Gross Domestic Product (GDP) and the balancing of the annual accounts
GDP can be estimated using three approaches:
the income approach,
the expenditure approach,
the production approach.
In theory the three different approaches should produce the same result. However, as they are based on different surveys and administrative data sources, they all produce results which, like all statistical estimates, are subject to errors and omissions. A definitive GDP estimate can only emerge after a process of balancing and adjustment. The Office for National Statistics (ONS) believes the most reliable estimate of the current price level of GDP is derived using the annual Supply and Use balancing framework. Accordingly, for the years when these estimates are available, GDP is set at the level derived from that year’s balance. For periods subsequent to the latest Supply and Use estimates, the level of GDP is carried forward using movements in income, expenditure and production totals.
The Supply and Use framework
The United Kingdom (UK) Supply and Use tables show the:
composition and value of goods and services entering into final demand,
outputs and incomes generated in the production process,
intermediate transactions which form inputs into these processes.
The analyses are constructed to show a balanced and complete picture of the flows of products in the economy and illustrate the relationships between producers and consumers of goods and services. On an annual basis, Supply and Use tables are used to achieve consistency in the economic accounts’ aggregates by linking the components of value added, inputs, outputs and final demand.
As the income, production and expenditure measures of GDP can all be calculated from the Supply and Use tables, a single estimate of GDP can be derived by balancing the supply and demand for goods and services and reconciling them with the corresponding value added estimates.
The annual Supply and Use balancing process produces a first balance for a given year approximately 18 months after the year end. Both full and summary Supply and Use tables are published as a separate web-only publication at the same time as the Blue Book.
Supply and Use tables
The Supply and Use tables are based on a framework incorporating estimates of:
The tables consist of two matrices:
the Supply table,
the Use table.
Both the Supply table and the Use table break down and balance 112 different industry groups at basic prices and 112 product groups at purchasers’ prices. However, for publication purposes, some adjacent industry and product groups may be combined for quality purposes.
At the very aggregate level the Supply table is represented in table 1.
Table 1, Supply table
Imports of goods
Taxes less subsidies
|Output by product|
The main part of the Supply table shows estimates of domestic industries’ output (total sales adjusted for changes in inventories of work in progress and finished goods) compiled at basic prices. Basic prices value the goods leaving the factory gate, but exclude any taxes on products and include subsidies on products. However, for the balancing process, the estimates of supply of products are required at purchasers’ prices (those actually paid by the purchasers to take delivery of the goods, excluding any deductible Value Added Tax). Estimates of domestic output valued at basic prices can be converted to the total supply of products valued at purchasers’ prices:
domestic output valued at basic prices
the value of imports of goods and services
distributors’ trading margins
taxes on products (for example, VAT)
subsidies on products (for example, agricultural and transport subsidies)
total supply of products valued at purchasers’ prices
The Use table reveals:
the input structure of each industry (in terms of combined domestic and imported goods and services),
the product composition of final demand,
for each industry the intermediate purchases adjusted for changes in inventories of materials and fuels.
The intermediate and final demand for products is represented in the rows of the balance, while intermediate consumption by industries, their demand for primary inputs (compensation of employees plus gross operating surplus and mixed income) and their outputs are represented in the columns. At the very aggregate level, the Use table can be considered in three parts.
Table 2, Use table
|Industry consumption||Final demand|
|Products consumed||Shows intermediate consumption of products by each industry in the production of their own output||Shows final demand categories (for example, households' expenditure, gross capital formation and exports) and the values of products going to these categories|
|Primary inputs||Shows the gross value added components of each industry, taxes less subsidies on production other than product specific taxes and subsidies, compensation of employees and gross operating surplus|
The body of the matrix, which represents consumption of products, is at purchasers’ prices and therefore already includes the product-specific taxes and subsidies separately in the Supply table.
The Supply-Use balance is effectively achieved when:
Inputs (from the Use table)
Outputs (from the Supply table).
Supply (from the Supply table)
Demand (from the Use table).
That is, when the data from the income, expenditure and production approaches used to fill the matrices all produce the same estimate of current price GDP at market prices. GDP at current market prices can be derived from the balances by taking the estimate of total gross value added at basic prices (from the Use table) and adding taxes on products and deducting subsidies on products (from the Supply table). It can also be derived by summing all the elements of final demand, plus exports (all from the use table) less imports (from the Supply table).
The balancing process
If the Supply and Use framework is considered as a column (industry) and row (product) matrix, the process of balancing consists of a series of alternating row and column confrontations of the data.
The first stage takes place before the Supply and Use framework is populated. It examines each of the individual rows and columns within the framework being reviewed for plausibility independently of each other. For example, estimates of household consumption expenditure by product are produced and analysed to ensure the overall picture of household spending and its breakdown by product presents a credible story. For components with an industry dimension, such as output, the initial stage scrutinises the data, ensuring the story for industries looks plausible. The first stage is carried out by the compilers of the original data.
The second stage is a confrontation within the framework of the rows (products) in the Supply and Use framework. The accounting relationship that supply is equal to demand is tested for each product and areas of inconsistency between the various sources are identified. Once investigated, the row data are then adjusted to achieve a balance. This adjustment process firstly reviews the quality of the data used to populate the individual cells within a row. It then applies adjustments to the components, based on their relative strengths, to achieve coherence between product supply and demand. This phase is carried out by the Supply and Use branch with assistance from data compilers.
The third stage of the balancing process is to confront the columns (industries). While the second stage of balancing equalises supply and demand by product, it does not ensure that, for each industry, the inputs to the process of production – goods and services consumed during production, plus primary inputs – equals its outputs. This third stage of balancing has the objective of confirming that this column identity is satisfied. Again, the quality of the data feeding into each cell is assessed and this information is used as the basis for consequent adjustments to bring each industry into balance. As for the second stage, this phase is carried out by the Supply and Use branch with assistance from data compilers.
Once each industry’s inputs and outputs have been balanced, there is a strong probability that the product supply and demand identities set during the second stage of the process will have become unbalanced. The second and third stages are therefore repeated until both the row and column identities are simultaneously satisfied. This iterative approach converges on an overall solution by reducing the remaining imbalances in the Supply and Use framework in each complete cycle of row and column adjustment.
Since 2012 the balancing process has also made use of a ‘Raking and Scaling’ (RaS) algorithm at the very last stage of balancing. The algorithm is used when final, credible economic pictures have been defined for all the economic components, but minor mathematical imbalances remain within the production matrix. The algorithm is superior to manual balancing at this stage as it quickly and proportionately distributes and removes these imbalances, without having any discernable impact on any one industry or product.
Throughout each stage of the process, a significant amount of knowledge of the compilation methods and quality of the basic data are used and balancers and data compilers’ area knowledge is supported by a broad and expanding range of secondary economic evidence. This evidence includes labour market and price statistics produced by the ONS, analyses from other government departments, data gathered by trade associations and observations by industry experts. During the process, the evolving balance is reviewed after each industry-product cycle to assess how the economic picture is developing and confirm that it is credible. The final balanced levels are only introduced into the National Accounts once they have been agreed with each compiler area and signed off by a panel of senior ONS statisticians and economists.