AI Glossary / Statistics and probability / Intermediate

Chain linking

Definition

Chain linking constructs an index or volume series by connecting changes measured between successive periods. Instead of comparing every period using one permanently fixed set of weights, the method can update weights for each link and multiply the resulting growth factors. It is widely used in economic statistics to reflect changes in the composition and relative importance of goods and services.

What it means

A fixed-base index can become less representative as prices, production or spending patterns change. Chain linking uses a sequence of shorter comparisons to connect periods while allowing the weighting basis to evolve. The details depend on the index formula and statistical application. A chained price index and a chained volume measure answer different questions, even though both may use linking. The reference period where the published index equals 100 is also distinct from the periods supplying the weights.

How it works

Choose an appropriate index formula and calculate each adjacent-period change using the required prices, quantities or weights. Starting from a reference value, multiply successive link factors to build the longer series. In a simple annual Laspeyres-type volume link, value the current and previous quantities at previous-period prices and take their ratio. Other formulas and quarterly linking procedures use different constructions. Preserve consistent definitions across links and document revisions. Re-referencing an already constructed series changes its scale without changing its growth rates, whereas changing the underlying weights or methodology can change the measured movements.

Three practical examples

• National accounts: construct volume measures that distinguish changes in production from price changes while updating the weighting structure.
• Retail analysis: follow sales volume as the mix of products and their relative values changes over time.
• Business dashboards: maintain a documented composite index whose component weights are updated at planned intervals, distinguishing linking from an arbitrary splice of incompatible series.

Think of it this way

Think of measuring a long journey through connected stages, using an appropriate local map for each stage. The final distance record depends on how those stages are defined and joined. Merely setting the starting display to 100 does not explain the route or weighting choices.

Worked example

Consider two products, A and B. In period 0, their prices are 1 and 10 and their quantities are 100 and 10, giving value 200. In period 1, quantities are 100 and 20. Valued at period-0 prices, they total 100 + 200 = 300, so the first volume link is 300/200 = 1.5. Suppose period-1 prices are now 2 and 10, giving period-1 current-price value 400. In period 2, quantities become 200 and 20. Valued at period-1 prices, they total 400 + 200 = 600, so the second volume link is 600/400 = 1.5. The chained volume index is therefore 100, 150 and 225. A fixed period-0-price comparison for period 2 gives (200 × 1 + 20 × 10)/200 × 100 = 200 instead. The difference arises because the second link gives A its updated relative price weight. The two 50% increases compound to 125%, not 100%. Re-referencing the chained series so period 1 equals 100 gives approximately 66.67, 100 and 150, preserving the same adjacent growth rates.

Illustrative case study

Fictional case: Harbor Retail tracks a composite sales-volume index while its product mix shifts toward a new category. Analysts document the annual weight update, calculate links on a consistent product definition and explain why the chained result differs from the old fixed-weight index. They keep price, value and volume measures separate and investigate classification changes before linking historical data. This fictional case illustrates index construction and does not claim that a particular index better measures every business objective.

Common misconceptions and mistakes

Chain linking is not simply adding percentage changes. Nor is it the same as seasonal adjustment, inflation forecasting or converting currencies. Independently chained component series generally do not add to the chained aggregate outside particular reference conditions, so summing published component levels can be misleading. Changing the reference year alone does not update economic weights. Frequent linking can introduce path dependence or chain drift in some settings, especially with substantial reversals, so more frequent updates are not automatically better. Missing or incomparable product definitions can break the meaning of a link even when the multiplication is correct.

Business use

Businesses using chained statistics should read the methodology before calculating contributions, shares or totals from published levels. Use the provider's recommended aggregation or growth-contribution methods where additivity is absent. For an internal index, specify the purpose, component universe, weight source, update frequency and revision policy. Preserve old versions when underlying data are revised. AI-generated commentary should describe what the measure actually tracks: a rise in a chained volume index is not necessarily a rise in nominal revenue or the physical count of every individual product.

People also ask

Does an index of 225 mean there are 225 units?

No. It is a relative level determined by the index construction and reference value.

Does changing the reference year change percentage growth?

Pure re-referencing rescales the same series and leaves growth rates unchanged.

Can chained components always be added together?

No. Chaining generally loses additivity, so the relevant statistical guidance must be followed.

Check your understanding

Q: Why is the worked example's chained period-2 index 225 rather than 200?

Answer: A: It multiplies two volume links of 1.5, and the second uses updated relative prices. The fixed-base calculation continues using period-0 prices for both products.

Related terms

• Index number
• fixed-base index
• Laspeyres index
• Fisher index
• chain volume measure
• reference year
• weighting and nonadditivity

Learn next

reproduce the worked example and change the period-1 price of A to see how the second link changes. Then compare price, quantity and value indices, and examine why aggregate growth contributions require more care than summing separately chained component levels.

More in Statistics and probability

• Centerpoint - A centerpoint is an experimental run with quantitative factors set at the middle of their selected low and high settings.
• Central composite design - A central composite design (CCD) is an experimental design for fitting a second-order response surface.
• CERNO test - The CERNO test, short for Coincident Extreme Ranks in Numerical Observations, is a rank-based enrichment test.
• Chauvenet's criterion - Chauvenet's criterion is a traditional rule for flagging an unusually distant observation under an assumed normal distribution.
• Chen distribution - The Chen distribution is a continuous probability model for positive lifetimes that can represent either increasing or bathtub-shaped hazard rates.
• Chi-square test - A chi-square test is a statistical hypothesis test that compares a calculated statistic with a chi-squared reference distribution.

Sources

Sources: Office for National Statistics, chain-linking methods in national accounts: https://www.ons.gov.uk/economy/nationalaccounts/uksectoraccounts/methodologies/chainlinkingmethodsusedwithintheuknationalaccounts ; Eurostat, quarterly national accounts metadata and annual-overlap linking: https://ec.europa.eu/eurostat/cache/metadata/en/ei_qna_esms.htm ; ONS, reference-period and chain-linking terminology: https://www.ons.gov.uk/economy/grossdomesticproductgdp/articles/nationalaccounts/proposedmethodologicalchangestochainlinkingforukpublicationsandinternationaltransmissions .

Previous: Chain graphical model Next: Chain-of-thought prompting