Fact-checked by Grok 2 weeks ago

Quasilinear utility

In , quasilinear utility describes a preference structure where an agent's utility function is linear with respect to one commodity, known as the numeraire (often ), and potentially nonlinear with respect to others, taking the general form u(\mathbf{x}, y) = v(\mathbf{x}) + y, where \mathbf{x} represents quantities of non-numeraire , v(\mathbf{x}) is a capturing diminishing in those , and y is the numeraire good. This representation implies that indifference curves in the space of non-numeraire are vertically parallel, shifting outward uniformly with increases in the numeraire, reflecting the absence of income effects on the demand for \mathbf{x}. A key property of quasilinear utility is the independence of demand for non-numeraire goods from total or , as any income increase is fully absorbed by consumption of the numeraire without altering the between goods. This eliminates income effects in problems, simplifying the analysis of effects alone and making the Marshallian demand for \mathbf{x} coincide with the Hicksian (compensated) demand. Consequently, quasilinear preferences avoid complications from wealth distribution in equilibrium models, such as those in , where aggregate demands remain stable across income levels for the non-numeraire components. Quasilinear utility is widely applied in applied due to its straightforward demand structure, where individual demands depend solely on relative rather than , facilitating the monetary measurement of consumer surplus as the area under the . In this context, changes in can be directly quantified in numeraire units without interpersonal comparisons, making it a standard assumption for evaluating policy interventions like changes or subsidies. Beyond analysis, the framework is prevalent in and , where the linearity in money enables the use of side payments to incentivize truthful revelation of private values, as seen in Vickrey-Clarke-Groves mechanisms that achieve efficiency under quasilinear assumptions. This property ensures that agents can compensate one another via transfers without distorting allocations, supporting incentive-compatible outcomes in problems.

Definitions

Preference-based definition

In consumer theory, preferences over consumption bundles are represented by a \succsim defined on the set of bundles (x, m), where x \in \mathbb{R}^l_+ denotes a of l and m \in \mathbb{R}_+ is the numeraire, often interpreted as or a linear composite good. These preferences are assumed to be complete (every pair of bundles is comparable), transitive (if (x_1, m_1) \succsim (x_2, m_2) and (x_2, m_2) \succsim (x_3, m_3), then (x_1, m_1) \succsim (x_3, m_3)), and continuous (the upper and lower contour sets are closed), which together ensure the existence of continuous representations under standard conditions. Quasilinear preferences, with respect to the numeraire m, satisfy a specific invariance property in their indifference relation: for any bundles (x, m) and (x', m') such that (x, m) \sim (x', m'), it holds that (x, m + c) \sim (x', m' + c) for any constant c such that both adjusted bundles remain in the consumption set. Equivalently, the between any good in x and the numeraire m depends only on the quantities of the in x and not on the level of m. This structure implies that indifference sets in the (x, m)-space exhibit a separability where changes in the numeraire simply translate preferences without altering relative trade-offs among the non-numeraire . Geometrically, this manifests as parallel indifference curves in the space of one good from x and the numeraire m: all curves are identical vertical translations of a base , shifted upward or downward by amounts of m, reflecting constant slopes (marginal rates of ) across different levels of or numeraire . For multiple goods, the indifference surfaces in (x, m)-space are such that horizontal slices (fixing m) yield the same shape, with vertical shifts preserving the structure. This parallel property facilitates analysis of consumer behavior where money acts as a scaler of without distorting good-specific trade-offs. The concept of quasilinear preferences has roots in Alfred Marshall's cardinal theory of value and consumer surplus in the late 19th century, assuming constant of money, and was further developed within theory in the mid-20th century to model separability without strong assumptions, providing tractable representations for and analysis.

