{
  "id": "price-index",
  "title": "Price Index",
  "lang": "en",
  "source_lang": "en",
  "summary": "A price index measures changes in the prices of a specified set of goods, services, or assets across time or locations.",
  "category": "economics",
  "subcategory": "macroeconomics",
  "type": "concept",
  "aliases": [
    "Price index number"
  ],
  "generated_at": "2026-11-03",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/price-index/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/price-index/",
      "json": "https://aiwiki.page/zh/price-index.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "A price index is a statistical measure that combines the prices of multiple goods, services, or assets into a single number, usually to describe how prices change over time. It makes heterogeneous observations—such as rents, electricity tariffs, food prices, and medical fees—comparable by assigning them weights and applying an aggregation formula. A reference period is commonly assigned the value 100, so that an index of 125 indicates that the measured price level is 25 percent above its reference-period level.\n\nPrice indices are central to measuring inflation, converting monetary values into measures of volume or purchasing power, and adjusting payments under contracts and public programs. They are not interchangeable: a consumer price index, a producer price index, and a broad output price index represent different transactions and answer different questions. Their interpretation depends on the population covered, the prices observed, the treatment of changing products, and the economic assumptions embodied in the formula."
    },
    {
      "heading": "Measurement scope and reference periods",
      "text": "A price index measures price change for a defined domain, not an observable price level that exists independently of statistical choices. Its domain may be household purchases, the output of domestic factories, imported merchandise, residential property, or another set of transactions. Specification of the domain determines which prices enter the measure and what those prices represent. A retail price including sales tax differs from a producer’s basic price, while a quoted list price may differ from the amount actually paid after discounts.\n\nThree reference concepts must be distinguished. The price reference period supplies the prices against which current prices are compared. The weight reference period supplies the expenditure or quantity information used to determine relative importance. The index reference period is the period assigned a numerical value, usually 100. These periods can coincide, but often do not. An index published as “2020 = 100” may use expenditure weights collected later and price comparisons linked through several intervening periods.\n\nRe-referencing changes the scale, not the underlying measured movement. If an existing index is \\(I_t\\), and period \\(r\\) is to become the reference period, the rescaled series is\n\n\\[I_t^{(r=100)}=100\\frac{I_t}{I_r}.\\]\n\nAll ratios between observations remain unchanged. Rebasing in the broader sense, however, may also involve new weights, new products, or methodological revisions. Statistical publications therefore distinguish a simple change of reference scale from a substantive update of the index.\n\nAn index level is also distinct from a percentage change. A rise from 120 to 126 is an increase of 6 index points but 5 percent. Conversely, two countries reporting index levels of 120 and 150 do not thereby have comparable price levels: each series may have its own reference period, coverage, and currency. Temporal indices describe movement relative to a reference, whereas comparisons of absolute price levels require additional information.\n\nThe distinction between prices and expenditure is equally fundamental. Household spending can rise because prices increase, quantities increase, or consumers shift toward more expensive products. A price index attempts to isolate the first component while controlling, as far as its purpose requires, for changes in quantities and product characteristics. The corresponding separation underlies measures of real gross domestic product and other volume indicators."
    },
    {
      "heading": "Aggregation formulas and weighting",
      "text": "Let \\(p_i^0\\) and \\(p_i^t\\) denote the prices of item \\(i\\) in a reference and comparison period, and let \\(q_i^0\\) and \\(q_i^t\\) denote the corresponding quantities. The price relative is \\(r_i=p_i^t/p_i^0\\). Simply averaging price levels is generally meaningless when items use different units: the numerical result would change if fuel were measured in liters rather than gallons. Price relatives avoid that problem, although the choice of weights remains decisive.\n\nThe Laspeyres price index measures the change in the cost of reference-period quantities:\n\n\\[P_L^{0t} =\\frac{\\sum_i p_i^tq_i^0}{\\sum_i p_i^0q_i^0} =\\sum_i s_i^0r_i,\\]\n\nwhere \\(s_i^0=p_i^0q_i^0/\\sum_jp_j^0q_j^0\\) is the reference-period expenditure share. The Paasche price index instead uses comparison-period quantities:\n\n\\[P_P^{0t} =\\frac{\\sum_i p_i^tq_i^t}{\\sum_i p_i^0q_i^t} =\\left(\\sum_i\\frac{s_i^t}{r_i}\\right)^{-1}.\\]\n\nBoth expressions are price ratios; multiplying by 100 gives conventional index-number notation.\n\nThese formulas can differ substantially when quantities respond to relative prices. Suppose two goods initially cost 1 each and households buy 10 units of each. Later, the first price doubles while the second is unchanged; purchases become 5 and 15 units respectively. The Laspeyres ratio is \\(30/20=1.50\\), while the Paasche ratio is \\(25/20=1.25\\). The difference reflects the quantities used to evaluate the price change, not an arithmetic error.\n\nThe Fisher price index is the geometric mean of the two:\n\n\\[P_F^{0t}=\\sqrt{P_L^{0t}P_P^{0t}}.\\]\n\nThe Törnqvist price index combines price relatives geometrically using average expenditure shares:\n\n\\[\\ln P_T^{0t} =\\sum_i\\frac{s_i^0+s_i^t}{2}\\ln r_i.\\]\n\nFisher and Törnqvist indices treat both periods symmetrically and have important economic-theoretic properties. They generally require more current expenditure information than an index based on previously established weights.\n\nOfficial indices frequently use related formulas rather than a literal textbook Laspeyres index. A Lowe index prices a fixed basket whose quantities can come from a period different from the price reference period. A Young index averages price relatives using fixed expenditure shares. The distinction matters when expenditure weights are collected well before publication. Describing every fixed-weight index as “Laspeyres” obscures differences in how weights are brought forward and how relative price movements affect their effective importance."
