Imagine a very small economy that produces only bread. In one year, its bakery makes 100 loaves. Each loaf sells for 10 units of money. The money value of that year's bread is therefore 1,000 units.
In the next year, the bakery makes 110 loaves and each loaf sells for 12 units. The money value is now 1,320 units. It has risen by 32 per cent.
But the bakery did not produce 32 per cent more bread. It produced only 10 more loaves, which is a 10 per cent increase. The rest of the rise in money value came from the higher price.
This is the central problem of this chapter. The money value of production can rise because more goods and services were produced, because their prices rose, or because both happened. A money total alone cannot tell us how much production changed.
We therefore need three connected measures. One records production at the prices of the period being measured. Another values the quantities at a steady comparison price basis. A third shows the price movement that connects the first two. The idea begins with ordinary multiplication:
`value = price × quantity`
The same logic remains true for a large economy. The calculation becomes harder because an economy produces millions of changing goods and services rather than one kind of bread. But the question does not change: how much of the change in money value came from production, and how much came from prices?
Separate the money total from the amount produced
Return to the bakery. Its first-year output was worth 1,000 units because 100 loaves were valued at 10 each. Its second-year output was worth 1,320 units because 110 loaves were valued at 12 each.
The value calculated with each year's own prices is called a current-price or nominal value. The second-year nominal value uses the second year's price and quantity. It answers a useful question: what was that period's production worth at the prices then prevailing?
To see how the amount produced changed, hold the comparison price steady. Value the second year's 110 loaves at the first year's price of 10. The result is 1,100 units. This is the second year's production expressed on the first year's price basis.
`110 loaves × 10 units = 1,100 units`
The move from 1,000 to 1,100 reflects a 10 per cent rise in output. The price used for both periods is the same, so a price rise cannot create this difference.
A measure constructed to show this change is called a volume measure. It is also commonly called a real or constant-price measure. These labels do not mean that actual prices stayed constant. The bakery's actual price rose from 10 to 12. Only the price basis used for comparison was held steady.
This distinction prevents a common error. Nominal production rose by 32 per cent, while real production rose by 10 per cent. Saying that the economy “grew by 32 per cent” would confuse a rise in money value with a rise in output volume.
The two measures can even move in opposite directions. Nominal value can rise while real output falls if prices rise enough. Real output can rise while nominal value falls if prices fall enough. The direction of one measure cannot be assumed from the other.
Volume means more than counting objects
For identical loaves, quantity is easy to count. Most production is not so simple. A hospital provides treatments of different kinds. A software service may become more capable without producing more physical units. A new machine may last longer or work faster than the machine it replaces.
An economic volume measure therefore includes quantity, quality and the composition of output. Ten better machines may provide more productive service than ten older machines. A shift from simple to complex medical treatment can change measured volume even if the number of patients stays the same.
Calling the result “real” does not make it a direct physical observation. Statisticians must decide how to recognise quality change, combine unlike products and estimate activities that are incompletely observed. A real measure is a carefully constructed comparison, not a quantity that exists without assumptions.
See value, volume and price as three indices
An index makes change easier to compare by setting one period equal to 100. In the bakery example, let the first year be 100.
The nominal value rises from 1,000 to 1,320. Its value index is therefore 132. This means that the money value is 32 per cent above the first-year value.
The constant-price value rises from 1,000 to 1,100. Its volume index is 110. This means that production volume is 10 per cent above the first-year volume.
The loaf price rises from 10 to 12. Its price index is 120. This means that the price is 20 per cent above the first-year price.
These three changes fit together by multiplication. The output factor is 1.10 and the price factor is 1.20. Their product is 1.32, the factor for nominal value.
`1.10 × 1.20 = 1.32`
This is why subtracting the real growth rate from nominal growth is not exact. Thirty-two per cent minus 10 per cent gives 22 per cent, not the correct 20 per cent price change. The exact calculation divides the value factor by the volume factor.
`1.32 ÷ 1.10 = 1.20`
Index points and percentage change must also be kept separate. If an index rises from 120 to 126, it rises by six index points. Its percentage increase is only 5 per cent because six is 5 per cent of the starting value of 120.