Utility function-based definition

In economics, quasilinear utility functions offer a concrete representation of preferences through an additive structure that is linear in one commodity, often designated as the numeraire good. The canonical functional form is U(\mathbf{x}, m) = v(\mathbf{x}) + m, where \mathbf{x} \in \mathbb{R}^n_{\geq 0} denotes a vector of n non-numeraire goods, m \geq 0 is the numeraire good (typically money or a composite commodity), and v: \mathbb{R}^n_{\geq 0} \to \mathbb{R} is a strictly increasing and concave function capturing the utility from the non-numeraire goods. This linearity in m implies constant marginal utility for the numeraire, which simplifies analysis by eliminating income effects on the demand for \mathbf{x}. The assumptions on v ensure that the represented preferences are monotone and convex: strict monotonicity follows from v being strictly increasing, while concavity of v guarantees quasiconcavity of U, yielding convex indifference sets. The domain restricts consumption to non-negative quantities, aligning with standard economic models of feasible bundles. A more general form allows for scaling and shifting in the numeraire term, U(\mathbf{x}, m) = v(\mathbf{x}) + \alpha m + \beta, where \alpha > 0 and \beta \in \mathbb{R} are constants; however, without loss of generality, the analysis often normalizes to the canonical case with \alpha = 1 and \beta = 0 by rescaling units. For instance, with a single non-numeraire good x > 0, a representative v(x) = \ln x yields the utility function U(x, m) = \ln x + m, which is commonly used to illustrate properties like parallel indifference curves in the (x, m)-plane. Utility representations are ordinal, meaning that any positive affine transformation \tilde{U} = a U + b with a > 0 and b \in \mathbb{R} preserves the underlying preferences and the quasilinear structure, as the transformed function takes the form \tilde{U}(\mathbf{x}, m) = a v(\mathbf{x}) + a m + b, retaining linearity in m. This invariance underscores the defining role of the linear numeraire term in distinguishing quasilinear utility from other separable forms.

Mathematical Properties

Functional form and equivalence

A fundamental result in consumer theory establishes the equivalence between the preference-based and utility function-based definitions of quasilinear preferences. Specifically, a preference relation \succsim on the consumption set \mathbb{R}^L_+ \times \mathbb{R}_+, where the last coordinate m represents the numeraire good (such as ), is with respect to m there exists a function U(x, m) = v(x) + m that represents it, with v: \mathbb{R}^L_+ \to \mathbb{R} being a continuous and increasing function.http://www.columbia.edu/~md3405/Choice_PHD_Consumer_Proofs_1_18.pdf The proof proceeds in two directions. First, suppose the preferences are in the sense that all indifference surfaces are parallel vertical translations of each other, meaning that for any bundles (x, m) and (x', m'), (x, m) \sim (x', m') m - m' = \delta(x, x') for some \delta independent of the absolute levels of m and m'. To construct the representing , define v(x) = \inf \{ t \in \mathbb{R} \mid (t, 0) \succeq (0, x) \}, where the infimum exists and is finite under the assumptions of , , and strict monotonicity of \succsim. This v(x) satisfies v(x) + m = U(x, m), and it can be verified that U preserves the indifference relation because adding a fixed amount to m shifts the bundle vertically by that amount, matching the parallel translation . Strict monotonicity ensures v is strictly increasing, and of \succsim implies of v and U.http://www.columbia.edu/~md3405/Choice_PHD_Consumer_Proofs_1_18.pdf Conversely, suppose there exists a utility function U(x, m) = v(x) + m representing \succsim. Then, for any (x, m) \sim (x', m'), it follows that v(x) + m = v(x') + m', so m - m' = v(x') - v(x), which is independent of the baseline levels of m and m'. This confirms the parallel translation property of the indifference sets, establishing the quasilinearity of the preferences. The representation is unique up to affine transformations of the form v(x) + c for a constant c, as adding a constant to v can be offset by subtracting it from m without altering the ordinal ranking, but the coefficient on m is normalized to 1 by rescaling the numeraire unit if necessary.http://www.columbia.edu/~md3405/Choice_PHD_Consumer_Proofs_1_18.pdf The equivalence relies on standard conditions: the preference relation \succsim must be complete, continuous (in the topological sense on the consumption set), and strictly monotonic (more of any good is strictly preferred). These ensure the existence of a continuous representation and the well-definedness of the infimum in the construction of v. Without , the infimum may not be attained, potentially leading to non-representable preferences; without strict monotonicity, v may not be strictly increasing, violating quasilinearity.http://www.columbia.edu/~md3405/Choice_PHD_Consumer_Proofs_1_18.pdf In edge cases where v is concave but not strictly concave, the preferences remain quasilinear, but the indifference curves may exhibit flat segments in the x-direction, implying potential satiation points for the non-numeraire goods. However, local non-satiation still holds globally due to the strict monotonicity in the numeraire m, as any bundle can be improved by an arbitrarily small increase in m alone.https://nmiller.web.illinois.edu/documents/notes/notes4.pdf A key implication of the functional form is the independence of the (MRS) from the numeraire level. The MRS between a non-numeraire good x_i and the numeraire m is given by \text{MRS}_{x_i, m} = \frac{\partial U / \partial x_i}{\partial U / \partial m} = v'(x_i), where the partial derivative is evaluated at the relevant bundle, but notably, it depends only on x and not on m. To derive this, start with the definition of MRS as the slope of the indifference curve, satisfying dU = 0: \frac{\partial U}{\partial x_i} dx_i + \frac{\partial U}{\partial m} dm = 0 \implies \frac{dm}{dx_i} = -\frac{\partial U / \partial x_i}{\partial U / \partial m}. Thus, \text{MRS}_{x_i, m} = -\frac{dm}{dx_i} = \frac{\partial U / \partial x_i}{\partial U / \partial m}. Substituting U(x, m) = v(x) + m yields \partial U / \partial x_i = v'(x_i) and \partial U / \partial m = 1, confirming the result. This independence underscores the absence of income effects on the demand for non-numeraire goods.https://nmiller.web.illinois.edu/documents/notes/notes4.pdf

