RESEARCH DOSSIER
Subjective Value and Marginal Utility
Abstract. This Research Dossier investigates subjective value and marginal utility as foundations of economic choice, exchange, demand, and welfare measurement. It reconstructs the historical development of marginal valuation in Menger, Jevons, Walras, and Marshall; examines the transformation from psychological utility to ordinal preference representation; and derives the conditions connecting marginal trade-offs, budget-constrained optimization, monetary willingness to pay, compensated demand, and observed prices. The analysis distinguishes representation-dependent derivatives from invariant preference relations and substitution ratios, identifies the importance of income and distribution for interpreting monetary valuation, and examines finite-data rationalizability and selected laboratory and field research. Constructed counterexamples and reproducible calculations clarify the explanatory limits of utility maximization, ordinary consumer surplus, market aggregation, and interpersonal comparison. The original analytical contribution is an integrated assessment of the inference permitted by each type of economic evidence. No new named theory or original empirical estimate is proposed.
Keywords: Subjective value; marginal utility; marginalism; ordinal utility; preference representation; scarcity; opportunity cost; marginal rate of substitution; consumer choice; Slutsky equation; willingness to pay; consumer surplus; revealed preference; welfare economics; exchange
The economic object of subjective value
Subjective value is the economic significance of a good, service, or attainable change in circumstances to an evaluating decision maker. Its immediate domain is the relation among alternatives, needs, available resources, and the conditions under which alternatives can be obtained. The term subjective locates evaluation in the decision maker's ordering of attainable outcomes. Physical properties, productive capacity, institutional constraints, knowledge, and access constitute the objective conditions within which that ordering becomes economically effective.
Marginal utility concerns the incremental contribution of a change in consumption within a specified representation of preference. For a divisible good and a differentiable utility representation, it is the partial derivative of utility with respect to quantity, holding the other arguments fixed. For an indivisible good, it is a finite difference between feasible outcomes. These two definitions belong to different mathematical settings. An indivisible dwelling, a month of insurance, and a liter of water cannot be forced into a common smooth physical calculus merely because all can be purchased.
Three distinctions organize this inquiry. Preference orders alternatives; utility represents an ordering under stated conditions; monetary valuation expresses a trade-off between a contemplated change and resources forgone. A market price is the institutional exchange term at which transactions are possible. The four objects are connected by a model of choice, yet they have different units, sources of identification, and domains of validity. The literature on preference representation (Hansson and Grune-Yanoff, 2022) supplies the logical basis of the first two distinctions.
The purpose of the monograph is an integrated examination of these connections, from the marginalist texts to contemporary finite-data tests and field studies. The central inquiry is how much can be learned about subjective value from choice, and under what conditions an interpretation of marginal utility becomes economically informative. The answer developed here is conditional: a ranking may be recoverable without cardinal intensity; marginal trade-offs may be identified without an absolute marginal utility; monetary measures may be useful without constituting a universal welfare scale.
Historical background and the value paradox
Classical political economy investigated value through production, distribution, exchange, and the reproduction of economic society. Smith's distinction between use and exchange in Wealth of Nations (1776, Book I, chapter IV) illustrates the divergence with water and diamonds. What sustains life may exchange for little, while a scarce ornamental good may exchange for much. This observation provides a problem for any theory that maps the total usefulness of a class of goods directly onto exchange ratios.
A careful reconstruction distinguishes the usefulness of maintaining access to the entire good from the value of an additional unit at an existing stock. Imagine a settlement with abundant safe water and a small market for gems. The loss of its entire water supply would be catastrophic. A further liter of water, however, may displace a comparatively minor use; a further gem may satisfy a priority for which few substitutes exist. A monetary exchange ratio can reflect the latter comparison even where the former comparison is decisive for survival.
The explanation depends on accessible quantities rather than total natural abundance. Water may be abundant in a region and scarce at a particular household, season, or quality threshold; the cost of treatment, transport, and reliability may be decisive. A household facing a binding subsistence constraint has a different attainable set from a household connected to a reliable distribution network. Consequently the classic paradox is a question about marginal access, not an invariant ranking of physical substances.
There is a further methodological lesson. A value paradox is often manufactured by comparing different objects: total loss of a fundamental good and willingness to pay for a marginal increment of another. Before comparing values, specify baseline, unit, domain of feasible substitution, and whether the relevant change is marginal or extensive. This principle of good measurement follows from the comparison itself and carries into modern policy appraisal.
Menger and the dependence of satisfaction on scarce goods
In Principles of Economics (1871, chapter III; English edition 2007), Carl Menger develops value in relation to satisfactions whose achievement depends on command over goods. Scarce goods become economically significant when losing command over them entails losing some satisfaction. Where a stock supports multiple uses, a reduction in stock can remove the least important use that would otherwise be maintained. The ordinal priority of uses supplies the economic logic of marginal valuation.
Consider five homogeneous containers of grain allocated to essential meals, preservation, animal feed, preparation for a later period, and a discretionary use. Assuming each use requires one container and the priority order is stable, the household will protect the more important uses when one container disappears. The economically relevant loss is determined by the displaced use. The physical container has no unique intrinsic priority independent of the stock. Reallocation is central to the result; simply identifying the container physically removed would miss the economic opportunity cost.
The example is not a universal numerical law. Some uses demand complementary units. A minimum seed quantity must be maintained to plant a field; two containers together may enable an activity that either alone cannot. Storage losses and future prospects can also change priority. Menger's scarcity interpretation motivates a marginal inquiry while leaving the form of the feasible set open. An analyst must specify divisibility, substitutability, complementarity, and planning horizon.
The significance of this account is its treatment of valuation as the evaluation of dependency. Marginal value changes with the set of satisfactions preserved by available resources. As endowments, technology, and opportunities change, the affected satisfaction may change even when a person's broad purposes remain stable. This is one historical route to understanding why willingness to exchange is context-sensitive without presuming incoherent preferences.
Jevons and the final degree of utility
Jevons develops a mathematical language for the distinction between an aggregate benefit and the utility associated with the last increment. His Theory of Utility (1888 [1871], chapter III) discusses the final degree of utility in relation to quantity. The marginal form makes continuous variation analytically tractable and establishes a role for differentiation in the economic explanation of exchange.
Let U(q) denote a specified numerical utility index for quantity q. A small change dq produces a first-order increment dU = U'(q)dq. If U is twice differentiable and U''(q) is negative, marginal utility in that representation decreases as q increases. The relation concerns a derivative at a given quantity, not the total value of eliminating the entire stock. If U(q) is increasing, total utility can rise while successive equal increments contribute less in the chosen scale.
Jevons's Theory of Exchange (1888 [1871], chapter IV) brings marginal comparisons into exchange relations. In contemporary notation, a decision maker allocating expenditure across two divisible goods can be modeled as equalizing marginal utility per unit of expenditure at an interior optimum. The exchange implication requires both a substitution opportunity and a feasible allocation; it rests on conditions additional to the claim that marginal utility diminishes.
