at the finished shirt. The equality of those two totals is what makes value added a way of counting an economy rather than a way of counting invoices.

The careless way runs differently. Adding the three sale values instead of the three contributions gives Rs 1,600/-. The Rs 1,600/- counts the cotton three times, once when the grower sold it, again inside the yarn, and a third time inside the shirt. The Rs 1,600/- is not a bigger measurement of the same economy. The total is a wrong one, and it gets more wrong the more stages a product passes through. Counting sale values would reward an economy for having longer supply chains rather than for producing more.

There is a second half to the definition that gets skipped far more often than the first, and it is the half the distinction turns on. Value added is measured at basic pricesThe amount a producer actually receives for a unit of output, counting any subsidy received on that product and excluding any tax charged on it. It is the seller side of the price rather than the buyer side., and that phrase means the amount the producer actually receives for the thing it made. Not the amount printed on the customer bill. The amount that ends up with the producer. Gross value added is the producer side of the economy, measured in the money producers keep, and that is what makes it the natural base for any question about who made what.

The distinction is familiar from ordinary life even where the term is new. A caterer quotes Rs 500/- a plate for a wedding. A tax is added on top, so the final bill shows more than Rs 500/- a plate. The caterer never touches that extra amount. Asked how much catering the caterer did, Rs 500/- a plate is the honest number, and the bill total answers something else entirely.

What does gross domestic product value, and at whose price?

Gross domestic product measures the same physical output as gross value added. The same shirts, the same cylinders of cooking gas, the same haircuts, the same tonnes of steel. Nothing extra is counted and nothing is left out. The price the output is valued at changes, and that single change is the entire difference between the two measures. How gross domestic product is put together, and the three separate routes to arriving at it, are covered where that subject is treated in its own right. Only one of its properties matters for the comparison.

Gross domestic product values output at market pricesThe amount a buyer actually pays for a unit of output, including any tax charged on that product and after any subsidy that has reduced what the buyer hands over. It is the buyer side of the price rather than the seller side., and that is the amount buyers actually hand over. And the amount a buyer hands over is the same number as the amount a producer receives in exactly one situation: where the product carries no tax and no subsidy. Everywhere else the two numbers are pulled apart, and by a known amount.

Two products make the whole point. A shirt leaves the shirt maker at Rs 900/-, a tax on productsA tax charged on a product because it was made or sold, so it moves with the quantity sold rather than with the profit earned. The buyer pays it and the producer never keeps it. of Rs 90/- is charged on it, and the buyer pays Rs 990/-. The producer receives Rs 900/- and the buyer parted with Rs 990/-, and the Rs 90/- difference never belonged to the producer. Now a cylinder of cooking gas: the producer receives Rs 800/-, of which the state pays Rs 200/- as a subsidy on productsAn amount the state pays on a product so the buyer can pay less than the producer receives. It moves with the quantity sold, and it flows to the producer rather than to the buyer., so the buyer pays only Rs 600/-. Here the producer receives more than the buyer paid. A tax on a product makes the buyer pay above what the producer receives and a subsidy on a product makes the buyer pay below it, so the two measures can be pulled apart in either direction.

How are the two measures connected?

Both measures are now built, each on its own terms, so they can be set against each other. Gross domestic product equals gross value added, plus taxes on products, less subsidies on products. The relationship is not a finding anybody discovered by looking at data. The relationship is an identityA statement that is true by the way the terms are defined rather than by observation, so no year of data can support it and no year can contradict it., true by construction. The only difference between the two measures is which side of the tax and subsidy the price is read from. The relationship holds in every economy and in every year, so the interesting question is never whether it holds but how large the gap is and what has moved it.

The gap appears constantly in national accounts, and it has a name worth keeping. Taxes on products less subsidies on products is called net product taxes, and the shorter name used here is the wedgeNet taxes on products, meaning product taxes less product subsidies. It is the single number that separates output valued at buyer prices from value added measured at producer prices.. The wedge is what drives the two measures apart. The wedge is a single number, it can be positive or negative, and it is the only thing standing between the two totals.

Taking the shirt and the cylinder as a two product economy makes the identity concrete. Value added is Rs 900/- from the shirt maker chain plus Rs 800/- from the cylinder producer, a total of Rs 1,700/-. Taxes on products are Rs 90/-. Subsidies on products are Rs 200/-. The wedge is Rs 90/- less Rs 200/-, or minus Rs 110/-. So output at market prices is Rs 1,700/- less Rs 110/-, or Rs 1,590/-, and that checks directly against what the two buyers actually paid: Rs 990/- plus Rs 600/- is Rs 1,590/- too. In this small economy the measure at buyer prices is smaller than the measure at producer prices. The buyer side falls below the producer side whenever subsidies on products exceed taxes on them, and that kills the idea that one measure is permanently the larger of the two.