Indifference curves and marginal utility

In the standard representation of quasilinear preferences, the utility function takes the form U(\mathbf{x}, m) = v(\mathbf{x}) + m, where \mathbf{x} is a vector of goods, m is the numeraire (often interpreted as money), and v is an increasing, concave function. In the (x, m)-space for a single good x, the indifference curves are vertically parallel, meaning that higher utility levels correspond to uniform upward shifts of the curve without changing its shape along the x-axis. This parallelism arises because the marginal rate of substitution (MRS) between x and m depends solely on x and is independent of the level of m or total wealth. Specifically, the slope of an indifference curve is given by -\MRS = -v'(x), which remains constant along any vertical line in the space./03:_Optimal_Choice/3.02:_More_Practice_and_Understanding_Solver) The constant marginal utility of the numeraire good underpins these properties. In the canonical form, the partial derivative \frac{\partial U}{\partial m} = 1, reflecting a linear and thus constant marginal utility of money across all consumption levels. For the other goods, \frac{\partial U}{\partial x_i} = v'(x_i) for each component x_i of \mathbf{x}, assuming v is differentiable./03:_Optimal_Choice/3.02:_More_Practice_and_Understanding_Solver) This structure implies no income effects on the demand for \mathbf{x}, as additional income is entirely allocated to m without altering the optimal quantities of the other goods. Furthermore, the cross-partial derivative \frac{\partial^2 U}{\partial x_i \partial m} = 0 for all i, which ensures additive separability between \mathbf{x} and m in the utility representation./03:_Optimal_Choice/3.02:_More_Practice_and_Understanding_Solver) Geometrically, these indifference curves exhibit distinct features in the (x, m)-space. Since m = U - v(x) along a curve of constant utility U, and assuming v(0) = 0 with v increasing, the curves intersect the m-axis at m = U \geq 0 but never enter the negative m-region, reflecting non-negativity constraints on consumption. If v is concave (v''(x) < 0), the curves are convex to the origin, with their absolute slope |dm/dx| = v'(x) decreasing as x increases, making higher portions of the curve flatter relative to the x-axis./03:_Optimal_Choice/3.02:_More_Practice_and_Understanding_Solver) Quasilinear utility generalizes linear utility functions, where v(x) is itself linear, resulting in straight, parallel indifference curves with constant MRS and representing perfect substitutability. By allowing v to be nonlinear (typically concave to capture diminishing marginal utility of x), quasilinear forms introduce realism while preserving the key simplifying properties of linearity in money.

