Gross Domestic Product: What It Counts, What It Misses, and How Real Against Nominal Growth Is Read

Gross domestic product (GDP) is the value of everything produced inside a country in a period, counted once. Nominal GDP measures it at that year prices, so it climbs when prices climb. Real GDP holds prices at a base year, so it climbs only when more is actually made. Almost every argument about growth turns out to be an argument about which of the two somebody quoted.

Underneath that answer sits one idea and one problem. The idea is that a country worth of unlike things can be added up by measuring each one in money instead of in tonnes and hours. The problem is that money itself moves. A rupee of output this year and a rupee of output last year are not the same amount of stuff, so a total measured in rupees rises for two entirely different reasons, and the whole apparatus below exists to pull those two reasons apart.

A reader arrives here knowing what a price is, what a market is and how a price is settled between buyers and sellers. Because this is the first aggregate, nothing about aggregates is assumed. The definition comes apart word by word below, followed by what the measure was never built to see, the three routes to the same total and why they disagree slightly, the separation of a nominal reading from a real one, exactly how much the popular shortcut for combining them is wrong by, and why an economy can add output in a year that adds no work at all.

The Republic of Sankhya, a country constructed for this walkthrough, supplies every rupee crore that follows. Its figures were chosen so the arithmetic lands exactly rather than roughly. Real measurement behaves the other way round: levels arrive with a margin of error and rates rarely divide cleanly. Every rate below is recomputed from the two levels it sits between, and the panel further down recomputes the central one live as its two dials move.

What is Gross Domestic Product, and what is each of its five words doing?

Gross domestic productThe total value of goods and services produced inside a country during a stated period, with each item counted once. Gross means before any allowance for the machinery and buildings worn out along the way. is a sentence pretending to be an acronym, and the sentence is short enough to take apart a word at a time. Slowly, then: the value of everything produced inside a country in a period, counted once. Five separate instructions are hiding in there, and each one rules something out. Missing any of them means misreading a published figure without ever knowing it.

Start with value. A country makes onions, haircuts, cement, bus rides and software in the same twelve months, and there is no unit that adds those together. Kilograms will not do it and hours will not do it. Money will. Everything on that list has a price, so each item can be converted into rupees and the rupees can be added. Pricing everything is the foundation of the whole measure, and it plants the first limitation too. Anything without a price becomes invisible to the total.

Then produced. The measure counts making, not changing hands. When a neighbour sells a five year old scooter, a payment happens, but no scooter came into existence this year, so nothing is added. The dealer commission on that sale is a service produced this year and does count. The test is never whether money moved, it is whether something new came into being in the period being measured. Inside a country is a statement about geography and not about who takes the profit home: a plant standing on Sankhya soil counts in Sankhya output whoever the shareholders are and wherever they live. In a period makes it a flow rather than a stock. A flow is why output is quoted per year and never as an accumulated pile.

Why does counting the flour and the bread in full count the same wheat three times?

One short street in Sankhya carries the whole point. A grower sells wheat to a mill for Rs 20/-. The mill grinds it and sells flour to a bakery for Rs 32/-. The bakery bakes and sells bread over the counter for Rs 50/-. Three producers, three sales, and a very tempting mistake sitting right there: adding the three sale values gives Rs 102/-.

Look at what that Rs 102/- actually contains. The grower Rs 20/- of wheat is inside the wheat sale, inside the flour sale and inside the bread sale, so it has been counted three times over, contributing Rs 60/- to the total. The mill own contribution of Rs 12/- sits inside the flour sale and inside the bread sale, so it has been counted twice, contributing Rs 24/-. The bakery Rs 18/- is counted once. Sixty plus twenty four plus eighteen is exactly the Rs 102/-, so the inflation of the total is not vague, it can be traced rupee by rupee.

