{
  "id": "price-elasticity-of-demand",
  "title": "Price Elasticity of Demand",
  "lang": "en",
  "source_lang": "en",
  "summary": "Price elasticity of demand measures how strongly quantity demanded responds to a change in a good’s own price, holding other influences constant.",
  "category": "economics",
  "subcategory": "microeconomics",
  "type": "concept",
  "aliases": [
    "Own-price elasticity of demand",
    "Demand price elasticity",
    "PED"
  ],
  "generated_at": "2026-11-03",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/price-elasticity-of-demand/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/price-elasticity-of-demand/",
      "json": "https://aiwiki.page/zh/price-elasticity-of-demand.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "Price elasticity of demand measures the percentage change in the quantity demanded of a good or service associated with a 1 percent change in its own price, holding other relevant influences constant. It is a central concept in microeconomics because it distinguishes a strong purchasing response from a weak one without depending on the units used to measure prices or quantities. The concept applies to individual consumers, particular products, firms, and entire markets, although elasticities measured at these different levels need not be similar.\n\nDemand is described as elastic when quantity responds more than proportionately to price, inelastic when it responds less than proportionately, and unit elastic when the proportional responses are equal. These distinctions help explain how price changes affect business revenue, how taxes alter purchases, and why consumers adjust differently across products and over time. Elasticity is not normally an unchanging property of a commodity: it depends on the price level, available alternatives, the population studied, the adjustment period, and the circumstances generating the price change."
    },
    {
      "heading": "Definition, sign conventions, and measurement",
      "text": "Let the quantity demanded be represented by \\(Q=D(P,Y,\\mathbf{P}_{-},\\mathbf{Z})\\), where \\(P\\) is the good’s own price, \\(Y\\) is consumer income, \\(\\mathbf{P}_{-}\\) represents other prices, and \\(\\mathbf{Z}\\) contains additional influences such as tastes and demographic characteristics. The point own-price elasticity is\n\n\\[\\varepsilon_D = \\frac{\\partial Q}{\\partial P}\\frac{P}{Q} = \\frac{\\partial\\ln Q}{\\partial\\ln P}.\\]\n\nThe partial derivative holds the other arguments of demand constant. This qualification is essential: a simultaneous change in income or product quality cannot simply be attributed to the price change. Price elasticity of demand is one application of the broader concept of price elasticity, which also applies to supply and other price-responsive quantities.\n\nFor an ordinary downward-sloping demand curve, \\(\\varepsilon_D\\) is negative. A value of \\(-0.4\\) means that a sufficiently small 1 percent price increase reduces quantity demanded by approximately 0.4 percent. Many textbooks and business reports instead state the positive magnitude \\(e=|\\varepsilon_D|\\). Under this convention, demand is elastic if \\(e>1\\), inelastic if \\(0<e<1\\), and unit elastic if \\(e=1\\). An elasticity reported as “0.4” is therefore ambiguous unless its sign convention is specified.\n\nPoint elasticity describes responsiveness at a particular point. When only two observations are available, a common finite-change measure is the midpoint, or arc, elasticity:\n\n\\[\\varepsilon_{\\mathrm{arc}} = \\frac{(Q_2-Q_1)/[(Q_1+Q_2)/2]} {(P_2-P_1)/[(P_1+P_2)/2]}.\\]\n\nThe midpoint denominators make the calculation independent of which observation is treated as the starting point. If price rises from 10 to 12 and quantity falls from 100 to 90, the midpoint quantity change is approximately \\(-10.53\\) percent and the midpoint price change is approximately \\(18.18\\) percent. Arc elasticity is consequently about \\(-0.58\\). A logarithmic arc measure, \\(\\ln(Q_2/Q_1)/\\ln(P_2/P_1)\\), is another symmetric approach for positive prices and quantities.\n\nElasticity is dimensionless, unlike the slope \\(\\partial Q/\\partial P\\). Changing measurement from liters to gallons alters the numerical slope but not elasticity. A vertical demand curve represents zero price responsiveness over the relevant range; a horizontal curve represents the limiting case of perfectly elastic demand. These are useful theoretical benchmarks rather than typical descriptions of entire markets. Ordinary percentage and logarithmic measures also require care at zero prices or quantities, where their denominators or logarithms are undefined."