Economic Implications

Consumer demand and income effects

In consumer theory, quasilinear utility functions simplify the derivation of Marshallian demand for the non-numeraire good. Consider a utility function of the form u(x, m) = v(x) + m, where x is the quantity of the non-numeraire good, m is the numeraire good (often money), v(x) is strictly increasing and concave (v'(x) > 0, v''(x) < 0), prices are p for x and 1 for m, and total is w. The consumer maximizes utility subject to the p x + m = w, which substitutes to maximizing v(x) + w - p x. The first-order condition yields v'(x) = p, so the Marshallian demand is x(p, w) = [v']^{-1}(p), depending only on price p and independent of w. This independence implies zero income effects for the non-numeraire good x, as \partial x / \partial w = 0. Any change in wealth adjusts consumption solely through the numeraire m = w - p x(p), with expenditure on x fixed at p x(p). In contrast, general utility functions exhibit both substitution and income effects, where a price change for x affects demand through relative price adjustments (substitution) and real purchasing power changes (income), leading to normal, inferior, or behaviors. Quasilinear preferences eliminate the income effect, isolating the substitution effect and making demand more predictable. For illustration, suppose v(x) = \sqrt{x}. The first-order condition v'(x) = \frac{1}{2\sqrt{x}} = p solves to demand x(p) = \left( \frac{1}{2p} \right)^2, with expenditure on x of p x(p) = \frac{1}{4p}. The residual wealth w - \frac{1}{4p} is spent on m. This example highlights how demand for x remains unchanged regardless of wealth levels above the minimum required. The absence of income effects also simplifies the Slutsky equation, which decomposes price effects into substitution and income components: \frac{\partial x}{\partial p} = \frac{\partial h}{\partial p} - x \frac{\partial x}{\partial w}, where h(p, u) is Hicksian demand. For quasilinear utility, \frac{\partial x}{\partial w} = 0, so Marshallian demand equals Hicksian demand (x(p, w) = h(p, u)), and the total price effect is purely substitutional. This equivalence facilitates welfare analysis by aligning uncompensated and compensated measures.

Welfare analysis

In , quasilinear utility functions enable precise measurement of consumer surplus as the integral of the difference between the marginal valuation and the price along the , specifically \int_{0}^{x(p)} (v'(q) - p) \, dq, where v(x) represents the from the good and p is its price. This exact welfare metric arises because the linearity in money (the numeraire good) implies a constant of , allowing changes to be evaluated directly in monetary terms without or income effects distorting the calculation. Producer surplus complements this by representing the excess of over production costs, and under quasilinear preferences, the total surplus in a competitive equals the difference between the total valuation and at the efficient , v(x^*) - c(x^*), where x^* is the where marginal valuation equals . This formulation simplifies welfare assessment, as the absence of income effects ensures that the sum of and surpluses accurately captures net benefits without requiring interpersonal comparisons. Quasilinear preferences guarantee that competitive equilibria are Pareto efficient, as the first welfare theorem holds without complications from income redistribution, since agents' marginal rates of substitution for the numeraire are constant and equal to unity across individuals. This property facilitates efficiency analysis by aligning individual optimization with social optimality, bypassing the need to compare utilities across heterogeneous agents. However, the assumption of constant of money underlying quasilinear utility has been critiqued for its lack of realism, particularly in settings with heterogeneous agents or significant income variations, where diminishing of income would more accurately reflect behavioral responses. In general equilibrium settings, quasilinearity across all agents implies no effects, preserving the of equilibria and simplifying the aggregation of demands without concerns.

Applications

Mechanism design

In quasilinear settings, the revelation principle asserts that any social choice function implementable in equilibrium by some mechanism can also be implemented by a , mechanism where agents truthfully report their private valuations, provided individual rationality and are satisfied. This simplifies by focusing on truthful mechanisms, as indirect mechanisms can be replicated . A canonical truthful mechanism under quasilinearity is the Vickrey-Clarke-Groves (VCG) mechanism, which selects the allocation maximizing the sum of s' reported valuations net of costs and imposes payments on i according to the rule t_i = h_i(\theta_{-i}) - \sum_{j \neq i} v_j(x^*), where x^* denotes the efficient allocation based on all reports \theta, v_j is j's valuation based on reports, and h_i is an arbitrary depending only on others' reports \theta_{-i}. This structure ensures dominant-strategy , as each 's payment internalizes the imposed on others without affecting the overall . The VCG mechanism, building on foundational work in auctions and team incentives, achieves truthful revelation while respecting the linear separability of money in utilities. Quasilinear utilities enable efficient by allowing the separation of the outcome , which maximizes \sum_i v_i(x_i) - c(x), from monetary transfers that enforce and individual rationality. Side payments adjust utilities without distorting the efficient outcome, as the of money is and identical across s. For public goods provision, quasilinearity facilitates voluntary contributions through the Clarke pivot rule, a VCG variant where each pays only if their report is pivotal in changing the efficient provision level, equal to the loss in others' they cause. This ensures truth-telling leads to efficient supply levels, as s internalize their impact without free-riding on misreports. However, free-rider problems persist even under ity, as agents contribute only when pivotal, often resulting in zero payments and potential underprovision if no one is decisive; while the mitigates strategic misrepresentation compared to non- cases where transfers fail to align incentives effectively, it generally violates budget balance, requiring external subsidies. The assumption itself faces critiques for inapplicability to nonpecuniary public goods, where utilities are not additively separable in , undermining .