    },
    {
      "heading": "Economic theory and index-number tests",
      "text": "Index-number theory examines both the economic meaning of an index and the mathematical properties an aggregation formula should possess. The economic approach asks what hypothetical comparison the statistic represents. For consumers, its central construct is the cost-of-living index, developed from the expenditure needed to attain a specified level of satisfaction at different prices.\n\nLet \\(e(p,u)\\) be the minimum expenditure required to achieve utility \\(u\\) at price vector \\(p\\). A conditional cost-of-living index is\n\n\\[K(p^t,p^0;u) =\\frac{e(p^t,u)}{e(p^0,u)}.\\]\n\nThe concept derives from consumer choice theory and utility. It permits consumers to substitute among goods when relative prices change, rather than requiring them to purchase an unchanged basket. It is conditional because tastes and relevant circumstances are held fixed; changes in household composition, environmental conditions, or access to public services are not automatically price changes.\n\nUnder expenditure-minimizing behavior and unchanged preferences, a Laspeyres index provides an upper bound to the cost-of-living comparison evaluated at reference-period utility. Purchasing the old basket remains one way to attain that utility, but substitution may provide a cheaper way. A Paasche index provides a lower bound for the comparison evaluated at comparison-period utility. Because the two bounds refer to different utility levels, they do not establish a universal ordering for every dataset. Observed Laspeyres indices often exceed Paasche indices, but this is not an unconditional mathematical rule.\n\nThe axiomatic approach evaluates formulas through tests. The identity test requires no measured change when all prices remain unchanged. Proportionality requires an index to rise by a common factor when every price does so. Time reversal requires the index from period 0 to period \\(t\\) to be the reciprocal of the index measured backward. Factor reversal asks whether a price index and its corresponding quantity index multiply to the total value ratio. Fisher’s formula satisfies time reversal and factor reversal, properties emphasized in its historical development.\n\nNo formula satisfies every attractive property under all circumstances. Direct comparisons may conflict with comparisons chained through intermediate periods, and formulas with strong theoretical properties may require data unavailable at publication time. “Superlative” indices, including Fisher and Törnqvist indices, have precise approximation properties for flexible representations of economic behavior. The term does not mean that they eliminate sampling error, quality-measurement problems, or disagreements over the intended scope of a price statistic."