Derive the GDP deflator from the same production
The bakery used one directly observed price. An economy-wide production measure combines many different price movements. The broad price measure built into the comparison between nominal and real gross domestic product is called the GDP deflator.
The deflator is obtained after nominal GDP and real GDP have been estimated on compatible boundaries. It is not usually created by visiting shops and pricing one fixed GDP basket. This is why it is described as an implicit price measure.
The relationship is:
`GDP deflator = nominal GDP ÷ real GDP × 100`
In the bakery economy, nominal output in the second year is 1,320 and real output is 1,100. Dividing one by the other and multiplying by 100 gives a deflator of 120. This matches the 20 per cent price increase in the one-product example.
The identity can also be rearranged:
`nominal GDP = real GDP × GDP deflator ÷ 100`
If a value of 1,320 is deflated by the price factor 1.20, the result is 1,100. The arithmetic works in both directions because the nominal and real measures describe the same production boundary and period.
In a real economy, statisticians do not normally apply one simple price index to the entire current-price total. Different parts of output have different products, prices and data. A suitable price index or volume indicator is matched to each part before the pieces are combined. Where both output and intermediate inputs are important, their volume changes may need to be estimated separately. This avoids assuming that their prices always move together.
Converting output and intermediate consumption separately before subtracting one from the other is called double deflation. The resulting volume value added is the difference between two large, separately estimated components. A modest change in either estimate can therefore create a noticeable change in value added. This sensitivity calls for careful interpretation; it does not by itself show that the method has failed.
Why the GDP deflator is not CPI or WPI
The GDP deflator, the Consumer Price Index and the Wholesale Price Index all describe prices, but they answer different questions.
The GDP deflator covers final goods and services produced within the domestic production boundary. Its composition can change as the pattern of domestic output changes. A newly important service can gain weight because it has become a larger part of production.
The Consumer Price Index, or CPI, follows the prices faced by households for a defined consumption basket. It is designed to examine consumer prices and living costs, not the price of everything produced domestically. An imported consumer good can affect CPI because households buy it. The imported good is not itself domestic output covered by the GDP deflator.
The Wholesale Price Index, or WPI, tracks selected goods at the wholesale or producer-facing stage. Its coverage does not include the full range of domestic final services measured in GDP. It therefore cannot be treated as another name for the GDP deflator.
The measures also use different weights. CPI weights follow household consumption patterns. WPI weights follow its selected goods. The GDP deflator's weights reflect the changing composition of measured domestic output. For these reasons, their inflation rates can move differently without any calculation being wrong.
The GDP deflator is useful for separating the value and volume of domestic production. It is not automatically the best measure of the cost of living faced by a household. The detailed construction and use of consumer, wholesale and producer price indices belong to the inflation chapters.
Understand what a base year actually does
The bakery calculation valued both years at the first year's price. That first year supplied the comparison price basis. In a fixed-base system, a base year provides the prices and often the weights used to compare output across periods.
The year being measured is not automatically the base year. We could measure production in a later comparison year while still expressing it at the base year's prices. The phrase “at constant prices” means “valued on the chosen comparison basis,” not “observed in a world where prices never changed.”
Three time ideas must therefore be distinguished. The base period supplies the valuation or weighting structure. The reference period is the period whose index is displayed as 100. The comparison period is the period whose level or change we are examining.
These periods can coincide, but they need not always do so. An index can be re-referenced so that a newer year equals 100 without rebuilding all the weights and methods underneath it. Re-referencing changes the displayed scale. Rebasing is a deeper statistical operation.
Weights matter because the economy produces unlike things. If food forms a large part of output, its movement has more influence on the total than the same percentage movement in a very small activity. A base-year structure gives each component a consistent role in the combined measure.
The problem is that economies change. New industries grow, old products disappear and relative prices shift. If the fixed base becomes too distant, old weights can describe a production pattern that no longer exists. A computer service that was once tiny may have become important, while an older product may have lost most of its market.