The historical significance of Jevons lies in formalizing incremental evaluation, including the role of variation in availability. The contemporary significance must be stated more narrowly. Differentiating a particular utility function is an analytical operation. Identifying its numerical derivative as a measurable psychological magnitude is a separate proposition, requiring a scale and a measurement strategy. The subsequent history of utility theory reworked precisely this distinction.
Walras and the simultaneous determination of exchange
Walras's marginalist program connects individual evaluations with the simultaneous adjustment of markets. The historical reconstruction by Cervera-Ferri and Insa-Sanchez (2023), Rareness in Walras's theory, traces rarete as an intellectual and formal category in his value theory. Their account supports the historical attribution of the intensity of the final satisfied want and its relationship to marginal formulation. The present dossier uses that scholarship for Walras's intellectual development; it does not claim an independent inspection of the original French editions.
The formal problem of simultaneous exchange can be set out independently. Suppose individuals have endowments, preferences over goods, and access to common exchange prices. Each may demand a bundle whose price does not exceed the value of the individual's endowment. The prices supporting a competitive equilibrium must make the sum of individual demands equal to aggregate available resources, with production incorporated where appropriate. The same set of prices appears in each individual's feasible choice problem.
This structure separates individual valuation from the system that coordinates it. Relative prices are not an arithmetic average of utility numbers. They reflect the compatibility of choices, ownership, production, and scarcity. Multiplying all nominal prices by the same positive scalar does not change the real exchange problem in this elementary model. Nominal price determination requires additional structure.
The value of the simultaneous perspective is particularly clear when one good has several productive or consumption uses. A change in its availability affects the opportunity sets of many agents; equilibrium adjustments alter the scarcity comparisons of other goods. A partial marginal valuation remains useful but its baseline may itself be endogenous. A comprehensive analysis therefore specifies whether prices and resources are held fixed for a household optimization or solved for across interacting markets.
Marshall and the emergence of monetary surplus
Marshall's Gradations of Consumers' Demand (1920 [1890], Book III, chapter III) relates marginal demand to willingness to pay and to the marginal significance of money. His Value and Utility (Book III, chapter VI) develops consumer surplus and the qualifications attached to money as a measure of benefit. The treatment anticipates an enduring practical problem: monetary measures are valuable precisely because money enters the feasible alternatives, yet its significance changes with income and circumstances.
A demand schedule expresses the quantities chosen under specified prices and income, with other relevant conditions held constant. An inverse demand schedule can provide a money-valued marginal willingness to pay under additional assumptions about the choice setting. Integrating under a demand curve, above the price line, then constructs a surplus measure. The monetary accounting is straightforward; the welfare interpretation requires care.
For example, a consumer may be willing to pay a large amount for the first available unit of a reliable service and very little for additional units. At a common unit price, inframarginal units can generate monetary gains over the price paid. Those gains may be large even where the market price is small. Conversely, a high market price need not imply large gains if the purchaser's reservation price is only slightly higher.
Marshall's approximation is especially informative when the income effects of the expenditure under consideration are small. In modern theory, exact money-metric welfare changes are constructed with compensated demand or expenditure functions. Treating an ordinary demand integral as exact in every setting would suppress the distinction between changing purchasing power and the pure trade-off associated with a price change.
From psychological utility to preference representation
Utility has undergone substantial semantic and methodological transformations. Witt's Transformations of Utility Theory (2016) examines the behavioral significance of those transitions. Slutsky's 1915 Theory of the Consumer Budget develops demand analysis in a mathematical setting that does not depend on observing a numerical psychological feeling. The later ordinal and revealed-preference programs further separate the consistency of decisions from the measurement of affect.
Let X be a set of alternatives. A weak preference relation, written x ≽ y, means that x is at least as preferred as y. Completeness permits a comparison of each pair in the specified domain; transitivity requires that x ≽ y and y ≽ z imply x ≽ z. A utility representation U satisfies x ≽ y exactly when U(x) ≥ U(y). Strict preference and indifference follow from these comparisons. A finite complete and transitive ordering admits a numerical representation by ranks. On continuous domains, representation by a continuous real-valued function requires additional regularity; the formal claims depend on the topology and assumptions of the domain.
Ordinal representation is informationally economical. Assigning the same outcome the number 10 instead of 1 has no economic effect if every ordering relation is preserved. If U is a representation and f is strictly increasing on its range, V = f(U) represents the same preferences. The numerical distances between levels can therefore vary while every choice predicted from those preferences remains unchanged.
This is not a denial of subjective experience. An account of experienced satisfaction may have its own research methods, including psychometric observations. Ordinal choice theory establishes what follows from rankings and feasible choice without treating a latent psychological quantity as an observed variable. This separation is fundamental to the interpretation of all subsequent derivative expressions.
Marginal utility and invariance under transformation
Consider a differentiable representation U(x,y). Its marginal utilities are U_x and U_y. The derivatives are expressed in arbitrary utility-index units per physical unit. Under the transformation V = f(U), with f differentiable and f' positive at the point of interest, V_x = f'(U)U_x and V_y = f'(U)U_y. Their absolute magnitudes change by a common factor. Their ratio is preserved.
The second derivative transforms differently: V_xx = f''(U)(U_x)^2 + f'(U)U_xx. Its sign can change even when the underlying preferences are identical. For x > 0, U(x) = √x is increasing and strictly concave, whereas V(x) = U(x)^4 = x² is increasing and strictly convex. These representations generate the same ordering of quantities. The example establishes that diminishing own marginal utility, treated as the negative second derivative of an arbitrary ordinal representative, is not an invariant behavioral property.
This qualification changes how the marginalist proposition should be formulated today. Under sufficient differentiability, the ratio U_x/U_y measures the marginal rate of substitution of x for y: the amount of y the consumer would give up locally to obtain a further unit of x while maintaining the represented ranking. Along a smooth indifference curve with U_y nonzero, dy/dx = −U_x/U_y. Its units are units of y per unit of x. In contrast to the numerical level of U_x, the trade-off corresponds to the geometry of the ordering.
Decreasing marginal rates of substitution can be associated with convex preferences along appropriate indifference curves; this property does not require a unique cardinal utility representation. Convexity describes the structure of preferred sets. Concavity describes a particular representing function. The distinction is operationally important: empirical tests of convex demand rationalization should not be interpreted as measuring the second derivative of a person's psychological utility.
Scarcity, budget sets, and opportunity costs
A standard price-taking consumer faces a budget set B(p,m) consisting of nonnegative bundles x satisfying p·x ≤ m, with p a strictly positive price vector and m nonnegative income available for the modeled commodities. Opportunity cost emerges from the boundary. Obtaining one additional unit of commodity i at price p_i absorbs purchasing power that could buy p_i/p_j units of commodity j if other expenditures remained unchanged.