Auction theory

In , quasilinear utility functions play a central role in modeling bidder behavior under the independent private values (IPV) framework, where each bidder i has a private valuation \theta_i drawn independently from a common F, and their is given by U_i(\theta_i, x_i, t_i) = \theta_i x_i - t_i, with x_i \in \{0,1\} indicating allocation of the single item and t_i the payment. Bids serve as signals of the bidder's valuation \theta_i, enabling the auctioneer to extract information while ensuring , as the linear form in money simplifies analysis by eliminating wealth effects and allowing focus on value maximization. A foundational result is the revenue equivalence theorem, which states that in the IPV model with risk-neutral bidders, any auction that is efficient (allocates the item to the highest-valuation bidder), symmetric, and Bayesian incentive-compatible yields the same expected for the seller, equal to the expected value of the pivotal bidder's virtual valuation. This equivalence holds because quasilinear utilities ensure that bidders' strategies adjust payments in a way that equalizes expected surplus across mechanisms, with determined by the lowest valuation at which the item is sold in . Myerson's optimal auction maximizes seller revenue by allocating the item to the bidder with the highest virtual valuation \phi(\theta_i) = \theta_i - \frac{1 - F(\theta_i)}{f(\theta_i)}, where f is the of F, and incorporating reserve prices to exclude low-valuation bidders. This design, implementable via a modified with entry fees or personalized reserves, achieves the highest possible expected revenue under constraints, leveraging the structure to separate allocation from payment rules. For example, in a second-price sealed-bid with bidders, each submits their true valuation \theta_i as a dominant , and the winner pays the second-highest bid, ensuring efficiency and truthful revelation without strategic shading. This mechanism exemplifies , as its expected revenue matches that of a first-price where bidders shade bids downward. Extensions reveal limitations: in common value settings, where valuations are affiliated rather than independent, revenue rankings across auction formats diverge due to the , breaking equivalence. Similarly, bidder alters bidding strategies, favoring first-price auctions for revenue maximization over second-price ones, as risk-averse bidders bid more aggressively to reduce uncertainty.