    },
    {
      "heading": "Substitution, expenditure shares, and adjustment over time",
      "text": "The availability of substitutes is a principal determinant of price responsiveness. When closely comparable alternatives are readily available, a small price increase can induce substantial switching. A narrowly defined brand of bottled water will generally face more substitution opportunities than bottled water as a category, which in turn has more alternatives than drinking water broadly defined. Elasticity therefore depends on how the product and the relevant market are delineated. A product does not have one elasticity that remains valid under every aggregation.\n\nSubstitution depends on more than physical similarity. Search costs, compatibility requirements, contracts, habits, and switching costs can prevent consumers from moving to an otherwise attractive alternative. A software user may remain with an existing application because changing requires retraining or converting files. A commuter’s response to a rail fare increase depends on bus routes, car ownership, parking charges, and journey times. The relevant comparison is the full cost and usefulness of available choices, not merely their posted prices.\n\nThe share of expenditure devoted to a good also matters. A substantial proportional increase in the price of an inexpensive, infrequently purchased item may have little effect on a household’s budget. The same proportional increase in rent or fuel can require broader expenditure adjustments. Nevertheless, a large budget share does not mechanically imply elastic demand: housing may absorb a large share of income while offering few immediate substitutes. Likewise, describing a product as a necessity or luxury provides context but does not establish a numerical elasticity.\n\nThe distinction between the short run and long run is especially important when purchases depend on durable equipment or infrastructure. After a gasoline price increase, drivers can initially combine journeys or reduce discretionary travel. Over several years, they can buy more fuel-efficient vehicles, change residence, or alter commuting arrangements. These additional margins often make long-run fuel demand more elastic than short-run demand. Electricity consumers similarly face different adjustment possibilities before and after replacing heating systems, appliances, or insulation.\n\nLong-run elasticity is not universally larger in magnitude. A temporary discount on a durable good can cause households to bring forward purchases they would otherwise make later. Sales may respond strongly during the discount period but much less when measured over several years. Stockpiling creates a related distinction between purchases and actual consumption: a supermarket promotion can sharply increase sales without producing an equally large increase in use. Expectations, inventory capacity, and the timing of replacement therefore help determine what a measured elasticity represents. Any estimate needs a defined horizon and a clear account of whether it concerns orders, purchases, usage, or consumption."
    },
    {
      "heading": "Consumer theory and compensated demand",
      "text": "In consumer theory, ordinary demand results from choosing a preferred bundle subject to a budget constraint. A price change alters both the relative attractiveness of goods and the purchasing power of a given money income. These are analytically distinct channels. The substitution effect reflects movement toward relatively cheaper goods, while the income effect reflects the change in attainable consumption caused by the price change. Their combination determines the ordinary, or Marshallian, price response.\n\nCompensated demand, also called Hicksian demand, isolates the substitution response by adjusting expenditure so that the consumer remains at the original utility level. It does not mean that compensation is actually paid. It is a theoretical comparison used to separate changes in relative prices from changes in real purchasing power. Under standard assumptions about preferences and optimizing behavior, a good’s compensated own-price response is nonpositive.\n\nThe Slutsky equation connects compensated and ordinary demand. For good \\(i\\), it can be written in elasticity form as\n\n\\[\\varepsilon_{ii}^{M} = \\varepsilon_{ii}^{H}-s_i\\eta_i,\\]\n\nwhere \\(\\varepsilon_{ii}^{M}\\) is the Marshallian own-price elasticity, \\(\\varepsilon_{ii}^{H}\\) is the compensated own-price elasticity, \\(s_i=p_iq_i/Y\\) is the good’s expenditure share, and \\(\\eta_i\\) is its income elasticity of demand. The equation explains why budget shares matter: the income-effect contribution depends jointly on the share and income responsiveness. Holding nominal income constant is therefore not equivalent to holding utility constant.\n\nFor a normal good, \\(\\eta_i>0\\). A price increase reduces real purchasing power, and this income effect reinforces the substitution-induced reduction in demand. For an inferior good, \\(\\eta_i<0\\), the income effect works in the opposite direction. Most inferior goods still have downward-sloping ordinary demand because substitution dominates. A Giffen good is the exceptional case in which the opposing income effect is sufficiently large to make ordinary quantity demanded rise with its own price. This does not contradict the nonpositive compensated response.\n\nDemand theory also places restrictions on a system of elasticities. If all prices and income rise in the same proportion, a consumer’s feasible consumption bundles remain unchanged; ordinary demand is homogeneous of degree zero in prices and income. For good \\(i\\), this implies\n\n\\[\\sum_j \\varepsilon_{ij}^{M}+\\eta_i=0,\\]\n\nwhere the sum includes its own-price and cross-price elasticities. Compensated cross-price effects also satisfy symmetry restrictions when expressed in derivative form. These relationships matter in empirical work: estimating every elasticity independently can produce a collection of numbers inconsistent with any coherent consumer demand system."