If income is finite and all prices are positive, this budget set is compact. A continuous utility representation attains a maximum on the set. Local nonsatiation ensures a maximizing bundle exhausts the budget: any slack permits a sufficiently small feasible improvement. Strict quasiconcavity can ensure uniqueness on a convex budget set. These properties supply existence, boundary, and uniqueness conclusions respectively; each has a distinct role.
Actual feasible sets often incorporate fixed costs, indivisibilities, contractual minimums, time, borrowing restrictions, administrative eligibility, rationing, and geographic access. A nominally affordable item may be inaccessible to a person who lacks the relevant infrastructure. A consumer may also have unused money because some future alternatives are outside the model. Such cases call for an expanded or differently specified feasible set, not an automatic attribution of irrationality.
Scarcity is a relationship between purposes and attainable means. A large physical stock does not eliminate scarcity when access is costly or its allocation restricts other valuable purposes. Individual choice models express that relation by feasible alternatives; institutional and production analysis explains the origin of those alternatives.
Interior allocation and unequal marginal trade-offs
For an interior optimum of a differentiable utility function with a binding budget, a Lagrangian can be written as U(x) + λ(m − p·x). The first-order conditions are U_i(x*) = λp_i for every consumed good i. Hence U_i/p_i = U_j/p_j at that optimum. An elementary illustration of this allocation condition appears in Consumption Choices (Greenlaw, Shapiro, and MacDonald, 2022).
This equation has a straightforward economic derivation. If a small expenditure transfer from j to i is feasible, the utility gain from the transfer is proportional to U_i/p_i − U_j/p_j. A strictly positive difference indicates a direction of improvement. At an interior optimum there can be no such profitable local reallocation. Concavity of the objective is a sufficient condition for the first-order conditions to characterize a global maximum in this linear-budget setting.
For a good consumed at zero, the condition changes. Under differentiability and the usual constraint qualifications, U_i ≤ λp_i at that boundary. A good with marginal benefit below its opportunity cost need not be purchased. For perfect substitutes, an interior allocation can occur only when the relevant preference weights and prices create an exact tie; otherwise a corner allocation is natural. Perfect complements may generate a kink at which the smooth derivative argument cannot be used at all.
The multiplier λ is the marginal change of the optimum utility index with additional income under envelope conditions. It is representation dependent. A monetary marginal trade-off is obtained by dividing U_i by λ when λ is positive. Consequently the first-order calculation produces meaningful relative conditions while its unnormalized utility numbers remain dependent on the selected index.
Indivisibility, complements, and thresholds
Marginal value can refer to a discrete change. Consider a constructed choice problem in which money is a numeraire with linear utility and successive units of a good contribute 12, 8, 5, and 2 monetary-equivalent units. At a price of 6 per unit and sufficient income, the consumer purchases two units: the first two increments exceed the cost, the third and fourth do not. The resulting net surplus is (12 − 6) + (8 − 6) = 8. These numbers are an illustrative numerical schedule, not observations.
The result depends on separability and on the feasibility of purchasing a prefix of the sequence. Suppose instead an installation brings no benefit unless two components are acquired together, after which the benefit is 20, while each component costs 7. Evaluating each component in isolation would reject both; evaluating the attainable pair yields a net gain of 6. The relevant increment is the complementary project. A smooth marginal condition at zero would fail to recognize the threshold.
Thresholds are especially important for access to transport, connectivity, medical treatment, and durability improvements. The value of the first component can depend on whether the other components are already available. Quantity, quality, reliability, and timing jointly determine the service obtained. A physical-unit demand schedule that ignores those dimensions can combine changes in structurally different goods.
Discrete valuation thus requires comparison of the total utility of feasible alternatives or a discrete choice model. Marginal analysis retains a role in changes within an already functioning system; it must be applied to an economically appropriate unit of decision.
An explicit model of diminishing willingness to pay
Consider a constructed quasilinear preference representation U(x,z) = 12 ln(1+x) + z, where x is a nonnegative quantity and z is consumption of a numeraire priced at one. The budget is px + z ≤ m. Assuming enough income for z to remain positive, utility maximization reduces to maximizing 12 ln(1+x) − px. The interior condition is 12/(1+x) = p, giving x = 12/p − 1 when p < 12; demand is zero at prices of 12 or above. The income requirement for the displayed interior solution is m > p x.
At p = 3, the chosen quantity is 3 and expenditure on x is 9. At p = 2, quantity rises to 5 and expenditure is 10. The net monetary surplus at p = 2 is 12 ln 6 − 10, approximately 11.5011. The example illustrates how a specified declining derivative creates a declining inverse demand schedule and how the monetary normalization of the numeraire supports its interpretation.
The functional form is pedagogical. It is not a claim that real households have logarithmic preferences or a constant marginal utility of income. In fact, if income is less than the required expenditure, the nonnegative numeraire constraint binds and the displayed solution ceases to apply. This is precisely why quasilinear demand and surplus must be interpreted on their valid domain.
A strictly increasing transformation of the entire utility index leaves the choices and money-equivalent valuations unchanged but generally removes the quasilinear numerical appearance. The money interpretation depends on the economic substitution structure, not on preservation of a particular algebraic expression.
Income, willingness to pay, and the meaning of money
Monetary valuation is always relative to an opportunity. Define an individual's willingness to pay for an improvement as the maximum monetary reduction in initial resources compatible with accepting the improvement rather than retaining the baseline. Where the relevant indifference condition has a solution, this is a compensating money amount. Willingness to accept a deterioration or surrender of an entitlement solves a related problem in a different baseline.
A transparent example uses U(x,z) = ln x + ln z, with both arguments strictly positive, and baseline x = 1. A change to x = 2 maintains the initial utility when z falls to half its former quantity: ln 2 + ln(z/2) = ln z. An individual with 20 units of the numeraire can therefore give up 10, while an individual with 200 can give up 100. The two people have identical specified preferences; their feasible compensations differ solely because their baselines differ.
The example rules out a common interpretation of monetary demand. Greater willingness to pay need not imply greater need or a stronger subjective feeling. Richer people can surrender more units of money at a given trade-off because their resource positions differ. Conversely, inability to pay is not evidence of the absence of value. It is evidence about choice under a constraint.
For practical valuation, the money metric must name the initial entitlement, income, price vector, access conditions, and change of outcome. In distributive analysis it is appropriate to present the distribution of valuations and the distribution of ability to pay separately, rather than treating a monetary sum as a direct measure of the strength of all preferences.
Demand systems and aggregation
Consider Cobb-Douglas preferences U(x,y) = x^a y^(1−a), with 0 < a < 1 and positive prices and income. The expenditure shares at an interior optimum equal a and 1−a, yielding x = am/p_x and y = (1−a)m/p_y. Under a = 1/2, m = 120, p_x = 2, and p_y = 4, the optimum is (30,15). If p_x rises to 4, the optimum becomes (15,15). This illustrates an ordinary demand response at fixed nominal income.