    },
    {
      "heading": "Statistical production and chain linking",
      "text": "Producing an official price index begins with a target population and a classification of expenditures or outputs. Consumer indices commonly organize spending into categories such as food, housing, transport, and recreation. Expenditure surveys, national accounts, business surveys, and administrative records provide weights. Sampling then selects locations, sellers, and representative products. A price collector must identify not merely “bread” or “a telephone plan,” but sufficiently precise specifications to make observations comparable across periods.\n\nAggregation usually occurs in stages. At the lowest level, detailed expenditure weights may be unavailable, so an elementary index combines sampled price observations within a relatively homogeneous group. A Jevons index uses the geometric mean of price relatives; a Dutot index compares arithmetic mean prices; and a Carli index takes an arithmetic mean of price relatives. These formulas have different properties. In particular, a ratio of mean prices is sensitive to the units and composition of the observations unless products are appropriately comparable.\n\nHigher-level aggregation combines elementary indices with expenditure or output weights. Since spending patterns change, statistical offices periodically update those weights. A chain index links successive short-period comparisons:\n\n\\[P_{\\mathrm{chain}}^{0t} =\\prod_{\\tau=1}^{t}P^{\\tau-1,\\tau}.\\]\n\nThis permits the basket to evolve without discarding the earlier history. Annual weight updates are common in major statistical systems, although the frequency and reference data differ by index.\n\nChain linking has costs as well as advantages. A chained comparison need not equal a direct comparison between the initial and final periods. Repeated changes in prices and quantities can generate chain drift, particularly in volatile data with pronounced reversals. Chain-linked volume components also generally cease to be additive away from their reference setting: individually chained consumption, investment, and other components need not sum exactly to a chained aggregate.\n\nOperational procedures address missing prices, seasonal products, seller closures, and product turnover. Temporary absence may require an estimated price movement based on related observations; permanent disappearance generally requires replacement or another explicit treatment. A missing observation is not automatically evidence of no price change. Seasonal adjustment, when published, is an additional statistical transformation intended to remove recurring calendar patterns rather than a change to the underlying concept of price.\n\nPublication also requires decisions about revisions and validation. Consumer indices are often published rapidly and may have restricted revision policies, whereas national-accounts price measures are revised as source data improve. Automated checks can flag unusually large changes, but deleting every outlier would suppress genuine economic events. Verification must distinguish recording errors from real discounts, shortages, tax changes, and abrupt price increases."
    },
    {
      "heading": "Quality change, new products, and housing",
      "text": "Price measurement requires comparisons of economically equivalent products. A laptop selling for the same amount as its predecessor may offer a faster processor, more storage, and a better display. Treating the entire transaction as an unchanged price can overstate the price of obtaining a given set of capabilities. Conversely, a smaller package sold at an unchanged price represents an increase in the price per unit, provided that the contents remain comparable.\n\nSeveral methods address replacement and quality change. Direct comparison is suitable when old and new products are effectively equivalent. Quantity adjustment handles changes in package size. Overlap pricing uses information from periods in which both models are sold, under assumptions about how their price difference reflects quality. Explicit adjustments may use estimates of the value of added characteristics or relevant production-cost information, although production cost and consumer value are not generally identical.\n\nHedonic price measurement models product prices as functions of measurable characteristics. A simplified regression is\n\n\\[\\ln p_{jt} =\\alpha+\\delta_t+\\sum_k\\beta_k x_{jkt}+\\varepsilon_{jt},\\]\n\nwhere \\(x_{jkt}\\) describes attributes of product \\(j\\), and \\(\\delta_t\\) captures period effects conditional on those attributes. Such methods have been used for computers and other differentiated products. Their results depend on specification, coverage, and the availability of relevant characteristics; unmeasured reliability, design, or service quality can remain important.\n\nNew products create a problem beyond ordinary replacement. Before a product exists, there is no observed transaction price with which to compare its introduction price. Economic theory can describe a hypothetical reservation price at which demand would have been zero, but estimating that price is difficult. Delayed inclusion of new goods may miss early price declines and benefits from expanded choice. Free digital services pose a related boundary issue: their value to users is not necessarily represented by a monetary price paid by households.\n\nHousing illustrates how the target concept determines treatment. Rent is a purchased flow of housing services. An owner-occupied dwelling is both a source of services and an asset, so its purchase price cannot automatically be treated like the price of current consumption. Rental equivalence estimates what an owner would pay to rent a comparable dwelling. Net-acquisitions approaches focus on specified purchases entering the household sector, while other systems use payments-based or user-cost concepts.\n\nThese approaches respond differently to land prices, financing conditions, depreciation, and transaction costs. A change in interest rates may directly affect a payments-based measure but not enter rental equivalence in the same way. Consequently, differences in owner-occupied housing treatment can produce meaningful divergence between otherwise similar consumer indices."