References

  1. [1]
    [PDF] Chapter 4: Topics in Consumer Theory - Nolan H. Miller
    A preference relation is quasilinear if there is one commodity, called the numeraire, which shifts the indifference curves outward as consumption. 88. Page 3 ...
  2. [2]
    [PDF] Preferences and Utility - UCLA Economics
    As we will see later, quasilinear preferences have the attractive property that the consumption of x1 is independent of the agent's income (ignoring boundary ...
  3. [3]
    [PDF] When Do Market Games Have Transferable Utility?
    ' However, quasilinear utility implies that individual demands for all goods except one are independent of income. For many economic problems, this is not an ...
  4. [4]
    [PDF] Rationalizing and Curve-Fitting Demand Data with Quasilinear Utilities
    Jul 2, 2004 · If the consumer's choices are inconsistent with maximizing a quasilin- ear utility function over her budget set, then we compute the “best”.
  5. [5]
    [PDF] EC9D3 Advanced Microeconomics, Part I: Lecture 1
    Indifference is invariant to adding quantities to a particular good. ... In terms of the utility function, preferences are quasilinear if and only if ...
  6. [6]
  7. [7]
    [PDF] The quasilinear quadratic utility model: An overview - HAL
    Oct 17, 2019 · Fact 2 (Quasilinear Quadratic Utility). At least from the mid 1960s, Levitan and Shubik founded their linear demand system on a quasilinear ...
  8. [8]
    [PDF] Intermediate Microeconomics 8th Edition: A Modern Approach
    ... Varian. ISBN 978-0-393-93424-3. W. W. Norton & Company, Inc., 500 Fifth Avenue ... Preferences. Consumer Preferences 34. Assumptions about Preferences 35.
  9. [9]
    [PDF] 千葉大学 経済研究 第12卷第3号(1997年12月) - OPAC
    Recalling that an affine function takes the form f(x) = a+bx, choose to be an affine function and define the quasilinear utility. µ(u(x,y))=U(x,y) =y+u(x) as a ...<|control11|><|separator|>
  10. [10]
    4.13 Quasilinear Preferences - EconGraphs
    One common use of a quasilinear utility function is when we're thinking about one good in isolation, or more precisely in comparison to “all other goods.” In ...Missing: properties | Show results with:properties
  11. [11]
    Quasi Linear Preferences | PDF | Utility | Economic Theories - Scribd
    Rating 5.0 (2) The document discusses quasilinear utility functions and provides an example. It defines quasilinear utility functions as being additively separable and ...
  12. [12]
    [PDF] QUASILINEAR PREFERENCES Utility additive, and linear in y: U(x ...
    Quasilinear preferences have a utility function U(x, y) = F(x) + y, where F(x) = x^1/2, and indifference curves are vertically parallel at y = u - F(x).
  13. [13]
    [PDF] Econ 101A — Solution to Problem Set 2
    In the case of quasi-linear utility, there is no income effect on the purchase of good. 1. Whenever income increases, the agent just purchases more of good 2. 5 ...
  14. [14]
    [PDF] Week 3: Demand Theory and Welfare Analysis
    When preferences are quasi-linear, there is no income effect on the amount of x demanded, and so the three demand curves are identical, and so. CV=CS=EV. px.<|control11|><|separator|>
  15. [15]
    [PDF] C ' S l Consumer's Surplus
    S h th h ili. So when the consumer has quasilinear utility,. CV = EV = ∆CS. But, otherwise, we have: Relationship 2: In size, EV < ∆CS ...
  16. [16]
    Quasi-linear preferences – The Economy 1.0 - CORE Econ
    The assumption of quasi-linear preferences makes it possible to measure gains and losses of utility in terms of money. Expand all. An example. An example of a ...
  17. [17]
    [PDF] The Assumption of Equal Marginal Utility of Income - EconStor
    In this paper we use data from seven contingent valuation studies to illustrate that if one instead assumes equal marginal utility of the public good for all.<|control11|><|separator|>
  18. [18]
    [PDF] Equilibrium with Indivisibilities & Income Effects - arXiv
    Jun 30, 2020 · When agents utility functions are quasilinear, they do not experience income effects. When all agents have quasilinear utility functions, we say ...
  19. [19]
    [PDF] Mechanism Design Theory - Nobel Prize
    Oct 15, 2007 · In these papers, the set of possible allo- cations was unidimensional, and the agents had quasi-linear preferences that satisfied a single- ...
  20. [20]
    [PDF] Counterspeculation, Auctions, and Competitive Sealed Tenders
    Author(s): William Vickrey. Source: The Journal of Finance , Mar., 1961, Vol. 16, No. 1 (Mar., 1961), pp. 8-37. Published by: Wiley for the American Finance ...Missing: quasilinear | Show results with:quasilinear
  21. [21]
    Multipart pricing of public goods | Public Choice
    Clarke, E.H. Multipart pricing of public goods. Public Choice 11, 17–33 (1971). https://doi.org/10.1007/BF01726210. Download citation. Issue date: September ...
  22. [22]
    [PDF] Incentives in Teams
    However in this paper the methods of team theory are used in viewing the incentive problem as a problem of team formation or of the formulation of mechanisms ...
  23. [23]
  24. [24]
    A fair pivotal mechanism for nonpecuniary public goods
    The Clarke pivotal mechanism is inappropriate for nonpecuniary public goods, because the assumption of quasilinear utility is invalid, and because the ...
  25. [25]
    [PDF] Optimal Auction Design Roger B. Myerson Mathematics of ...
    Oct 19, 2007 · Optimal Auction Design. Roger B. Myerson. Mathematics of Operations Research, Vol. 6, No. 1. (Feb., 1981), pp. 58-73. Stable URL: http://links ...
  26. [26]
    [PDF] a theory of auctions and competitive bidding
    If the independent private values model is to be applied, one must make two assumptions: that each bidder knows his value for the painting, and that the values ...
  27. [27]
    [PDF] Optimal Auctions John G. Riley; William F. Samuelson The American ...
    Oct 19, 2007 · Optimal Auctions. John G. Riley; William F. Samuelson. The American Economic Review, Vol. 71, No. 3. (Jun., 1981), pp. 381-392. Stable URL ...