    },
    {
      "heading": "Demand-curve shape and total revenue",
      "text": "Elasticity and slope describe different properties of demand. For linear demand,\n\n\\[Q=a-bP,\\qquad a,b>0,\\]\n\nthe slope is the constant \\(-b\\), but elasticity is\n\n\\[\\varepsilon_D=-\\frac{bP}{a-bP}.\\]\n\nDemand becomes increasingly elastic as price rises and quantity approaches zero. At the midpoint of the linear curve, elasticity equals \\(-1\\). On the high-price, low-quantity portion it is elastic; on the low-price, high-quantity portion it is inelastic. Consequently, a straight demand curve is not a constant-elasticity curve, even though it has a constant slope.\n\nA constant-elasticity specification instead takes the form\n\n\\[Q=AP^{-\\alpha},\\qquad A>0,\\quad \\alpha>0,\\]\n\nwith \\(\\varepsilon_D=-\\alpha\\) everywhere. Taking logarithms produces \\(\\ln Q=\\ln A-\\alpha\\ln P\\), which explains the popularity of log–log equations in empirical research. The convenience comes with restrictions: one parameter must describe responsiveness throughout the modeled price range. If \\(\\alpha=1\\), the curve is a rectangular hyperbola and expenditure on the good remains constant. Such specifications may fit a limited observed range while behaving implausibly near zero prices or at extremely high prices.\n\nThe relationship with total revenue follows from \\(R(P)=PQ(P)\\):\n\n\\[\\frac{dR}{dP} = Q+P\\frac{dQ}{dP} = Q(1+\\varepsilon_D).\\]\n\nWhen demand is inelastic, a small price increase raises revenue because the proportional quantity loss is smaller than the proportional price gain. When demand is elastic, it lowers revenue. At unit elasticity, revenue is locally stationary. For the linear demand curve, revenue reaches its maximum at the midpoint. This relationship concerns revenue, not profit: production costs can change when output changes.\n\nThe derivative result is local and does not justify multiplying a large price change by a single point elasticity. Under constant elasticity, the exact finite-change relationship is\n\n\\[\\frac{Q_2}{Q_1} = \\left(\\frac{P_2}{P_1}\\right)^{-\\alpha}.\\]\n\nIf \\(\\alpha=2\\), a 10 percent price increase reduces quantity by about 17.4 percent, not exactly 20 percent. Revenue falls by about 9.1 percent. The percentage approximation becomes more accurate as the price change becomes smaller.\n\nObserved revenue movements alone cannot establish elasticity. Revenue may rise because demand shifted outward, because quality improved, or because the mix of products changed. The revenue test applies to movement along a given demand relationship, with other determinants held fixed. It is also sensitive to measurement: gross customer expenditure, a retailer’s receipts net of tax, and a platform’s commission revenue are different quantities and need not respond identically."