    },
    {
      "heading": "Major families of price indices",
      "text": "Consumer price indices measure price change for goods and services within a defined household-consumption domain. Coverage varies by resident population, geographic area, expenditure concept, and treatment of purchases abroad or by visitors. An expenditure-weighted aggregate gives more influence to spending categories with larger monetary outlays. It therefore need not represent the inflation experience of a household with unusually high spending on food, rent, or medical care.\n\nA producer price index measures prices received by producers for specified outputs or, in some systems, prices paid for inputs. Output indices generally aim to exclude price changes introduced solely by downstream retail distribution, although valuation conventions differ. Modern producer price programs can cover services and construction as well as mining and manufacturing. Stage-of-processing or final-demand classifications help distinguish goods sold for further production from those supplied to final users.\n\nImport and export price indices measure price movements in internationally traded products. They are used to deflate trade values and to examine terms of trade, commonly expressed as an export price index divided by an import price index. Results depend on currency denomination, transaction terms, and coverage. Customs unit values—trade value divided by recorded quantity—are useful data but are not pure price indices: changes in the mix of products within a customs category can alter unit values even when individual prices do not change.\n\nThe GDP deflator is an implicit price measure derived from nominal and volume measures of gross domestic product. In conventional index notation,\n\n\\[\\text{GDP deflator}_t =100\\frac{\\text{nominal GDP}_t}{\\text{real GDP}_t}.\\]\n\nIt covers domestically produced final output rather than a fixed household-consumption basket. Exports are included, while imports are not part of domestic production. Imported prices can nevertheless affect domestic costs and prices indirectly. Its composition changes with production and expenditure patterns, so it should not be interpreted as simply a broader consumer basket.\n\nOther indices address particular markets. Construction price indices may track the prices of completed structures, tender prices, or input costs; these are different objects. Residential property indices track asset transactions rather than the current consumption of shelter. Commodity indices combine prices of raw materials, with results strongly influenced by weighting and contract specifications.\n\nFinancial-market indices require further distinctions. A share-price index measures movements in selected securities under a weighting rule, while a total-return index also incorporates distributions such as dividends. These are price-related indices but are not measures of the general cost of living. Their scope is a portfolio of assets, not the goods and services purchased for current consumption."
    },
    {
      "heading": "Interpretation, deflation, and indexation",
      "text": "The inflation rate between two observations is the percentage change in the relevant price index:\n\n\\[\\pi_{a,t} =100\\left(\\frac{I_t}{I_a}-1\\right).\\]\n\nMonthly, quarterly, and twelve-month changes answer different questions. A twelve-month rate compares the current index with the same month a year earlier; a monthly rate captures more recent movement but is often more exposed to seasonal patterns. An annualized monthly rate compounds one month’s change as though it continued for a year:\n\n\\[\\pi_{\\mathrm{annualized}} =100\\left[\\left(\\frac{I_t}{I_{t-1}}\\right)^{12}-1\\right].\\]\n\nAnnualization is a transformation, not a forecast.\n\nBase effects explain why an annual inflation rate can change even when recent price movements are modest. A large increase twelve months earlier drops out of the comparison as the window advances. A declining inflation rate therefore does not imply declining prices. Slower positive price growth is disinflation; a sustained fall in the general price level is deflation. The index level preserves the cumulative effect of earlier changes that a current growth rate alone can obscure.\n\nPrice indices also convert nominal values into real values. If a nominal expenditure series is \\(V_t\\) and a suitable price index is expressed with reference value 100, a deflated measure is\n\n\\[V_t^{\\mathrm{real}}=\\frac{V_t}{I_t/100}.\\]\n\nThe qualifier “suitable” is essential. Deflating machinery expenditure with a household food index does not isolate machinery volume. In national accounting, separate deflation of output and intermediate inputs can be necessary to estimate real value added, especially when their prices move differently.\n\nIndexation applies an index to payments, tax parameters, or contracts. A payment fully adjusted from period \\(a\\) to period \\(t\\) becomes \\(W_t=W_a(I_t/I_a)\\). Actual arrangements may contain lags, caps, thresholds, exclusions, or partial adjustment. Tax thresholds can be indexed to limit the increase in effective taxation caused by nominal incomes rising with prices, while pensions and other transfers may be adjusted using legally specified measures.\n\nFor monetary policy, statistical authorities and central banks distinguish headline inflation from measures intended to reveal persistent movements. Core inflation may exclude selected volatile categories or use trimmed distributions of component changes. Such measures do not imply that excluded purchases are unimportant to households.\n\nContributions to aggregate inflation also differ from component inflation rates. A small category with a very large increase may contribute less than a major category with a moderate increase. Exact contribution calculations depend on the aggregation formula and linking structure; multiplying a component’s inflation rate by its expenditure weight is sometimes an approximation rather than a universal identity."