The formulation highlights the distinction between household and aggregate demand. Consider two consumers with different shares, a_A = 0.8 and a_B = 0.2, facing p_x = 1. Total income is 100. With incomes 80 and 20, aggregate demand is 0.8 × 80 + 0.2 × 20 = 68. With incomes reversed, aggregate demand is 32. Total income, preferences, and prices remain unchanged; the income distribution changes. The constructed comparison proves that aggregate expenditure cannot generally be represented as a function of total income and prices alone without restrictions on the distribution or individual demand structures.
An aggregate consumer is therefore a modeling device whose legitimacy depends on aggregation conditions or the research objective. A representative agent can simplify a price system, but the simplification may erase precisely the distributional information needed to interpret changes in subjective value. The observation is particularly important when subsistence expenditures differ markedly across income groups.
Empirical demand estimation should distinguish heterogeneous preference parameters, heterogeneous constraints, and the composition of the observed population. These factors can produce similar movements of aggregate demand while implying different policy responses. The choice of aggregation is a substantive scientific assumption, not a consequence of summing individual utilities.
Substitution, compensation, and the Slutsky relation
A Marshallian demand x(p,m) maximizes preference subject to nominal expenditure m. A Hicksian demand h(p,u) minimizes expenditure subject to obtaining utility level u in a selected representation. The expenditure function e(p,u) gives that minimum cost. At a regular interior solution with relevant duality conditions, h(p,U(x(p,m))) = x(p,m).
The compensated and uncompensated responses differ because a price change alters purchasing power. The Slutsky equation for commodity i and price j is
∂x_i/∂p_j = ∂h_i/∂p_j − x_j ∂x_i/∂m,
evaluated at the corresponding utility and expenditure. Slutsky's historical consumer budget analysis (1915) is the foundational source for this separation; the notation here is a modern derivation. To obtain the equation, differentiate x(p,e(p,u)) = h(p,u) with utility held fixed and use ∂e/∂p_j = h_j under differentiability. Rearrangement gives the displayed formula.
The compensated own-price derivative is nonpositive under the standard expenditure-minimization structure, wherever the derivative exists. One proof compares the costs of Hicksian bundles at two price vectors using their minimizing properties; adding the inequalities yields (p'−p)·(h'−h) ≤ 0. If only one price changes, the corresponding compensated quantity cannot increase with that price.
An inferior good has negative income derivative in the relevant domain. For such a good the second term in the own-price equation is positive; it may offset the nonpositive compensated term. If the offset is sufficiently large, ordinary demand increases when the price rises. The Giffen possibility is therefore consistent with preference maximization and a nonpositive compensated response. A universal downward-sloping ordinary demand claim would be stronger than the theory warrants.
Giffen behavior and subsistence constraints
Jensen and Miller (2008), in Giffen Behavior and Subsistence Consumption, report experimental price-subsidy evidence from poor households in China, with strong evidence for rice in Hunan and weaker evidence for wheat in Gansu. The interpretation is linked to subsistence consumption. The finding is important because it turns a theoretical exception into a context-specific empirical question rather than a logical curiosity. This dossier relies on the accessible abstract for that reported result and does not re-estimate its effects.
The mechanism can be described without assigning cardinal utility. A household near a nutritional threshold allocates expenditure among a low-cost staple and more desirable alternatives. An increase in the staple's price reduces real purchasing power. If preserving minimum consumption requires sacrificing sufficiently expensive alternatives, the household may shift toward greater consumption of the staple despite its higher price. This is an uncompensated response. Maintaining previous welfare through compensation would change the comparison.
The mechanism yields a testable conditional prediction rather than a universal characteristic of staples. The response depends on the initial income range, relative prices, nutrient requirements, access to substitutes, and how consumption is measured. At more extreme poverty, a household may lack the resources to maintain the threshold; at greater income, ordinary substitution may dominate. A field design should therefore permit nonlinear and heterogeneous responses rather than impose a single population-wide slope.
For interpretation, a Giffen observation does not establish that marginal utility rises with quantity. It concerns the joint effect of price and income on choice. It is also distinct from demand increases caused by quality inference, status effects, or expectations of future price increases. Those mechanisms change perceived outcomes or information rather than generating the same fixed-preference comparative static.
Exact money-metric welfare and ordinary surplus
Consumer surplus based on ordinary demand is often a useful approximation. Exact compensation measures derive from the expenditure function. For a price increase from p0 to p1 at constant nominal income m, let u0 = v(p0,m) and u1 = v(p1,m) be the initial and final utility levels. A positive loss measured by compensating variation is CV = e(p1,u0) − m. A positive loss measured by equivalent variation is EV = m − e(p0,u1). The former restores the old standard at new prices; the latter measures how much could have been taken at old prices to induce the new standard.
In the Cobb-Douglas example with a = 1/2, e(p,u) = 2u√(p_x p_y). At the initial bundle (30,15), u0 = √450. At the final bundle (15,15), u1 = 15. The price of y remains 4 while that of x doubles from 2 to 4. The compensating variation is 8√450 − 120 ≈ 49.7056. The equivalent variation is 120 − 30√8 ≈ 35.1472. The integral of ordinary demand over the price increase is 60 ln 2 ≈ 41.5888. All three are monetary losses, yet they answer different questions.
The expenditure function provides exact compensation under fixed preferences, the specified price system, and feasible expenditure minimization. A strictly monotone transformation merely relabels the utility target and leaves its cost unchanged. Thus the money metrics are compatible with ordinal preference representation. They do not require measurement of numerical happiness.
The example also explains why Marshall's consumer surplus analysis is best interpreted with attention to the variation of purchasing power. The ordinary integral coincides with both compensation measures in the quasilinear setting where income effects for the good vanish over the relevant domain. Once income effects are material, the three quantities separate. Reporting their definitions and the sign convention is part of valid welfare accounting.
Subjective valuation and exchange prices
Suppose a buyer has successive reservation values of 14, 10, and 6, while a seller has successive incremental costs of 3, 7, and 11, in a common monetary unit. Under a simple separable exchange model with no external effects, the first two units generate net gains of 11 and 3. A third loses 5. At a unit price strictly between 7 and 10, the buyer wants two units and the seller supplies two. At the boundaries, indifference can create additional feasible outcomes. The efficient quantity in this constructed example is two and the total gains are 14, regardless of the division of those gains at a clearing price within the interval.
The reservation values are subjective, but the seller's cost schedule and common price mechanism are also essential. A buyer's strong preference does not by itself determine the price or the feasible amount supplied. The actual price can depend on negotiation, competitive entry, capacity, property rights, and the strategic alternatives available to each side. A market clearing price need not uniquely determine the surplus division in settings with indivisibility.