How big is the wedge in Sankhya, and does the arithmetic run both ways?

Sankhya publishes both measures, and two of its years are enough to see everything. In year 2, output at market prices was Rs 17,64,000 crore. Taxes on products came to Rs 1,84,000 crore and subsidies on products to Rs 64,000 crore, so the wedge was Rs 1,20,000 crore, and value added was Rs 17,64,000 crore less Rs 1,20,000 crore, or Rs 16,44,000 crore. In year 3, output at market prices was Rs 19,26,288 crore, taxes on products Rs 2,00,000 crore and subsidies on products Rs 40,000 crore, so the wedge was Rs 1,60,000 crore and value added was Rs 17,66,288 crore.

The relationship runs the other way too, and a relationship that travels in only one direction is a formula memorised rather than a mechanism understood. Year 3 output of Rs 19,26,288 crore less the wedge of Rs 1,60,000 crore gives value added of Rs 17,66,288 crore. Value added of Rs 17,66,288 crore plus the same Rs 1,60,000 crore returns Rs 19,26,288 crore. Every published national accounts table can be walked in both directions like this, and that ability is what allows a figure to be checked rather than accepted. If the two directions do not reconcile, one of the three numbers is wrong, and that has been established in about fifteen seconds. The check has a limit, though. The relationship is an identity, so reconciling shows only that the three figures are mutually consistent, never that any one of them is correctly measured.

Why did output grow faster than value added in a year when nothing was produced differently?

Here is the case the whole argument exists for. Sankhya from year 2 to year 3, with both growth rates computed. Output at buyer prices went from Rs 17,64,000 crore to Rs 19,26,288 crore, up by Rs 1,62,288 crore, and that works out at 9.20 per cent. Value added went from Rs 16,44,000 crore to Rs 17,66,288 crore, up by Rs 1,22,288 crore, and that works out at 7.44 per cent. Same country, same year, same output, two growth rates 1.76 points apart.

So what happened in Sankhya during year 3? Nothing, on the production side. Not one extra shirt, not one fewer cylinder. The wedge changed, and it changed in two places at once. Subsidies on products were cut from Rs 64,000 crore to Rs 40,000 crore, a withdrawal of Rs 24,000 crore. Taxes on products rose from Rs 1,84,000 crore to Rs 2,00,000 crore, an increase of Rs 16,000 crore. Together the wedge went from Rs 1,20,000 crore to Rs 1,60,000 crore, up by Rs 40,000 crore, and that is 33.33 per cent in a year when the economy under it grew 9.20 per cent. The wedge grew nearly four times as fast as the economy it sits on top of, and that single fact is the whole 1.76 point gap.

Notice what is not in that explanation. No new factories. No productivity gain. No collapse in any sector. Nobody worked differently, nobody bought differently, and no shirt maker changed what a shirt sells for at the gate. A measure moved 1.76 points on grounds entirely unconnected to what it is supposed to be measuring, and no argument for knowing what a number includes before using it is sharper than that.

What happens to the gap if the wedge is held where policy left it?

Everything above is still an assertion until it is located precisely, and there is a clean way to locate it. The year 3 total is not in dispute and production is not being changed, so keep Sankhya year 3 output at exactly the Rs 19,26,288 crore it was. Then ask what value added would have been if the wedge had simply kept its year 2 share of output instead of being raised. In year 2 the wedge of Rs 1,20,000 crore was 6.80 per cent of output of Rs 17,64,000 crore. Carry that same share into year 3 and the wedge becomes Rs 1,31,040 crore, the year 2 wedge grown at the same 9.20 per cent as everything else.

Now recompute. Value added would have been Rs 19,26,288 crore less Rs 1,31,040 crore, or Rs 17,95,248 crore. Set against year 2 value added of Rs 16,44,000 crore, that is Rs 1,51,248 crore more, and it works out at 9.20 per cent exactly. Hold the wedge at its old share and value added grows 9.20 per cent, precisely the same as output, and the 1.76 point gap disappears to nothing. In the published year the wedge took 6.80 per cent of output and then 8.31 per cent of it, and that rise in the share is the whole of the gap.

The obvious version of that counterfactual gives a different answer, and it is worth working because most readers reach for it first. Freeze the wedge in rupees instead, at the year 2 figure of Rs 1,20,000 crore, and value added becomes Rs 19,26,288 crore less Rs 1,20,000 crore, or Rs 18,06,288 crore. Against Rs 16,44,000 crore that is a rise of 9.87 per cent. The freeze does not close the gap at all. The freeze reverses it, putting value added 0.67 points ahead of output. A wedge frozen in rupees is a shrinking slice of a growing economy, so freezing the amount overshoots while freezing the share lands exactly. The two ideas sound interchangeable and are not: one lands the gap on nil and the other lands it 0.67 points the wrong side of nil.