    },
    {
      "heading": "Spatial comparisons and purchasing power",
      "text": "A temporal price index compares different periods within a specified economy or market. A spatial price index compares locations at approximately the same time. The two problems share aggregation methods but differ in their practical difficulties. Comparing prices across countries requires equivalent products, consistent valuation, and weights representing different consumption patterns. Converting local prices at the market exchange rate does not remove differences in domestic purchasing power.\n\nPurchasing power parities are currency conversion rates designed to equalize purchasing power for specified expenditure aggregates. If a comparable basket costs 100 units of currency A and 200 units of currency B, its elementary parity is 2 units of B per unit of A. An economy-wide parity requires aggregation across many products and expenditure categories; it is not simply the price ratio of one familiar consumer good.\n\nThe International Comparison Program, coordinated globally by the World Bank, produces purchasing power parities and price-level measures through international statistical cooperation. Countries collect prices for specified products, and national-accounts expenditure data provide weights. Consumer goods are only part of the task: construction, equipment, government services, health, and education require specialized procedures because comparable market prices may be difficult to observe.\n\nRepresentativity and comparability can conflict. A product widely purchased in one country may be uncommon and unusually expensive in another. Highly detailed specifications improve comparability but can reduce the number of genuinely representative observations. Differences in service arrangements, package sizes, brands, and outlets complicate apparently straightforward matches. These problems are particularly important when comparing economies with very different income levels and consumption structures.\n\nMultilateral methods seek comparisons that are consistent across many countries. A collection of independent bilateral Fisher indices will not generally be transitive: the comparison between A and C need not equal the product of A-to-B and B-to-C comparisons. Methods such as Gini–Éltető–Köves–Szulc aggregation construct transitive comparisons from bilateral information. This gains system-wide consistency but means a country’s measured position can depend on the set of countries included.\n\nA price-level index is commonly formed by dividing a purchasing power parity by the corresponding exchange rate and expressing the result relative to a reference. It describes comparative prices for the covered aggregate, not currency misalignment by itself. National averages also conceal regional variation, especially in housing and locally supplied services. Extending benchmark parities through time using national price indices is useful, but the resulting estimates need not equal a fresh spatial comparison because temporal and international baskets differ."
    },
    {
      "heading": "Historical development and new data methods",
      "text": "Early price comparisons arose from practical questions about money, contracts, and changes in purchasing power. William Fleetwood’s Chronicon Preciosum, published in 1707, assembled historical price information to examine the changing value of money. In the eighteenth century, writers including Gian Rinaldo Carli and Joseph Lowe explored methods of combining prices and using price comparisons in contractual adjustment. Lowe’s 1823 work is particularly associated with the proposal to stabilize monetary obligations through a broadly defined basket.\n\nThe nineteenth century established formulas that remain central. Étienne Laspeyres presented his fixed-reference-quantity formula in 1871, and Hermann Paasche developed the comparison-period-quantity formula in 1874. Their approaches made explicit that a general price measure depends not only on observed prices but also on the quantities used to weight them. Expanding industrial economies and official statistics created growing demand for measures of wholesale and retail price movements.\n\nIn the early twentieth century, research connected formulas to systematic tests and economic behavior. Irving Fisher examined numerous formulas in The Making of Index Numbers in 1922 and favored the geometric mean of the Laspeyres and Paasche indices. Alexander Konüs’s 1924 work established a foundational expenditure-based treatment of the cost of living. Later contributions associated with Ragnar Frisch, Leo Törnqvist, and W. Erwin Diewert developed the relationships among economic theory, aggregation, and approximation.\n\nOfficial price statistics expanded alongside industrial wage bargaining, national accounting, and macroeconomic policy. The United States consumer price program developed from studies of workers’ living costs during and after the First World War. Postwar statistical systems extended coverage and formalized methods. International manuals, including the Consumer Price Index Manual: Concepts and Methods issued in 2020 through cooperation among international organizations, codify concepts while recognizing that national circumstances require different implementations.\n\nRecent developments concern the scale and structure of source data. Retail scanner data can provide transactions, quantities, and expenditures for large numbers of products. Web-collected prices expand coverage of e-commerce, but advertised prices may omit delivery charges, personalized discounts, or actual transaction volumes. Frequent product replacement and dynamic pricing make matching harder even when observations are abundant.\n\nLarge datasets have encouraged multilateral methods over rolling time windows, designed to use changing product sets without some forms of chain drift. They also create a tension between improved measurement and revision stability: extending a multilateral comparison can change earlier results unless special linking procedures are adopted. Access to commercially held data introduces questions of continuity, confidentiality, and reproducibility. Statistical offices must document how product identifiers, expenditure coverage, replacement rules, and linking procedures transform these records into a published index."
    }
  ],
  "links": [
    "inflation",
    "consumer-price-index",
    "real-gdp",
    "cost-of-living-index",
    "utility-economics",
    "national-accounts",
    "interest-rate",
    "producer-price-index",
    "terms-of-trade",
    "gdp-deflator",
    "gross-domestic-product",
    "deflation",
    "value-added",
    "monetary-policy",
    "core-inflation",
    "exchange-rate",
    "purchasing-power-parity",
    "world-bank",
    "world-war-i",
    "e-commerce"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}