    },
    {
      "heading": "Firm pricing, market power, and product differentiation",
      "text": "A firm uses the elasticity of the demand it faces, which may differ greatly from market demand. Even when total demand for a product category is inelastic, a single supplier can lose many customers if competitors offer close alternatives. In the ideal model of perfect competition, an individual firm faces a horizontal demand curve at the market price despite a downward-sloping industry demand curve. A monopolist serving the entire market instead faces market demand, subject to the market’s definition and the availability of outside substitutes.\n\nFor a single-product firm choosing price or quantity to maximize profit, with differentiable demand and an interior optimum, marginal revenue satisfies\n\n\\[MR=P\\left(1+\\frac{1}{\\varepsilon_D}\\right).\\]\n\nEquating marginal revenue to marginal cost, \\(MC\\), gives the Lerner pricing condition:\n\n\\[\\frac{P-MC}{P} = -\\frac{1}{\\varepsilon_D} = \\frac{1}{|\\varepsilon_D|}.\\]\n\nUnder these assumptions, less elastic demand permits a larger proportional markup. If marginal cost is positive, the ordinary interior optimum lies on the elastic portion of demand. On an inelastic portion, reducing output raises revenue while also reducing variable production costs, so that point cannot be the optimum in the standard unconstrained model.\n\nThe condition is not a universal rule for all observed prices. Capacity constraints, regulation, uncertainty, discrete price choices, and dynamic investment can alter the relevant optimization problem. A measured markup is also not an elasticity estimate unless the maintained model is appropriate. In oligopoly, the demand relevant to a firm depends on rival products and strategic responses. A short-term unilateral price experiment measures a different response from an industry-wide price increase or a long-run change after competitors adjust.\n\nMultiproduct firms must consider cross-price effects. Raising one product’s price may divert customers to another product owned by the same company, partially offsetting the initial loss. Alternatively, reducing the price of a device can increase sales of complementary accessories or services. Own-price elasticity alone therefore cannot determine the profit-maximizing prices of an interconnected product portfolio. Product quality, advertising, and availability may also be chosen jointly with price.\n\nUnder price discrimination, a seller may charge different prices to identifiable groups or offer menus that induce customers to select different options. In the simplest separated-market model with common marginal cost, the group with less elastic demand receives the higher optimal price. This result requires effective segmentation and limited resale between groups. In two-sided markets, such as some payment and advertising platforms, prices on one side also affect participation on the other. A low or zero price to users cannot be interpreted using the single-product markup formula alone."
    },
    {
      "heading": "Estimation and causal identification",
      "text": "Estimating demand elasticity requires separating the effect of price from the other reasons that purchases change. A basic empirical specification is\n\n\\[\\ln Q_{it} = \\alpha+\\beta\\ln P_{it} +\\boldsymbol{\\gamma}^{\\prime}\\mathbf{X}_{it} +u_{it},\\]\n\nwhere \\(i\\) identifies a product, household, or market, \\(t\\) identifies time, and \\(\\mathbf{X}_{it}\\) contains controls. In a correctly specified constant-elasticity model, \\(\\beta\\) is the own-price elasticity. However, fitting a regression does not by itself establish causation. Prices are often chosen in response to expected sales, local competition, or product characteristics that are incompletely observed.\n\nThe classic problem arises because observed prices and quantities are joint outcomes of supply and demand. Strong demand can raise both market price and sales. A positive correlation between the two may consequently appear even when every underlying demand curve slopes downward. Conversely, promotions can coincide with advertising, favorable shelf placement, or changes in inventory. Attributing the entire sales increase to the lower price would overstate price responsiveness if these other changes also increased purchases.\n\nInstrumental variables provide one identification strategy. An instrument must predict price while affecting quantity only through the relevant price channel, conditional on the model’s controls. Input-cost changes, transportation costs, or tax changes sometimes provide useful variation, but none is automatically valid. For example, weather may change production costs while simultaneously changing consumer preferences. The exclusion restriction must be defended in the particular setting, and weak instruments can produce unstable estimates and misleading inference.\n\nRandomized pricing experiments can make price changes independent of demand conditions by design. Their interpretation still depends on implementation. Customers may share information, move between experimental locations, postpone purchases, or respond to the novelty of a discount. An experiment may identify the immediate effect on sales without identifying long-run consumption. Quasi-experimental designs, including difference-in-differences and regression discontinuity, can exploit policy changes or eligibility thresholds, but require their own assumptions about comparison groups and other changes occurring at the same time.\n\nData construction is equally consequential. Recorded unit values—expenditure divided by quantity—may change because consumers select higher-quality items rather than because the price of an unchanged product moved. Stockouts make observed sales smaller than unconstrained demand. Aggregation can conceal household differences, and zero purchases complicate logarithmic models. Estimates therefore need a clearly defined price, quantity, population, period, and uncertainty measure.\n\nStructural methods estimate a system of consumer choices rather than a single equation. Discrete-choice models are particularly useful for differentiated products such as cars, where consumers select among models and an outside option. Their estimated own-price and cross-price elasticities can support simulations of new products or mergers. Such exercises depend on assumptions about preferences, available choices, and firm behavior; an elasticity precisely estimated within one model can still be sensitive to alternative specifications."