The working counterfactual established something narrower than it looks, and easy to oversell. Hold the wedge at a constant share of output and the two measures must grow together: that is arithmetic, true by the way the identity is built, and no year of Sankhya data could have come out any other way. So the counterfactual is not independent evidence that policy caused anything, and treating an identity rearranged as a second opinion is one of the commonest ways a check gets mistaken for a confirmation. The counterfactual does establish where the 1.76 points live: entirely in a rise in the wedge's share of output, from 6.80 per cent to 8.31 per cent, and nowhere in what Sankhya produced. The claim is smaller than proof of a cause, and it is the one the arithmetic actually supports.

Two more readings from the same arithmetic turn the case into a portable rule. First, the size. Of the Rs 40,000 crore by which the wedge rose, Rs 11,040 crore is what a wedge holding its share of a growing economy would have taken anyway, and the remaining Rs 28,960 crore is the excess. The excess, set against year 2 value added of Rs 16,44,000 crore, is 1.76 per cent, the gap to the last decimal. Second, the direction. The gap is nil when the wedge grows at the same rate as value added, positive when the wedge grows faster, and negative when the wedge grows slower, so nothing about the arithmetic makes output the permanently faster measure.

Readings taken off that panel, written down so they survive without it. At the published setting the gap is 1.76 points with output ahead. Leave year 2 taxes and subsidies exactly as they were, at Rs 1,84,000 crore and Rs 64,000 crore, and output grows 6.93 per cent against value added at 7.44, so the gap is minus 0.51 and value added is now the faster measure. Push subsidies to Rs 1,20,000 crore with taxes at Rs 1,80,000 crore and output growth falls to 3.53 per cent, a gap of minus 3.91 points. Value added grows faster than output whenever subsidies on products are raised enough for the wedge to grow more slowly than value added, so anyone who has learned the Sankhya year as a rule about which measure runs ahead has learned the wrong lesson from the right example.

Which of the two applies, and to what?

Neither measure is the more correct one, and any argument that starts by trying to crown one of them has misunderstood what they are. The two measures answer different questions, so the choice follows from the question rather than from a preference. Value added is measured in what producers receive, so it is the right base for anything about production: which sectors expanded, how much a worker produced, how sectoral outputThe output of one part of the economy on its own, such as manufacturing, construction or trade, rather than the economy as a whole. is shifting between activities, and whether a rise in an industry came from making more or from charging more. A tax change would contaminate every one of those readings, and value added keeps it out.

Output at buyer prices is the measure of the economy at the prices actually paid, and it is the one called for when the question is about size, about the burden something represents, or about comparison across countries whose tax settings differ. Buyer prices are also what most headline growth figures are published on and what most international comparison uses. Buyer prices are therefore the number a reader is most likely to meet without being told which measure it is. The mistake is never picking the wrong measure. The mistake is comparing a figure on one measure against a figure on the other without saying so, and the difference will then be read as a real change.

What goes wrong when a chart switches measure part way through?

Both measures are published, both are called growth, and both are quoted as percentages with no unit to give the game away. The absence of any unit is what makes the next failure so easy to walk into. A chart carries a series of growth rates over several periods. Part way along, at the point where the tax and subsidy settings changed, the series is put together on the other measure. The line steps down by something close to the size of the wedge change and stays there, and the picture now shows a slowdown that never happened. A step that arrives exactly at a measurement change is the first thing that measurement change would produce, so check which measure a series is on before reading any turning point in it.

The tell is almost always the date. Real slowdowns are messy: they begin in one sector, spread unevenly, and rarely land as a clean step between two adjacent periods. A break that arrives precisely at the quarter a tax setting changed, in every sector at once, with the same size everywhere, is a measurement event wearing the clothes of an economic one. The change of measure therefore has to be labelled on the chart itself rather than mentioned in a footnote nobody reads.

How does an analyst actually use the split?

Practitioners do not treat the two measures as a definitional curiosity. Practitioners treat the difference between the two as a diagnostic, and the routine is short. When output growth and value added growth diverge sharply in the same period, the first question is never what happened to the economy. The first question is what happened to product taxes and subsidies, the one thing capable of producing a divergence without anything real occurring. Only when the wedge turns out to be steady is there something economic to explain, and at that point the divergence is worth real work.

The routine matters most to three readers in particular. An equity analyst building a sector view needs value added. A sector reading contaminated by a tax change will not reconcile with what the businesses in that sector report, and the analyst will spend a week hunting for a company level explanation of a national accounts artefact. A lender sizing exposure to an industry wants the same thing, for the same reason. And anyone comparing one country against another needs to know which measure each figure came from before the comparison means anything at all. The split is most useful as a first question rather than a final answer: check the wedge, and only then start looking for an economic story.

The Republic of Sankhya is invented.
Educational material. Not advice on any investment, tax, budget or market position.