For multiple buyers and sellers with different opportunities, relative valuations shape demand and supply while exchange institutions determine how bids and offers meet. Individual marginal rates of substitution can align with price ratios in a competitive model under the relevant interiority and access assumptions. Market power or rationing can produce wedges between individual trade-offs and exchange terms.
The explanatory role of subjective value is consequently substantial but bounded. It helps specify demand, reservation decisions, and gains from trade. Production and ownership explain feasible supply and the incidence of income. Market institutions explain transaction rules and bargaining positions. Price formation requires the interaction of these elements.
Production, cost, and the limits of a one-sided explanation
A marginal valuation account becomes incomplete when it ignores the transformation of resources into goods. Production creates economically available alternatives through technology, organized labor, capital, time, and access to inputs. The opportunity cost of output is the value of the best feasible alternatives sacrificed in producing it. In a competitive setting, marginal cost can characterize supply under convexity and regularity assumptions; with fixed costs and increasing returns, that condition alone may not describe viable operation.
Consider an infrastructure service with fixed cost F and constant variable cost c per unit. Total cost is F + cq. A marginal-cost price c can make consumption decisions efficient within the network once capacity and access are given, yet revenue cq does not recover F. The economics of access, financing, and distribution becomes central. A low willingness to pay for some additional units cannot by itself settle the decision to create the network or its financing arrangement.
This example is also a warning against identifying the classical and marginalist questions as interchangeable. Reproduction, labor allocation, rent, and the generation of production surplus concern structural processes. Marginal preference theory concerns the evaluation and choice of feasible alternatives. Both can enter a theory of exchange, while answering different explanatory questions.
Institutional constraints are equally consequential. Rights of access determine who can obtain a resource; contracts determine how risk and costs are divided; trade rules determine whether exchange can occur. These conditions affect effective scarcity and opportunity cost. Subjective valuation operates through the actual institutional arrangement, not in a vacuum preceding the economy.
Revealed preference as a test of choice consistency
Revealed preference analysis examines whether observations of selected bundles and their budgets are consistent with a stable ordering. It puts a testable restriction on choice without measuring intrinsic utility intensity. Let observations be indexed by t, with positive price vectors p_t and nonnegative selected bundles x_t. Bundle x_t directly reveals preference over x_s when p_t·x_t ≥ p_t·x_s. The alternative x_s was affordable when x_t was chosen.
Let R be the transitive closure of this direct relation. The generalized axiom of revealed preference (GARP) requires that if x_t R x_s, p_s·x_s ≤ p_s·x_t. A strict reversal would mean that the later choice rejected a strictly cheaper bundle indirectly revealed to be at least as preferred. Under local nonsatiation, such a reversal is inconsistent with the proposed stable ordering.
Afriat's 1967 Construction of Utility Functions established the finite-data construction; the mathematical content used in this dossier was checked against Varian's 1982 Nonparametric Approach to Demand Analysis, including its derivation. The accessible Afriat record establishes the original bibliographic attribution rather than an independently inspected full text.
For a finite dataset satisfying the theorem's conditions, there exist real numbers u_t and positive multipliers λ_t satisfying u_s ≤ u_t + λ_t p_t·(x_s − x_t) for all s,t. A rationalizing monotone concave utility representation can be constructed as the minimum across the affine functions u_t + λ_t p_t·(x − x_t). GARP and the inequalities provide a computational route to testing rationalizability under the maintained assumptions. They establish existence of at least one compatible representation; they do not determine a unique representation of utility or preference outside the observed constraints.
A reproducible finite-data example
Consider the following entirely constructed data with two commodities. In dataset A, observation 1 has prices (1,2), income 8, and the chosen bundle (4,2); observation 2 has prices (2,1), income 8, and the chosen bundle (2,4). At the first prices, the second bundle costs 10. At the second prices, the first bundle also costs 10. Neither is affordable in the other observation. With no conflicting direct revealed preference, these data satisfy GARP. They are consistent with the specific representation U(x,y) = √(xy).
Dataset B uses the same two price vectors. At prices (1,2), choose (1,3) at cost 7; its alternative (3,1) costs 5. At prices (2,1), choose (3,1) at cost 7; the alternative (1,3) costs 5. Each chosen bundle strictly reveals preference over the other, creating a cycle with strict affordability. These two observations fail GARP under stable, locally nonsatiated preferences and the stated budgets.
The comparison supplies both a positive and negative control for a demand-analysis implementation. Compute expenditures for every observation-bundle pair. Build a directed adjacency matrix from weak affordability; compute its transitive closure; reject when a path from t to s combines with strict reverse affordability at s. A numerical implementation must state tolerance for rounding and currency measurement. A tolerance should be selected independently of the desire to pass the test.
Failure of the test is evidence against the joint maintained conditions. A change of tastes between observations, missing goods, different quality, mismeasured prices, or an incorrect budget can produce an apparent violation. Passing has a narrower implication: at least one stable preference ordering can rationalize the observations within the specified setting. Both results are scientifically useful when interpreted at the correct logical level.
Identification, heterogeneity, and the evidential boundary
A single choice under a known budget can reveal that the selected bundle was preferred to the alternatives the consumer could have selected. It generally does not reveal the detailed ranking of alternatives outside that set. Additional choices at varying relative prices enlarge the tested comparisons; they do not automatically identify every aspect of preferences over the entire commodity space.
The finite-data result establishes a set of compatible orderings rather than a numerical latent utility. If a concave representation U rationalizes the observations, a positive rescaling and an additive constant preserve choices. Many other representations may do so as well, subject to the maintained regularity. Different rationalizations may have different willingness-to-pay implications for counterfactuals outside the observed comparisons. In applied work, bounds and sensitivity analysis are more transparent than unsupported point estimates.
Statistical identification adds problems absent from deterministic demonstrations. If households face different effective prices because of travel costs or discounts, then posted prices misrepresent feasible sets. If quantity is recorded at the household level while utility is attributed to an individual, the decision-making unit has changed. If the sample selection depends on unobserved preferences, market purchases may not identify population-level valuations. A stable ordering cannot be tested without a stable definition of the alternatives being compared.
A credible empirical design specifies observation units, relevant prices, restrictions, timing, information, and the hypothesized preference invariance. The question of whether valuations are identified is then separable from the question of whether choices satisfy a consistency restriction. This distinction is a central interpretive finding of the dossier: the existence of an explanatory representation should never be confused with its empirical uniqueness.
Experiments, consistency, and observed wealth
Laboratory budget-choice designs permit researchers to set feasible alternatives and observe repeated selections. Choi, Fisman, Gale, and Kariv (2007), Consistency and Heterogeneity, report experimental evidence from graphical portfolio-choice settings that enables an examination of individual consistency and heterogeneity under uncertainty. The accessible publisher abstract supports this design-level characterization. It does not supply a basis here for re-estimating the original preference parameters or their distributions.