    },
    {
      "heading": "Empirical patterns across goods and services",
      "text": "Empirical elasticities differ across countries, income groups, periods, and methods. A familiar pattern is that broad categories with few immediate substitutes have relatively inelastic short-run demand, while narrowly defined products face stronger substitution. Yet broad labels such as “food,” “energy,” or “transport” conceal substantial heterogeneity. The elasticity of all food consumption is not the elasticity of beef, and neither is the elasticity of one supermarket’s beef sales. Published estimates are most informative when compared at similar levels of aggregation and over similar adjustment periods.\n\nGasoline illustrates both limited immediate adjustment and the importance of historical context. In a 2008 study, Jonathan Hughes, Christopher Knittel, and Daniel Sperling compared United States gasoline demand in 1975–1980 and 2001–2006. Their reported short-run price-elasticity estimates were approximately \\(-0.21\\) to \\(-0.34\\) for the earlier period and \\(-0.034\\) to \\(-0.077\\) for the later period. These results describe particular samples and specifications rather than permanent properties of gasoline. Vehicle ownership, settlement patterns, incomes, and the price variation available for estimation all affect such comparisons.\n\nElectricity demand highlights the distinction between usage and equipment choices. Within a day, households may have limited ability to replace electric heating or essential appliances, although they can adjust thermostat settings or shift some activities. Over longer periods, changes in insulation, heating systems, and appliance efficiency expand the response. Under time-varying tariffs, another elasticity concerns substitution between peak and off-peak consumption. A substantial shift away from peak hours can occur even when total daily electricity use changes relatively little.\n\nFood demand demonstrates the role of income and the level of aggregation. Consumers can respond to a price increase by switching brands, retailers, package sizes, or ingredients before reducing total food intake. Lower-income households may face tighter expenditure constraints, but this does not imply a universal ranking of own-price elasticities across income groups. Available substitutes, subsistence requirements, and the expenditure share of the particular food all matter. Changes in quality also complicate measurement when households substitute cheaper varieties within a category.\n\nTransport and digital products introduce further measurement issues. Travel demand responds to fares alongside waiting time, reliability, and access to alternatives, so elasticity with respect to the money fare need not capture responsiveness to the full journey cost. For a zero-priced digital service, conventional percentage price elasticity is undefined at the initial price, although introducing a fee can be studied through participation changes or a choice model. Subscription services require distinctions among sign-ups, renewals, cancellations, and intensity of use. These are separate outcomes, and an elasticity estimated for one cannot automatically be used for another."
    },
    {
      "heading": "Tax incidence, welfare, and public policy",
      "text": "Demand elasticity helps determine how a tax changes market prices and quantities. In the standard competitive model, a small per-unit tax creates a gap between the price buyers pay and the price sellers receive. If \\(\\varepsilon_S>0\\) is supply elasticity and \\(e_D=|\\varepsilon_D|\\), the buyer’s share of the marginal tax burden is\n\n\\[\\frac{dP_b}{dt} = \\frac{\\varepsilon_S}{\\varepsilon_S+e_D}.\\]\n\nThe seller’s share is \\(e_D/(\\varepsilon_S+e_D)\\). The relatively less elastic side of the market bears more of the burden because it has fewer opportunities to adjust. This is the central elasticity result in tax incidence.\n\nFor example, if supply elasticity is 1 and the magnitude of demand elasticity is 0.25, buyers bear 80 percent of a small marginal tax in the simplified model. The result does not depend on whether the law requires buyers or sellers to remit the payment. Statutory liability and economic incidence are different concepts. The formula assumes competitive behavior and local adjustment; market power, existing taxes, cross-border purchases, and nonstandard supply responses can produce different pass-through patterns.\n\nElasticity also affects the quantity distortion associated with taxation. For a small per-unit tax introduced into an otherwise undistorted competitive market, a familiar approximation to deadweight loss is\n\n\\[DWL\\approx\\frac{1}{2}t|\\Delta Q|.\\]\n\nA smaller quantity response produces a smaller efficiency loss for a given tax wedge, other things equal. Accurate welfare analysis may require compensated rather than ordinary demand, especially when income effects are substantial. The transfer of revenue to government is not itself deadweight loss; the loss arises from mutually beneficial trades that no longer occur.\n\nThis reasoning contributes to Ramsey taxation and pricing. In a simplified setting with independent demands, no distributional weights, and a revenue requirement, relatively inelastic goods can bear higher proportional tax wedges while limiting aggregate distortion. The full theory incorporates cross-price effects and compensated responses rather than merely ranking isolated own-price elasticities. Distributional objectives also matter: an inelastic necessity may account for a large share of low-income households’ spending. An efficiency result alone does not determine a complete tax policy.\n\nWhere consumption or production creates an externality, reduced purchases can produce benefits not captured by private demand. A Pigouvian tax is conceptually related to marginal external damage rather than selected solely from the elasticity of demand. Elasticity then helps predict the behavioral response and revenue. Similar distinctions apply to subsidies and price controls: a low measured response can affect quantities and fiscal costs, but does not by itself establish whether an intervention improves welfare. Short-run estimates may also understate responses after durable equipment, production methods, or settlement patterns change."
