Marginal Utility

Marginal utility is the incremental change in total satisfaction or welfare (U) experienced by a decision-maker from consuming one additional unit of a good, service, or monetary wealth, holding the consumption of all other commodities constant. Formulated independently during the Marginal Revolution of the 1870s by William Stanley Jevons, Carl Menger, and Léon Walras—following earlier mathematical foundations by Daniel Bernoulli (1738) and Hermann Heinrich Gossen (1854)—the concept shifted business and economics theory from classical labor-cost theories of value to subjective marginalist valuation. Mathematically expressed as the partial derivative ∂U / ∂x_i of a differentiable utility function U(x₁, x₂, …, x_n), marginal utility provides the microeconomic foundation for downward-sloping supply and demand schedules, the equimarginal principle of consumer optimization, and expected-utility models of risk aversion.

Total Utility vs. Marginal Utility

Let a consumer's preferences over an n-dimensional commodity bundle x = (x₁, x₂, …, x_n) ∈ ℝ₊ⁿ be represented by a twice continuously differentiable utility function U(x). The marginal utility of commodity i, denoted MU_i, is the first partial derivative of total utility with respect to x_i:

Under the standard axiom of local non-satiation (more is preferred to less), marginal utility is positive across the relevant consumption domain (MU_i > 0) up to a satiation bliss point where MU_i = 0.

The Law of Diminishing Marginal Utility

Gossen's First Law, known universally as the Law of Diminishing Marginal Utility, states that as a consumer increases the consumption of any single commodity x_i within a fixed time interval while holding all other goods constant, the incremental utility gained from each successive unit decreases. In differential calculus, this corresponds to a negative second partial derivative:

For example, under a logarithmic utility specification U(x) = α · ln(x), marginal utility is MU(x) = α / x > 0, and its rate of change is ∂²U / ∂x² = −α / x² < 0. Psychophysically and behaviorally, consumers allocate their initial units of a resource toward their highest-priority physiological or subjective needs and assign subsequent units to progressively lower-priority uses.

For over a century after Adam Smith's The Wealth of Nations (1776), classical economists struggled to explain the paradox of value (the diamond–water paradox): why water, which is essential for human survival (possessing immense "value in use"), commands a near-zero market price, whereas diamonds, which serve primarily decorative functions (possessing minimal total "value in use"), command a high market price.

The marginalist synthesis resolved this paradox by distinguishing total utility (TU = ∫₀^Q MU(q) dq) from marginal utility (MU(Q)). Because fresh water is abundant in most temperate economies, aggregate consumption Q_water is very large, driving the marginal utility of the last liter consumed (MU_water(Q_water)) close to zero. Conversely, because gem-quality diamonds are geologically scarce, aggregate quantity Q_diamond is small, leaving the marginal utility of an incremental carat (MU_diamond(Q_diamond)) elevated. Because competitive market prices reflect marginal willingness to pay (P_i = MU_i / λ, where λ is the marginal utility of income) rather than total utility, diamonds command a high price while water commands a low price despite generating vastly greater consumer surplus.

Gossen's Second Law and the Equimarginal Principle

Consider a rational consumer maximizing U(x₁, …, x_n) subject to a linear budget constraint ∑_i=1ⁿ P_i x_i ≤ Y, where P_i > 0 are market prices and Y > 0 is nominal income (reflecting the opportunity cost of purchasing alternative bundles). Forming the Lagrangian function:

and setting the first-order partial derivatives ∂ℒ / ∂x_i = 0 for an interior solution yields Gossen's Second Law (the equimarginal principle):

Here, the Lagrange multiplier λ = ∂U^* / ∂Y represents the marginal utility of income—the shadow value of relaxing the budget constraint by one currency unit. At the optimum, the consumer allocates expenditure such that the last dollar spent on every commodity yields an identical increment of utility λ.

Cardinal vs. Ordinal Utility and the Marginal Rate of Substitution

In the early 20th century, Vilfredo Pareto (1906), Eugen Slutsky (1915), and John Hicks and R. G. D. Allen (1934) demonstrated that observable consumer demand behavior does not require cardinal utility (where numerical utility differences have absolute psychological meaning). Under ordinal utility theory, any monotonically increasing transformation V(x) = f(U(x)) with f′(U) > 0 represents the exact same preference ordering and generates identical demand functions and Pareto efficiency allocations.