Choi, Kariv, Muller, and Silverman (2014), Who Is (More) Rational?, report that a one-standard-deviation increase in consistency with economic rationality was associated with approximately 15–19 percent higher household wealth in their study. This is a reported association, not a universal causal effect of rationality. Wealth may influence opportunities for learning and financial decision making; common causes and measurement mechanisms may affect both outcomes. The present inquiry uses the reported estimate to explain the significance and limits of a consistency metric.
The research design has broader relevance. If a participant receives varying budget lines, an analyst can compare selected bundles with previously available alternatives. This approach avoids asking participants to directly report marginal utility in arbitrary numerical units. At the same time, the budget representation, salience of payment, understanding of the task, and the relationship between experimental stakes and ordinary decision contexts all influence interpretation.
Repeated-choice experiments answer an important question: whether behavior in a specified environment admits a coherent choice representation and with what degree of departures. They do not by themselves establish identical behavior across every domain, cardinal welfare comparisons across persons, or a permanent personal attribute of rationality.
Field evidence from pricing and product use
Field experiments introduce real access conditions and consequential purchases. Ashraf, Berry, and Shapiro (2010), Higher Prices and Product Use, examine the marketing of a water-treatment product in Zambia. Their reported distinction between screening and sunk-cost effects is directly relevant to the meaning of willingness to pay. The publisher abstract reports screening but no consistent evidence of the sunk-cost mechanism in their experiment. No estimate beyond the accessible source is asserted here.
A positive price can select purchasers who expect to use a product or are better positioned to obtain its benefits. Usage among purchasers at different prices can therefore differ even if paying more has no direct motivational effect. To distinguish selection from a causal effect of payment, a research design must separately vary the relevant price or payment conditional on the decision to acquire the product, under a defined randomization arrangement.
This observation matters for public welfare analysis. An intervention with strong health or social benefits may have low uptake because its intended recipients face limited liquidity, lack information, or face complementary costs. A screening effect can improve the targeting of actual use; a higher price can simultaneously exclude people who would benefit. The net decision depends on benefits, externalities, distribution, administrative costs, and the behavioral response among those who gain or lose access.
The field context demonstrates why a purchase decision is not a direct reading of the significance of a good. It is a decision within an actual distribution mechanism. Marginal subjective valuation helps explain price-sensitive uptake, while a separate causal and welfare inquiry determines what outcomes the acquisition produces and how they should be assessed.
Endowment dependence and reference conditions
Kahneman, Knetsch, and Thaler's 1990 Experimental Tests of the Endowment Effect provide evidence that valuation can depend on initial possession in experimental exchange. The bibliographic and abstract-level record supports the existence of such experimental results, but this dossier does not claim exhaustive assessment of the experimental procedures, magnitudes, or later replications.
Several conceptually distinct mechanisms can account for discrepancies between willingness to pay and willingness to accept. Ordinary income effects can produce differences even under stable preferences. A change in perceived reference state can alter which consequences are psychologically salient. Transaction costs, uncertainty about quality, strategic behavior, and unfamiliarity with elicitation can affect stated or revealed values. Identifying one mechanism requires evidence capable of distinguishing it from the others.
A well-defined valuation question therefore states whether the initial baseline includes legal possession, expected access, cash resources, and relevant substitutes. Loss of an entitlement and acquisition of the same physical object are different changes from different initial states. Equal physical descriptions do not guarantee equal economic questions. A direct comparison of the monetary values is meaningful only after the baselines and elicitation conditions are made explicit.
The implication is methodological rather than a new preference theory. The analyst should test stable-preference explanations first where the maintained assumptions are justified, document deviations, and introduce reference-dependent or institutional explanations where observations discriminate among them. A difference in valuations is an empirical object to be explained, not an automatic measure of a universal psychological parameter.
Uncertainty, expected utility, and the scope of cardinality
Choices over risk require distinctions from ordinary ordinal representations over certain bundles. In the expected utility model for lotteries, a numerical utility index enters a probability-weighted expectation. Under the representation's axioms, positive affine transformations preserve the lottery ordering. A general strictly increasing nonlinear transformation of the outcome utility does not necessarily preserve expected utility comparisons across lotteries. This is why curvature can have a decision-theoretic interpretation within the expected utility structure while curvature of an arbitrary ordinal representation over certain consumption cannot.
Consider an illustrative monetary gamble paying wealth 50 or 150 with equal probability. A risk-neutral expected-value comparison assigns mean wealth 100. If the stipulated expected utility of wealth is logarithmic, the certainty equivalent c solves ln c = (ln 50 + ln 150)/2, giving c = √7500 ≈ 86.6025. The risk premium is 100 − c ≈ 13.3975. These are calculated consequences of a specified model, not measured attitudes.
The marginal utility of wealth in such a model can help explain risk premiums and choices of insurance or saving. Interpretation nevertheless depends on the actual domain of wealth, background risk, the set of feasible contracts, and whether the probability structure is known. A utility index fitted to a restricted portfolio choice task may not transfer unchanged to all other decisions under uncertainty. The experimental literature on portfolio choice consistency illustrates the relevance of testing the maintained choice conditions.
Cardinality here is conditional on the axioms governing lotteries. It is not interpersonal cardinality. The affine normalization can differ from person to person. Neither comparing raw utility numbers across persons nor aggregating them as social benefits follows from expected utility representation alone.
Time, interdependence, and the boundaries of the commodity
Many economic decisions concern dated consumption and durable services. A heating system yields benefits over a stream of periods, a loan redistributes resources through time, and preventive treatment changes the probability of future adverse outcomes. An incremental purchase must be evaluated relative to a dated opportunity set and the state-contingent services the purchase provides.
A standard illustrative intertemporal objective is U(c_0,c_1) = u(c_0) + βu(c_1), where β is a specified positive discount weight. With a two-period present-value budget c_0 + c_1/(1+r) ≤ W and an interior optimum, the first-order condition is u'(c_0) = β(1+r)u'(c_1). This relation identifies a marginal trade-off conditional on discounting, the feasible interest rate, and the selected additive structure. It does not independently identify β and the curvature of u from one observed allocation.
Credit constraints can prevent households from realizing this interior allocation. Some consumers may highly value future benefits yet be unable to finance the initial cost. Joint household decisions can also depend on bargaining, individual consumption allocations, and shared goods. Treating a household as a single unit may be reasonable in an application, but its welfare interpretation needs a separate assumption about internal distribution.
Social interdependence adds another boundary. If an individual's outcomes depend on other people's consumption, the ranking is over a broader state than a private commodity bundle. Status, altruism, network participation, and environmental effects may alter the economically relevant alternatives. The scientific response is to specify what the good delivers in the observed context and distinguish these dependencies from a change in a numerical marginal utility parameter.