Move carefully from fixed bases to chain volume
A fixed-base comparison uses one base structure across a span of years. Its great advantage is clarity. Each period is compared through the same set of prices or weights, and the components can often be added in a straightforward way.
Its weakness grows with distance from the base. The farther the economy moves from the original production pattern, the less representative the old weights may become. Large changes in relative prices can also distort a comparison based on a distant year.
A chain volume approach updates the comparison more frequently. It first measures volume change between nearby periods using relevant weights. It then links, or chains, these short comparisons to create a longer series. This lets the weights follow structural change more closely.
Chain linking improves the relevance of weights, but it introduces a cost. Outside the displayed reference period, independently chained components may not sum exactly to the chained total. This property is called non-additivity. It is a mathematical feature of the linking method, not missing production.
Suppose a reader adds the chain-volume values for agriculture, industry and services and obtains a figure slightly different from total chain-volume GVA. The first question should be whether the published series is non-additive, not whether one sector was omitted. Current-price components, fixed-base volume components and chain-volume components cannot be mixed casually.
No single method removes every measurement problem. Fixed-base measures can rely on outdated weights. Chain measures can be harder to add. The choice depends on the purpose, the available evidence and the way the national accounts are constructed.
Accept that measuring a changing economy requires judgement
The small bakery kept the product unchanged. A real economy never does. Products improve, new products appear and old ones vanish. Services often have no simple physical quantity. Discounts, digital delivery and bundled products make the relevant price harder to observe.
Quality is especially important. A phone may become more expensive while gaining a better camera, longer battery life and more computing power. The entire price increase should not automatically be treated as pure inflation. Part of it may pay for a better product. Separating quality from price requires evidence and a method.
New goods create another difficulty. A new service has no price in an earlier year. A disappearing product has no price in a later year. The index must introduce or remove such items without pretending that the same unchanging basket exists forever.
Informal and small-scale activity can be difficult to observe regularly. Surveys, administrative records and benchmark studies may cover different parts of this activity at different times. A new survey can change the estimated level or growth of output because it supplies better information about work that was always present.
Changing product quality, unrecorded activity and imperfect source data create uncertainty. They do not make real GDP meaningless. They explain why a published estimate is conditional on its definitions, evidence and method. A responsible user should read it as an estimate rather than as a timeless, perfectly observed physical fact.
See rebasing as maintenance, not as rewritten history
Statistical agencies periodically rebuild a national-account series around a newer base. This process is called rebasing or a base revision.
A base revision can update far more than the year printed beside “constant prices.” It may introduce newer weights, product and industry classifications, price measures, surveys, administrative records and estimation methods. It may improve the treatment of new sectors or replace a weak proxy with a more direct source.
The physical past does not change when a series is rebased. What changes is the statistical lens used to estimate and compare that past. If better evidence suggests that an activity was larger, smaller or changing at a different pace, historical estimates may change.
Rebasing does not mechanically raise the growth rate. New data and weights can raise some estimates and lower others. The direction depends on how the revised structure differs from the old one. Treating every upward change as proof of improvement, or every revision as proof of manipulation, ignores the actual method.
A new base also creates a comparability problem. Values and growth rates from the old and new series may use different classifications, coverage, prices and methods. Simply joining them at the changeover date can create a false jump. A back series applies a sufficiently compatible framework to earlier periods so that longer comparisons can be made.
Re-referencing must again be separated from rebasing. If an index is merely rescaled so that a different year equals 100, its percentage path does not change. A base revision can change weights, methods and the estimated path itself.
As of August 2026, India's official annual and quarterly national-account series uses FY 2022–23 as its base year. The series was introduced on 27 February 2026 and replaced the FY 2011–12 series. These dates identify the applicable series; they are not current GDP values or growth forecasts.
Read every estimate together with its vintage
An annual GDP number often appears before all records for that year are complete. Early estimates may rely on partial surveys, budget information, company reports, production indicators and assumptions about the remaining period. Later releases can use fuller evidence.
