GDP can be derived by the production, expenditure and income approaches, which are identical in theory and differ in practice because each is built from different source data. Australia's headline quarterly figure is a simple average of the three in chain volume terms, labelled GDP(A). Annual estimates balanced through the supply-use tables carry no statistical discrepancy at all, but the discrepancy survives for the most recent year, for years before 1994-95, and for every quarterly estimate.
Gross domestic product is defined three ways at once. Each definition is complete, each is capable of producing the whole figure on its own, and each is built from data the other two do not use. In an ideal accounting they return the same number. They do not, and how Australia handles that gap is more interesting than the gap itself.
Key Takeaways
- Production, expenditure and income approaches each measure the whole economy by a different route.
- The headline Australian quarterly figure is a simple average of all three, in chain volume terms.
- Annual estimates balanced through the supply-use tables have no statistical discrepancy at all.
- The discrepancy survives quarterly, for the most recent year, and for years before 1994-95.
- The current price headline is derived from the averaged volume estimate, not averaged separately.
- Quarterly estimates are benchmarked so that they sum to the annual figure.
Three routes to the same total
The production approach measures what was made. It sums gross value added for each industry at basic prices, then adds taxes and subtracts subsidies on products. Its raw material is industry output and the cost of inputs consumed in producing it.
The expenditure approach measures what was bought. It sums all final expenditures, adds changes in inventories, adds exports and subtracts imports. Its raw material is household consumption, business investment, government spending and trade data.
The income approach measures what was earned. It sums compensation of employees, gross operating surplus, gross mixed income, and taxes less subsidies on production and imports. Its raw material is payroll, company profits and tax records.
The equivalence is a matter of identity rather than coincidence. Whatever is produced is bought by somebody, and the money paid for it becomes somebody's income. Every dollar should appear once in each account, from a different side.
Why the identity does not hold in practice
The identity is exact for the economy. It is not exact for the measurement of the economy, because the three estimates are assembled from entirely separate collections, each with its own coverage, timing, sampling error and definitional edges.
Business surveys, household surveys, tax records and trade statistics do not agree perfectly about a quarter. They are not supposed to; they were built for different purposes and they observe different things. The residual left over when the three are compared is the statistical discrepancy.
The word discrepancy invites the reading that somebody has made an error, and that reading is wrong. It is the visible remainder of measuring a very large thing three times by three methods. What would be alarming is a discrepancy of zero on unbalanced data, which would suggest the three estimates were not independent.
How Australia resolves it: annually, by balancing
The reconciliation happens once a year, through the supply-use tables. These trace each product from its supply, whether produced domestically or imported, to its use, whether consumed, invested, exported or absorbed as an intermediate input. Forcing supply and use to match for every product is what balances the accounts.
The result is stronger than a reduction in the gap. The Bureau of Statistics states that for years derived from balanced supply-use tables, the annual estimates using the income, expenditure and production approaches are identical, and that there is no statistical discrepancy for annual estimates from 1994-95 up to the year prior to the latest complete financial year, in either current price or volume terms.
So for the balanced span the three approaches do not merely approximately agree. They agree exactly, by construction.
Three periods sit outside that. The latest year, which has not yet been through the supply-use system. Years before 1994-95, which predate the tables. And every quarterly estimate, because balancing is an annual exercise.
How Australia resolves it quarterly: by averaging
Quarterly the gap cannot be balanced away, so it is handled differently. The Bureau compiles quarterly GDP in chain volume terms using all three approaches, and the headline measure is a simple average of the three separate measures - labelled GDP(A), with the A denoting average.
Averaging is a deliberate refusal to choose. No single approach is treated as the best estimate, so none of the three is privileged, and any one of them being unusually noisy in a quarter is damped rather than transmitted whole. The Bureau's position on attribution follows the same logic: the discrepancies across the measures offset one another at different times, so the difference cannot be pinned on any single component of GDP.
Two mechanical points follow, and both are easy to get wrong.
The averaging happens in chain volume terms, not in current prices. The current price estimate of GDP is then obtained by reflating the averaged volume estimate using the implicit price deflator derived from the expenditure-based estimates. There is no separate average of three current-price measures.
And quarterly estimates do not float free of the annual ones. For years balanced through the supply-use tables, quarterly GDP is benchmarked to annual GDP, so the four quarters of a financial year sum to the corresponding annual estimate.
What this means for reading a growth rate
The most useful consequence is that a quarterly growth rate and an annual one are different objects, not the same object at different resolutions. The quarterly figure is an average of three unreconciled estimates. The annual figure, once balanced, is a single reconciled one.
The second is that quarterly estimates for a year will change once that year is balanced. Not because they were wrong, but because they were provisional inputs to a reconciliation that had not yet happened.
The third is a caution about component detail. Because the headline is an average and the components come from individual approaches, arithmetic performed on published components will not necessarily reproduce the headline. The parts belong to three different accounts; the total belongs to none of them.
Where to look next
The current published growth figure and its history are on the GDP page, and the GDP entry defines the measure. The unemployment rate and inflation rate are the two series most often read alongside it, and the cash rate is the policy rate set partly against them.
Sources
- Australian Bureau of Statistics, Australian System of National Accounts methodology - the definitions of the income, expenditure and production approaches, the identical annual estimates under supply-use balancing, and the periods in which a discrepancy survives
- Australian Bureau of Statistics, National Income, Expenditure and Product methodology - quarterly compilation by all three approaches, the simple average and the GDP(A) label, the derivation of the current price estimate by reflation, and quarterly benchmarking to annual
- Australian Bureau of Statistics, Australian System of National Accounts, latest release - the annual balancing, and a year published with a discrepancy because its supply-use tables are not yet constructed
- Australian Bureau of Statistics, Quality of the national accounts - revisions - the benchmarking process and the revision windows that follow from it
Not financial advice, and not a recommendation of any product or institution. Reviewed by the Tradergy editorial team, September 2026.
Common questions
What are the three approaches to measuring GDP?
Production, which sums gross value added by industry at basic prices plus taxes less subsidies on products; expenditure, which sums final expenditures, inventory changes and exports less imports; and income, which sums compensation of employees, gross operating surplus, gross mixed income and taxes less subsidies on production and imports.
Which one is the published Australian figure?
None of them individually. The Bureau of Statistics compiles all three in chain volume terms and the headline measure is a simple average of the three, labelled GDP(A), with the A denoting average.
What is the statistical discrepancy?
The difference between an individual approach's estimate and the single published GDP figure. It exists because the three approaches are built from different source data that does not perfectly reconcile.
Is there always a statistical discrepancy?
No. For years balanced through the supply-use tables, the Bureau states there is no statistical discrepancy for annual estimates from 1994-95 up to the year before the latest complete financial year, in either current price or volume terms. It survives for the latest year, for years before 1994-95, and quarterly.
Does the discrepancy mean one measure is wrong?
It is not attributed to any single component. The discrepancies across the three measures offset one another at different times, so the difference cannot be traced to one part of the accounts being at fault.