Welfare, need, and interpersonal comparison
Individual preference satisfaction is an important economic object. Its relation to welfare depends on what individuals know, what outcomes they can attain, how their preferences are formed, and the normative framework adopted for assessing well-being. Hausman's Preference Satisfaction and Welfare (2011) develops a critical distinction relevant to this relationship. The accessible publisher summary supports the distinction; its full philosophical argument has not been independently reconstructed here.
Ordinal representations establish how an individual ranks alternatives. They do not establish whether equal utility-index changes have equal significance across individuals. Because the index can be independently transformed for each person, a sum of raw utility levels lacks an invariant meaning without further interpersonal normalization and a normative aggregation principle. The separateness of the two problems is essential: accurate individual demand estimation does not itself solve social welfare evaluation.
Suppose an improvement is valued at 10 monetary units by one household and 100 by another. Summing amounts can answer a specified question about total money-metric willingness to pay. Whether the latter household should receive priority in a public allocation depends on the objectives of the decision, the distribution of resources, entitlement, external consequences, and any adopted social weights. The monetary sum neither enacts nor justifies a distributive principle by itself.
A policy evaluation can use several independently interpretable objects: resource cost, demand or uptake, compensation measures, physical or health outcomes, distributional incidence, and explicit normative weights. Keeping these objects visible makes the evaluation more informative than suppressing them into an apparently neutral cardinal aggregation. This monograph advances no new welfare doctrine; it identifies the assumptions necessary for interpreting the standard instruments.
Measurement strategies and international applicability
A feasible empirical program begins with the distinction among price, quantity, expenditure, and reported subjective evaluations. Household expenditure surveys record outlays and characteristics using specific sampling and reporting methods. The U.S. Bureau of Labor Statistics' Consumer Expenditures and Income methodology describes interview and diary collections. Such data can inform models of expenditure behavior, budget shares, and heterogeneity. They do not directly record the derivative of an individual's ordinal utility representation.
A valuation study should document quality adjustment, units, in-kind acquisition, tax and subsidy treatment, price dispersion, missing purchases, household composition, and survey design. Price and quantity variation must be sufficient for the proposed demand parameters; endogenous prices may require an identification strategy. When observations are only expenditures, quantities and effective prices cannot always be recovered without auxiliary information. A credible estimate must match the observational resolution of the underlying data.
Cross-country comparisons add exchange-rate and purchasing-power questions. The World Bank's International Comparison Program methodology provides procedures for constructing internationally comparable purchasing power measures. These comparisons address the cost of specified goods and services and the comparison of economic aggregates. They do not establish that a monetary amount has identical subjective significance for persons in different countries.
The appropriate comparative design separates local effective prices, incomes, availability, institutions, and demographic composition. It specifies whether the comparison concerns nominal willingness to pay, constant-price expenditure, a purchasing-power-adjusted amount, or compensated welfare at a stated reference system. Each answers a different question. The strongest generalizable result of marginal analysis is the conditional structure of comparison, rather than an assertion that estimated magnitudes transfer mechanically across economies.
An empirical research protocol for future investigations
A research project designed to estimate marginal valuation should first define the outcome and its economically relevant increment. For a service, the increment may concern continuity, quality, or access probability rather than physical quantity. The project should then establish the baseline and the alternative uses of expenditure and time. The question determines whether a deterministic demand model, a discrete choice model, a field pricing experiment, or a compensated-money evaluation is appropriate.
For observational demand data, record the sampling population, prices actually available, quantities, income or expenditure resource, outside options, and the timing of observations. Test whether price variation is plausibly exogenous to unobserved demand. Where it is not, explicitly investigate policy-induced or experimental variation, or another defensible source of identification. Report uncertainty, model sensitivity, and identification limits. In finite-data revealed preference studies, report the tested restrictions, tolerance, accessible-budget comparisons, and any modifications needed because of nonconvex feasible sets.
For experimental valuation, distinguish hypothetical responses from consequential decisions. Establish the incentive and information conditions, report exclusions and attrition, and separate selection into an observed decision from the causal effect of the assigned price. Pre-specify heterogeneous analysis where subsistence, liquidity, or initial ownership is plausibly material. The Zambia pricing study and the China subsistence study illustrate distinct designs with distinct inferential objects; neither alone calibrates a global valuation function.
For welfare evaluation, state the counterfactual price or service environment and the sign and baseline of the monetary measure. Distinguish private compensation from social benefits and external effects. Report distribution by relevant characteristics before aggregating quantities with explicit weights. This research protocol is a methodological synthesis of the examined established literature and the dossier's internal analysis, not a proposed new economic model or empirical result.
The present monograph is theoretical, historical, and evidence-critical. No original survey, econometric estimate, or experimental dataset has been produced. All numerical examples are synthetic and fully specified, and selected empirical findings are attributed to their sources. These limitations are explicit to preserve the distinction between original interpretation and original measurement.
Analytical synthesis: what marginal utility explains
The combined historical, formal, and empirical analysis supports a determinate conclusion about the scientific object. Subjective value is relational: it concerns the significance of attainable changes to an evaluating decision maker. Marginal analysis specifies which change is being considered. Preference representation supplies the logical ordering; the economic constraint supplies its opportunity cost. Exchange and welfare analysis then introduce additional relations that translate selected comparisons into monetary trade-offs.
The historical authors identified essential components of this structure. Menger emphasized the dependency of satisfaction on scarce command; Jevons formalized incremental utility and exchange; Walras embedded marginal evaluation in simultaneous market coordination; Marshall interpreted demand and surplus through money and its changing significance. Subsequent ordinal analysis established that choice can be explained through orderings without assigning a unique number to psychological utility. Revealed-preference methods gave that interpretation empirically testable finite-data restrictions.
The mathematical results of this dossier connect these components. The derivative transformation establishes the dependence of absolute marginal utilities on ordinal scale. The substitution ratio establishes the invariance of an important trade-off under regular monotone transformations. The Slutsky relation establishes the distinct roles of compensation and purchasing power. The expenditure-function examples establish how alternative money metrics diverge in the presence of income effects. The two-household aggregation example establishes the importance of distribution in market demand. The revealed-preference examples distinguish existence of a rationalizing ordering from identification of a unique scale.
Taken together, these results clarify why subjective valuation can be economically rigorous without being directly observed as a numerical psychological object. A choice comparison is evidence about an ordering within a feasible set. A demand response is evidence about the interaction of substitution and resources under maintained conditions. A money-metric valuation is evidence about a specified trade-off and baseline. A market price is a realized or offered exchange term generated by an institutional system. Each object contributes a different explanatory layer.
The practical consequence is that the appropriate economic unit of measurement is determined by the decision being investigated. A marginal derivative is appropriate for a smooth local trade-off, a finite compensating amount for an attainable discrete change, a demand estimate for a price-response question, and a welfare comparison for a counterfactual under explicit normative assumptions. This is an analytical organization of existing results, not a novel named theory. Its contribution is to make the inferential boundaries explicit across research traditions.