The repair is to count only what each producer adds. The name for it is value addedThe sale value of what a producer sells, less the cost of the goods and services it bought in from other producers to make it. What is left is the part that producer itself created.: sale value less the cost of everything bought in from somebody else. The grower adds Rs 20/-. The mill adds Rs 32/- less Rs 20/-, or Rs 12/-. The bakery adds Rs 50/- less Rs 32/-, or Rs 18/-. Twenty plus twelve plus eighteen is Rs 50/-, the price of the bread. The sum of what everybody added is exactly the value of the final good. Value added and final output are therefore two routes to one number rather than two different measures. Double countingAdding the same production into a total more than once, which happens when the sale values of the steps in a chain are added instead of what each step contributed. is what the arithmetic on the left is called, and every national accounting system on earth is built to prevent it.

What does the measure not count, and is any of that a fault in it?

Four things sit outside it. Naming them precisely matters because vague complaints about the measure usually turn out to be one of these four stated badly.

The first is unpaid work. A parent who cooks two meals a day, cleans the house and cares for an elderly relative has produced a great deal, and none of it carries a price, so none of it enters the total. Hire somebody to do exactly the same tasks and the total rises, though the amount of cooking and cleaning in the world has not changed by a single meal. The second is anything produced outside the recorded economy: the cash trade nobody writes down, the workshop that never registered. Real production, invisible to the counters. The third is depletion, meaning whatever was used up to produce the output, from a machine that wore out to a forest that did not grow back. The word gross in the name is the honest admission of this: gross means before wear and tear is subtracted. The fourth is distribution. The total says how much was produced, never who received it, so two countries with identical output can have completely different lives inside them.

None of those four is a fault. Gross domestic product never claimed to measure welfare, sustainability or fairness, and a measure cannot fail at a job it was not built to do. The genuine risk is not the measure, it is the reader who treats one number as a report card on a country. Held to its scope, the number becomes useful again: this is a production counter, it counts what was made and priced, and it stops there.

How to read how GDP is calculated: which of the three routes produced the figure in hand?

There is one total and three ways to arrive at it, and the three are not rival estimates of different things. The three are different piles of paperwork that ought to end at the same place.

The production route adds up what each producer contributed, the value added arithmetic from the bread chain repeated across every producer in the country. The expenditure route adds up what was spent on final output: what households bought, what businesses put into new plant and stock, what government spent, and what foreigners bought from the country less what the country bought from abroad. The income route adds up what was earned in the making of it: wages, the operating surplus of businesses, and the mixed income of people who work for themselves and cannot separate their wage from their profit. The three agree in principle for a plain reason. Every rupee of value somebody produced was spent by somebody buying it, and became income for somebody making it. Same rupee, three vantage points.

In practice the three routes disagree. Each route is built from different records collected by different means at different times, so the gap between them is a measurement residual rather than a mystery. Suppose the production route lands Sankhya year 3 output at Rs 19,26,288 crore and the expenditure route, added up separately, reaches Rs 19,30,000 crore. A gap of Rs 3,712 crore is 0.19 per cent of the total. Nothing is wrong: two enormous counting exercises reached figures within a fifth of a per cent of each other. Statistical systems normally publish one route as the headline and carry the discrepancy openly as a line of its own. The route a figure came from is worth asking about. The routes are compiled on different timetables and revised on different schedules, so two published numbers for the same year can differ without either being an error.

Nominal GDP vs Real GDP: what is each of them, and only then, how do they differ?

Take them one at a time first. Most of the trouble in this area comes from meeting the two as a pair and never learning what either one is by itself.

NominalMeasured in the money of the day, with no adjustment for the fact that money buys different amounts of goods at different times. gross domestic product is the total measured at the prices actually charged in the year being measured. Every item counted is valued at what it sold for that year. Nominal output is the honest arithmetic of the year as it was lived, and it is what summing real invoices would give. Because prices are part of it, a nominal total climbs when more is made, when prices rise, or when both happen at once, and by itself it never says which.