Each release is therefore a vintage of the estimate. Labels such as advance, provisional, revised and final tell the reader how far the estimate has moved through the release process. “Final” means final within that stated cycle. It does not guarantee that no later base revision or methodological change will ever alter the historical series.
The annual release design checked through August 2026 describes successive stages called First Advance Estimate, Second Advance Estimate, Provisional Estimate, First Revised Estimate and Final Estimate. The annual estimate for FY 2025–26 released on 5 June 2026 carried the Provisional Estimate label. The fiscal year identifies the flow period; 5 June 2026 identifies the release date; “Provisional” identifies the vintage.
Regular revision and base revision are different. A regular revision updates an estimate as more complete information arrives within the existing series framework. A base revision can change the framework, sources, classifications and methods. Both can change a published number, but for different reasons.
Annual and quarterly estimates are also connected. Early quarterly estimates often use timely indicators because complete annual evidence is unavailable. When a stronger annual benchmark arrives, the quarterly path may be revised so that the quarters and the annual total remain consistent. A revision to one quarter does not stand alone if the whole year's benchmark has changed.
A precise estimate should therefore travel with at least its reference period, release date, estimate status, price basis, base or method, and vintage. Two figures with the same label “GDP growth” may be incomparable if any of these features differ.
Revision is not evidence of fraud by itself. Refusing to revise after better evidence arrives would preserve a known weakness. At the same time, a revision should be open to examination: users should be able to identify what changed in data, definition, coverage or method.
Compare annual and quarterly growth on the right denominator
A growth rate is incomplete until we know what the new level was compared with. Annual, year-on-year, quarter-on-quarter and annualised rates can describe the same economy yet give very different numbers.
An annual growth rate compares one full year with another full year. Each annual level normally combines production across all the quarters in that year. A flow for four quarters is a sum; multiplying only the fourth quarter by four is not the same calculation unless a special annualised convention is explicitly being used.
A year-on-year quarterly rate compares a quarter with the corresponding quarter one year earlier. The April–June quarter is compared with the previous April–June quarter. This comparison reduces some seasonal mismatch because the two quarters occupy the same part of the year.
A quarter-on-quarter rate compares a quarter with the immediately preceding quarter. It captures recent momentum more quickly, but raw quarters can follow strong seasonal patterns. Festival production, harvests, weather, holidays and the number of working days can make neighbouring quarters naturally different.
Seasonal adjustment estimates and removes recurring seasonal and calendar patterns so that underlying short-run movement is easier to see. It does not create real production. It changes the comparison basis. A seasonally adjusted rate must not be placed beside an unadjusted rate as though their treatment were identical.
Seasonal adjustment also cannot remove every irregular shock or measurement error. An adjusted series is an estimate of the underlying short-run path, not a perfectly cleaned observation.
An annualised quarterly rate takes a short-period change and expresses the rate that would result if that pace continued and compounded for a year. If a seasonally adjusted quarter grows by 1 per cent, simply calling this “4 per cent annual growth” is only an approximation. Compounding gives about 4.06 per cent.
The annualised figure is not a forecast that the pace will continue. It is a rescaling convention. A year-on-year rate, a quarter-on-quarter rate and an annualised quarter-on-quarter rate use different denominators and cannot be compared directly without conversion.
Calendar effects can matter even beyond the usual seasons. A quarter may contain a different number of working days, or a movable holiday may fall in another quarter. A raw change can therefore combine economic momentum with the arrangement of the calendar.
Keep levels, growth rates and contributions separate
An output level records an amount produced during a period. A growth rate reports the proportional change from a compatible starting level. A higher growth rate does not necessarily mean a larger economy.
Suppose Economy A has real output of 1,000 units and grows by 2 per cent, while Economy B has output of 100 units and grows by 10 per cent. Economy A remains far larger after the change even though Economy B grows faster. Size and speed answer different questions.
A lower positive growth rate also does not mean output fell. If output rises from 100 to 120, growth is 20 per cent. If it then rises from 120 to 126, growth is 5 per cent. Growth slowed, but the level continued to rise.
A negative rate says that the new level is below its stated comparison level. It does not prove that output is below every earlier year. If output rose from 100 to 130 and then fell to 120, the latest growth rate is negative against 130, but output remains above the original 100.