    },
    {
      "heading": "Historical development and interpretation",
      "text": "Nineteenth-century economists developed demand analysis before the modern statistical estimation of elasticities. Antoine Augustin Cournot’s Researches into the Mathematical Principles of the Theory of Wealth appeared in 1838 and analyzed price-dependent demand and monopoly revenue mathematically. Jules Dupuit’s 1844 work on the utility of public works examined willingness to pay and the benefits of infrastructure. These contributions helped establish the analytical relationship between prices, quantities, revenue, and consumer benefits.\n\nAlfred Marshall gave elasticity a prominent place in Principles of Economics, first published in 1890. His treatment emphasized proportional responsiveness rather than the raw slope of demand and connected it to practical questions about substitutes, expenditure, and adjustment over time. The terminology made it possible to compare markets measured in different physical units. Subsequent textbook presentations standardized the distinctions among elastic, inelastic, and unit-elastic demand, as well as the connection between elasticity and revenue.\n\nTwentieth-century consumer theory clarified what must be held constant when interpreting a price response. Eugen Slutsky’s 1915 analysis separated substitution and income effects. John Hicks and R. G. D. Allen’s work in 1934 further developed the ordinal analysis of consumer choice. These developments established that a single observed price response combines distinct mechanisms, and that welfare comparisons often require compensated responses rather than changes measured at constant money income.\n\nEmpirical work progressed from single equations toward jointly estimated demand systems and differentiated-product models. Richard Stone’s 1954 work on the linear expenditure system provided an important framework for analyzing household spending. Angus Deaton and John Muellbauer introduced the Almost Ideal Demand System in 1980, combining flexible expenditure relationships with restrictions from consumer theory. Steven Berry, James Levinsohn, and Ariel Pakes’s 1995 study of automobile demand and supply became influential in estimating differentiated-product demand while addressing price endogeneity and heterogeneous preferences.\n\nModern interpretations also consider attention, expectations, and the presentation of prices. Behavioral economics examines why consumers may respond differently to equally costly changes framed as fees, discounts, or future payments. An elasticity based on a conspicuous promotional price may therefore differ from one based on a less visible billing adjustment. Such findings do not eliminate the elasticity concept; they make the conditions attached to its measurement more important. A reported coefficient is interpretable only with its product definition, price measure, population, adjustment horizon, and identification strategy. A response observed for a small temporary change cannot automatically describe a large permanent reform, especially when alternatives, product quality, or market structure also change."
    }
  ],
  "links": [
    "microeconomics",
    "price-elasticity",
    "demand-curve",
    "substitute-and-complementary-goods",
    "income-effect",
    "giffen-good",
    "total-revenue",
    "perfect-competition",
    "marginal-revenue",
    "marginal-cost",
    "oligopoly",
    "price-discrimination",
    "causation",
    "supply-and-demand",
    "instrumental-variables",
    "difference-in-differences",
    "regression-discontinuity-design",
    "tax-incidence",
    "deadweight-loss",
    "externality",
    "price-controls",
    "alfred-marshall",
    "behavioral-economics"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}