RealMeasured at the prices of one chosen year, so that the figure moves only when the quantity produced moves and not when prices do. gross domestic product is the same production valued at the prices of one fixed year, the base yearThe year whose prices are held fixed so that quantities from several years can be compared on one price ruler. Statistical systems change the base year occasionally, and a change of base changes the level of every real figure in the series.. Hold the price ruler still and the total can only move when the quantities move. Real output is an artificial figure because nobody paid those prices in the later years. And that is exactly the point: the figure answers one question only, whether more stuff was produced.

Now set them side by side on Sankhya. Real output was Rs 16,00,000 crore in year 1 and Rs 16,80,000 crore in year 2. Dividing 16,80,000 by 16,00,000 gives 1.05, a rise of 5.00 per cent. Nominal output was Rs 16,00,000 crore and then Rs 17,64,000 crore. Dividing 17,64,000 by 16,00,000 gives 1.1025, a rise of 10.25 per cent. Same country, same year, same production, and two published growth rates that differ by a factor of two, with the entire difference being that prices in Sankhya rose 5.00 per cent that year. The household version of this is a monthly grocery bill. The bill goes from Rs 8,000/- to Rs 8,800/-, up 10 per cent, and yet the trolley leaving the shop holds the same items in the same quantities. Nominal grocery spending rose 10 per cent. Real grocery consumption did not move at all.

Why is nominal growth not simply real growth plus inflation?

The arithmetic that follows costs half a minute to learn, and it removes a small permanent error from every growth figure read afterwards.

Growth compounds, it does not add. Output in rupees this year is last year output multiplied by one plus real growth, and then multiplied again by one plus the price change. So nominal growth is one plus the first, times one plus the second, less one. On Sankhya year 2 that is 1.05 times 1.05, or 1.1025, so nominal growth is 10.25 per cent, not 10. On year 3 it is 1.04 times 1.05, or 1.0920, so nominal growth is 9.20 per cent, not 9. The extra bit is the cross term, and it is exactly the two rates multiplied together: 5 times 5 divided by 100 is 0.25 points, and 4 times 5 divided by 100 is 0.20 points. In plain language the cross term is the price rise applied to the new output as well as to the old. The extra production made this year is also sold at this year higher prices.

The gap is small when both rates are small and it grows with both of them, so the shortcut is a habit that behaves well in the conditions where accuracy hardly matters and fails hardest exactly where it does. At 2 per cent and 2 per cent the shortcut is wrong by 0.04 points, an error nobody would notice. At 20 per cent and 20 per cent it is wrong by 4.00 points, a fifth of the entire number. And the shortcut can flip a sign. If real output rises 5 per cent while prices fall 5 per cent, the shortcut says nominal output was flat, when in fact 1.05 times 0.95 is 0.9975, so nominal output fell by 0.25 per cent. The rupee total went down in a year that produced more.

What is the deflator doing in this arithmetic, and what is it not doing here?

One number carries a total across from one series to the other. The deflatorA price index for everything an economy produces, set to 100 in the base year, used to convert a total measured at current prices into one measured at base year prices. is a single number for the price level of everything produced, set to 100 in the base year. Real output times the deflator, divided by 100, is nominal output. Turn it around and nominal output divided by the deflator, times 100, is real output.

Work it once on Sankhya so the mechanism is not abstract. In year 3 the deflator stands at 110.25 against the base year 100, so prices for output as a whole are 10.25 per cent above the base year. Real output of Rs 17,47,200 crore multiplied by 110.25 and divided by 100 gives Rs 19,26,288 crore, the nominal figure. And the year on year price change comes out of the deflator itself: 110.25 divided by 105.00 is 1.05, so prices rose 5.00 per cent in year 3, exactly as they did in year 2.

The deflator is the lever that converts one series into the other, and nothing more than that. How the price level is actually measured, what prices go into it, how a basket is chosen and refreshed and why different price measures disagree with each other, is a full subject of its own, taught under inflation and prices.