Consecutive percentage changes compound. If output of 100 grows by 10 per cent and then by 20 per cent, it becomes 132. The cumulative rise is 32 per cent, not 30 per cent.
`100 × 1.10 × 1.20 = 132`
The same rule explains why a 20 per cent fall does not cancel a 20 per cent rise. A value rising from 100 to 120 and then falling by 20 per cent ends at 96 because the fall is calculated from the newer base of 120.
Base effects change the rate, not the truth of the change
Every growth rate has a denominator. An unusually low comparison level can make a later recovery produce a high percentage rate. An unusually high comparison level can make a similar absolute increase produce a lower rate. This influence of the starting denominator is called a base effect.
A base effect does not make growth fake. It tells us why the percentage rate may look dramatic. The answer is to examine both levels and rates over several comparable periods, not to discard the arithmetic.
The term “base” here means the denominator used in a growth comparison. It is not necessarily the national-account base year used for constant-price weights. Confusing these two meanings produces needless mistakes.
A large sector can contribute more while growing more slowly
A sector's share tells us its size relative to the total. Its growth rate tells us how fast its own comparable output changed. Its contribution to aggregate growth depends on both its starting weight and its growth.
Consider a fictional two-sector economy. Sector A forms 80 per cent of output and grows by 2 per cent. Sector B forms 20 per cent and grows by 5 per cent. In a compatible weighted illustration, Sector A adds 1.6 percentage points to aggregate growth, while Sector B adds 1 percentage point.
The smaller sector grows faster, but the larger sector contributes more. Exact published contributions can be affected by changing weights, chaining and revisions, so shares and rates from incompatible series should not be multiplied casually.
Per-capita real output is still only an average
Once output has been made comparable across time, it can be divided by a compatible resident population estimate. Real output of 1,100 units divided by 100 residents gives 11 units per person in a fictional example.
This calculation helps separate changes in total output from changes in population. If population grows faster than real output, real output per person can fall even while total real output rises.
The result remains an arithmetic average. It does not show how output or income is distributed, whether unpaid work increased, whether environmental damage occurred, or whether people's health and security improved. Those are development and welfare questions, not hidden meanings inside real GDP per capita.
Read a growth statement as a complete measurement sentence
A sound growth statement identifies what was measured, how prices were handled, and which periods were compared. It also identifies the series, estimate vintage and seasonal treatment when those features matter.
The same discipline applies when GDP is a denominator. Suppose a debt amount is unchanged but revised GDP is larger. The debt-to-GDP ratio will fall because its denominator changed, not because any debt was repaid. Every ratio using GDP inherits relevant revisions to GDP.
“GDP rose” is incomplete. It could describe a rise in current-price value caused mainly by prices. “Real GDP grew” is still incomplete if the denominator and period are missing. “Quarterly real GDP grew” remains ambiguous until we know whether the comparison is year-on-year or quarter-on-quarter and whether seasonal adjustment or annualisation was used.
The careful reader therefore asks a short sequence of connected questions. Is this a nominal value or a real volume measure? Which production boundary and price basis apply? What are the new and comparison periods? Is the rate annual, year-on-year, quarter-on-quarter or annualised? Was seasonal adjustment used? Which base, method and estimate vintage apply?
These questions do not weaken the usefulness of GDP. They make the number interpretable.
The entire chapter returns to one simple story. Money value combines price and production. Current-price GDP keeps both changes. Real GDP uses a steady comparison basis to show volume change. The GDP deflator connects the compatible nominal and real totals through an implicit price measure.
Index weights and base periods organise millions of different products. Chain methods can keep weights relevant, though their components may not add exactly. Rebasing updates the statistical picture as the economy changes. Revisions replace early information with fuller evidence. Quarterly and annual rates become meaningful only when their denominators and adjustments are stated.
No single precise estimate is timeless. But a measure with a clear boundary, price basis, comparison period, method and vintage can answer a precise economic question. That is the discipline that turns a large number into an understandable account of change.