While the individual marginal utility ∂V/∂x_i = f′(U) · MU_i changes under a nonlinear transformation, the ratio of marginal utilities—defined as the Marginal Rate of Substitution (MRS₁₂), or the absolute slope of an indifference curve—is strictly invariant to monotonic transformations:

Note that diminishing marginal utility (∂²U/∂x_i² < 0) is a cardinal property that is not invariant to positive monotonic transformations; the ordinal condition guaranteeing convex indifference curves and unique interior maxima is diminishing marginal rate of substitution (quasi-concavity of U).

Cardinal marginal utility remains indispensable in decision-making under uncertainty. In John von Neumann and Oskar Morgenstern's (1944) expected utility framework—foundational to game theory and modern portfolio theory—an agent evaluates risky lotteries by the expected value of a cardinal utility-of-wealth function E[u(W)]. Diminishing marginal utility of wealth (u″(W) < 0, strict concavity) implies risk aversion: via Jensen's inequality, E[u(W)] < u(E[W]), meaning a risk-averse consumer prefers a certain payout E[W] over a fair gamble with identical expected monetary value and is willing to pay a positive insurance risk premium. The degree of concavity is quantified by the Arrow–Pratt coefficient of absolute risk aversion:

In behavioral economics, Daniel Kahneman and Amos Tversky's (1979) prospect theory replaces total wealth W with gains and losses relative to a reference point, where diminishing sensitivity produces concave marginal valuation for gains and convex marginal valuation for losses, alongside loss aversion at the reference kink.

Can marginal utility ever be zero or negative?

Yes. As consumption of a commodity increases within a fixed period, marginal utility declines until it reaches zero at the satiation point (or bliss point), where total utility is at its local maximum. Consuming additional units beyond satiation (such as excessive food intake or industrial noise pollution) yields negative marginal utility (MU_i < 0), transforming the commodity from an economic "good" into an economic "bad" that reduces total utility and requires positive compensation to accept.

How does diminishing marginal utility explain progressive income taxation and insurance markets?

Under expected utility theory, if individuals exhibit diminishing marginal utility of income (u′(Y) = Y^−γ), the marginal utility of an additional dollar is substantially higher for a low-income household than for a high-income household. In utilitarian welfare economics, transferring a dollar from a high-income earner to a low-income earner increases aggregate social welfare (subject to labor-supply incentive constraints). In private markets, this same concavity explains why households pay insurance premiums exceeding expected actuarial losses to transfer wealth from high-income (no-loss) states to low-income (catastrophic-loss) states where marginal utility of wealth is elevated.

Why did economists transition from cardinal marginal utility to ordinal indifference curves for demand theory?

Cardinal utility cannot be observed directly by an external econometrician; only market choices (x^*) at given prices (P) and income (Y) are observable. Hicks and Allen proved that two consumers with different cardinal utility functions U(x) and V(x) = (U(x))³ exhibit identical Marginal Rates of Substitution (MU₁/MU₂) and identical price elasticity of demand, making ordinal preference rankings sufficient for deterministic consumer demand theory.

1. Jevons, W. S. (1871). The Theory of Political Economy. Macmillan and Co. https://doi.org/10.1057/9781137374158
2. Hicks, J. R., & Allen, R. G. D. (1934). "A Reconsideration of the Theory of Value." Economica, 1(1), 52–76. https://doi.org/10.2307/2548574
3. Pratt, J. W. (1964). "Risk Aversion in the Small and in the Large." Econometrica, 32(1/2), 122–136. https://doi.org/10.2307/1913738
4. Stigler, G. J. (1950). "The Development of Utility Theory. I." Journal of Political Economy, 58(4), 307–327. https://doi.org/10.1086/256962
5. Kahneman, D., & Tversky, A. (1979). "Prospect Theory: An Analysis of Decision under Risk." Econometrica, 47(2), 263–291. https://doi.org/10.2307/1914185