Employment Growth: what is it, and what does it have to do with output?

Employment growthHow far the count of people in work has moved from one period to the next, usually stated as a percentage of the earlier count. is the simplest measure in this guide. The measure counts the people working at the close of a year, counts them again at the close of the year before, and states the difference as a percentage of the earlier count. Sankhya had 8.00 crore people working in year 1 and 8.00 crore in year 2, so growth was nil. In year 3 it had 8.32 crore, and 8.32 divided by 8.00 is 1.04, so employment grew 4.00 per cent.

Set next to the output series, that count turns up something instructive. Year 2 delivered real growth of 5.00 per cent with not one additional worker. Output divided by workers shows where it came from. Year 1 gives Rs 2,00,000/- for each worker. Year 2 divides a larger Rs 16,80,000 crore by an unchanged 8.00 crore and gives Rs 2,10,000/-, so each worker produced 5 per cent more. Year 3 is the mirror image. Output rose 4.00 per cent, workers rose 4.00 per cent, and dividing Rs 17,47,200 crore by 8.32 crore returns Rs 2,10,000/- again, unchanged. Every point of year 3 growth came from more people working, and none of it from any worker producing more.

Output growth and employment growth are two separate series that happen to be published side by side, and neither one can be read off the other. Think of two workshops on the same lane. The first buys a second machine and hires nobody, and its output rises. The second hires two more hands to work the machine it already has, and its output rises by a similar amount. Both grew. One created work and one did not. A reader who sees only the output line cannot tell which lane they are looking at. How the two series relate to each other in full, and what a year of growth without jobs actually means for people, is taken up under output growth and employment.

How do all three Sankhya years look at once, with every rate worked from the levels?

Here is the whole case in one place. Read it downward and every rate in the table can be checked against the two levels sitting above it. Checking a rate against its own levels is the only way to hold a growth figure honestly.

Two checks are worth doing before moving on. Rs 17,47,200 crore multiplied by 110.25 and divided by 100 lands on Rs 19,26,288 crore exactly. And Rs 19,26,288 crore divided by Rs 17,64,000 crore gives 1.0920, the same 1.0920 that 1.04 times 1.05 gives. Every rate in that table is recoverable from the levels above it, and a growth figure whose levels are not visible is a figure that cannot be checked. For a sense of scale, one market inside this economy, the Sankhya onion market at Rs 2,000/- a quintal and 100 lakh quintals, comes to Rs 2,000 crore, or 0.125 per cent of year 1 output. An aggregate is built out of thousands of things that size.

What does an analyst actually do with a growth figure?

Less than might be expected, on its own. A growth rate arriving without its context is close to unusable, and the practitioner move is a short list of questions asked before the number is allowed anywhere near a model.

Which measure is this, real or nominal? Which route produced it? Is this a first estimate or a revised one, and what was the earlier version? And then the one that decides whether the number is any use: what does it imply for the specific line somebody is actually looking at? The most common practical error is a mismatch of measures. A company revenue line is a rupee figure at the prices actually charged, so it is a nominal series, and setting its growth beside real output growth compares two things measured on different rulers. If a lender is sizing whether a borrower is keeping pace with the wider economy, the honest comparator for a rupee revenue line is the nominal series. A borrower whose revenue rose 8 per cent in a year when nominal output rose 10.25 per cent lost ground, even though 8 looks comfortably ahead of the 5.00 per cent real figure that gets quoted in conversation.

The household version is the same discipline in miniature. A salary rose 6 per cent and the earner feels better off. Whether that is so depends on the price ruler. If prices rose 5 per cent the real gain is not 1 per cent but 1.06 divided by 1.05, or 0.95 per cent. Small difference here, and the same arithmetic that matters enormously when both numbers are large. Which prices a figure is measured at is worth establishing every time, before the figure is allowed to mean anything.