Conclusion
Subjective value and marginal utility constitute a fundamental domain of economic science because they connect individual evaluation with scarcity, resource allocation, and exchange. Their enduring explanatory power resides in the study of attainable changes and forgone alternatives. The historical marginalist contributions developed this problem in different languages; modern preference theory established the logical and mathematical conditions under which the problem can be analyzed without measuring a unique cardinal utility of certain consumption.
The analysis distinguishes marginal utility as a derivative or finite difference of a representation from invariant marginal substitution, monetary compensation, demand, and market price. The distinctions establish the limits of a numerical interpretation of marginal utility and the conditions under which money-metric valuation is meaningful. The allocation condition is valid at a regular interior optimum; corners, complementarity, and thresholds require appropriate alternative formulations. Compensated and uncompensated demand are scientifically different objects, as the Slutsky relation and the possibility of Giffen behavior demonstrate.
Subjective evaluation alone cannot explain the production of attainable goods, the distribution of resources, institutional access, or the simultaneous coordination of markets. Those conditions form the economic environment through which preferences become effective. A price can coordinate exchange without measuring total human need. The ability to pay affects monetary valuation; income distribution affects aggregate demand even when the underlying preferences are stable.
The empirical inquiry confirms the usefulness of carefully bounded claims. Revealed preference establishes restrictions on finite choice observations. Experimental and field evidence tests specific behavior under specified arrangements. Expenditure surveys and purchasing-power comparisons measure economic observables rather than absolute utility. Welfare conclusions require explicit criteria and transparent treatment of distribution.
The scientific position of this dossier is therefore that subjective value is a central explanatory category of choice and exchange, and that marginal utility is a powerful conditional analytical instrument. The instrument achieves its greatest precision when the representation, feasible alternatives, baseline, price conditions, and intended inference are specified together. The resulting economics connects the individual significance of goods to the institutional and productive order without confusing the different kinds of evidence that support each part of the explanation.
References
-
Afriat, Sydney N. 1967. The Construction of Utility Functions from Expenditure Data. International Economic Review 8 (1): 67–77. doi:10.2307/2525382. Construction of Utility Functions
-
Ashraf, Nava, James Berry, and Jesse M. Shapiro. 2010. Can Higher Prices Stimulate Product Use? Evidence from a Field Experiment in Zambia. American Economic Review 100 (5): 2383–2413. doi:10.1257/aer.100.5.2383. Higher Prices and Product Use
-
Cervera-Ferri, Pablo, and Pau Insa-Sanchez. 2023. Rareness in the Intellectual Origins of Walras's Theory of Value. Journal of the History of Economic Thought 45 (3): 467–485. doi:10.1017/S1053837221000547. Rareness in Walras's theory
-
Choi, Syngjoo, Raymond Fisman, Douglas Gale, and Shachar Kariv. 2007. Consistency and Heterogeneity of Individual Behavior under Uncertainty. American Economic Review 97 (5): 1921–1938. doi:10.1257/aer.97.5.1921. Consistency and Heterogeneity
-
Choi, Syngjoo, Shachar Kariv, Wieland Muller, and Dan Silverman. 2014. Who Is (More) Rational? American Economic Review 104 (6): 1518–1550. doi:10.1257/aer.104.6.1518. Who Is (More) Rational?
-
Greenlaw, Steven A., David Shapiro, and Daniel MacDonald. 2022. Principles of Economics 3e. Houston: OpenStax, Rice University. Section 6.1, Consumption Choices. Consumption Choices
-
Hansson, Sven Ove, and Till Grune-Yanoff. 2022. Preferences. In The Stanford Encyclopedia of Philosophy, substantive revision 2022 (first published 2006). Stanford University. Preferences
-
Hausman, Daniel M. 2011. Preference Satisfaction and Welfare. In Preference, Value, Choice, and Welfare, 77–87. Cambridge: Cambridge University Press. Online publication 2012. Preference Satisfaction and Welfare
-
Jensen, Robert T., and Nolan H. Miller. 2008. Giffen Behavior and Subsistence Consumption. American Economic Review 98 (4): 1553–1577. doi:10.1257/aer.98.4.1553. Giffen Behavior and Subsistence Consumption
-
Jevons, William Stanley. 1888 [1871]. The Theory of Political Economy. 3rd ed. London: Macmillan. Chapter III, Theory of Utility. Theory of Utility
-
Jevons, William Stanley. 1888 [1871]. The Theory of Political Economy. 3rd ed. London: Macmillan. Chapter IV, Theory of Exchange. Theory of Exchange
-
Kahneman, Daniel, Jack L. Knetsch, and Richard H. Thaler. 1990. Experimental Tests of the Endowment Effect and the Coase Theorem. Journal of Political Economy 98 (6): 1325–1348. doi:10.1086/261737. Experimental Tests of the Endowment Effect
-
Marshall, Alfred. 1920 [1890]. Principles of Economics. 8th ed. London: Macmillan. Book III, chapter III, Gradations of Consumers' Demand. Gradations of Consumers' Demand
-
Marshall, Alfred. 1920 [1890]. Principles of Economics. 8th ed. London: Macmillan. Book III, chapter VI, Value and Utility. Value and Utility
-
Menger, Carl. 1871. Principles of Economics. Translated by James Dingwall and Bert F. Hoselitz. Auburn, AL: Ludwig von Mises Institute, 2007. Chapter III, The Theory of Value. Principles of Economics
-
Slutsky, Eugen E. 1915. Sulla teoria del bilancio del consumatore. Giornale degli Economisti e Rivista di Statistica 51: 1–26. English translation by Olga Ragusa, On the Theory of the Budget of the Consumer, in Readings in Price Theory, edited by Kenneth E. Boulding and George J. Stigler, 27–56. Homewood, IL: Richard D. Irwin, 1952/1953 anthology edition; linked publisher reprint preview. Theory of the Budget of the Consumer
-
Smith, Adam. 1776. An Inquiry into the Nature and Causes of the Wealth of Nations. London: W. Strahan and T. Cadell. Book I, chapter IV, Of the Origin and Use of Money. Public-domain transcription. Wealth of Nations, Book I
-
U.S. Bureau of Labor Statistics. n.d. Consumer Expenditures and Income: Collections and Data Sources. Handbook of Methods. Accessed November 17, 2026. Consumer Expenditures and Income
-
Varian, Hal R. 1982. The Nonparametric Approach to Demand Analysis. Econometrica 50 (4): 945–973. doi:10.2307/1912771. Nonparametric Approach to Demand Analysis
-
Witt, Ulrich. 2016. The Transformations of Utility Theory: A Behavioral Perspective. Journal of Bioeconomics 18: 211–228. doi:10.1007/s10818-016-9235-6. Transformations of Utility Theory
-
World Bank. n.d. International Comparison Program: Methodology. Accessed November 17, 2026. International Comparison Program methodology