The substitution effect is the component of a price-induced change in demand that results from a change in relative prices, after compensating for the associated change in the consumer’s economic position. In microeconomics, it is distinguished from the income effect, which captures the response to the change in purchasing power. Under the standard model of consumer choice, an increase in a good’s own price produces a substitution effect that weakly reduces demand for that good; a decrease produces the reverse. The precise size of the effect depends on whether compensation holds utility constant or preserves the affordability of the original consumption bundle.
Economic meaning
A price change has two conceptually distinct consequences. It changes the terms on which one good can be exchanged for others, and it changes what the consumer can afford.
For example, if tea becomes cheaper while coffee and other prices remain unchanged:
- Tea becomes less expensive relative to other goods, encouraging substitution toward tea.
- The consumer can purchase the previous bundle with less money, increasing purchasing power.
The first response is the substitution effect. The second is the income effect. The observed change in tea purchases combines both, so it does not generally measure the substitution effect alone.
Substitution need not involve two obviously similar products. A lower transport fare can alter the allocation of spending among transport, housing, entertainment, and other goods. The relevant comparison concerns relative prices and alternative uses of a budget, not merely physical similarity.
The distinction between substitute and complementary goods also requires care. Goods can be classified according to their observed demand responses or their compensated demand responses; those classifications need not coincide because observed responses include income effects.
Budget constraints and indifference curves
Consider a consumer choosing quantities and , with prices and , and money income . The budget constraint is
With on the horizontal axis, the budget line has slope . A fall in , with income and unchanged, rotates the budget line outward.
Preferences are represented by a utility function, and an indifference curve contains bundles that provide the same level of satisfaction. Under differentiability and an interior optimum, the consumer chooses a bundle where the marginal rate of substitution equals the price ratio:
The substitution effect isolates the adjustment to the changed price ratio. Geometrically, it compares the original choice with a hypothetical choice on a budget line that has the new slope but an appropriately adjusted level of expenditure.
This compensated budget line is an analytical device. It does not imply that consumers actually receive compensation when prices change. Tangency is also not essential: corner solutions and kinked preferences can be analyzed using the underlying optimization problem.
Hicksian and Slutsky compensation
Two standard compensation rules yield different finite-change decompositions.
Hicksian substitution effect
The Hicksian approach, associated with John Hicks, holds the consumer’s original level of utility constant. At the new prices, income is adjusted to the minimum amount needed to attain that utility.
Let:
- and denote the initial and final price vectors;
- denote unchanged money income;
- denote initial utility;
- denote Marshallian demand for good ;
- denote Hicksian demand for good .
The Hicksian substitution effect is
The remaining income effect is
Therefore,
Graphically, the hypothetical compensated budget line has the new price ratio and supports the original indifference curve.
Slutsky substitution effect
The Slutsky approach adjusts income so that the consumer can still afford the original bundle at the new prices. If that bundle is , compensated income is
The Slutsky substitution effect is
The remaining income effect compares demand at the new prices with compensated income against demand at those prices with actual income.
The original bundle remains affordable under Slutsky compensation, but it need not remain optimal. By rearranging purchases, the consumer may attain higher utility than before. Hicksian compensation instead preserves utility exactly.
For finite price changes, the two measures generally differ. Under the usual smoothness and regularity conditions, they coincide to first order for an infinitesimal price change. Their development formed part of the twentieth-century formulation of demand theory in terms of ordinal preferences rather than measurable amounts of satisfaction.
The Slutsky equation
The Slutsky equation expresses the local decomposition mathematically. For demand for good and a change in the price of good ,
evaluated at , where is indirect utility.
The left-hand side is the observed price response at unchanged money income. The first term on the right is the compensated substitution response. The second term captures the purchasing-power effect of the price change.
For a good’s own price,
Under standard consumer-theory assumptions, the own-price compensated derivative satisfies
Thus, the substitution effect cannot by itself explain an increase in demand following a rise in the good’s own price.
More generally, where the required derivatives exist, the matrix of compensated price responses is symmetric and negative semidefinite. These restrictions are important in the estimation and testing of demand systems.
Normal goods, inferior goods, and Giffen goods
The total price response depends on how the substitution and income effects interact.
| Type of good | Income response | Effect of an own-price increase |
|---|---|---|
| Normal good | Demand rises as income rises | Substitution and income effects both reduce demand |
| Inferior good, without Giffen behavior | Demand falls as income rises | The income effect opposes but does not outweigh the substitution effect |
| Giffen good | Demand falls sufficiently strongly as income rises | The income effect outweighs the substitution effect, so demand rises |
For a normal good, a price increase reduces purchasing power, which further reduces demand. Both effects therefore support a downward-sloping demand curve.
For an inferior good, reduced purchasing power tends to increase demand, partially offsetting substitution away from the now more expensive good. If that opposing income effect is sufficiently strong, the good is a Giffen good.
Giffen behavior therefore does not reverse the compensated law of demand. It concerns the total, uncompensated response. Inferiority is necessary for Giffen behavior in the standard model, but most inferior goods are not Giffen goods.
Numerical example
Suppose a consumer has income and preferences represented by
Initially, both prices equal 1. The optimal bundle is
giving utility .
Now let the price of fall to , while the price of and money income remain unchanged. The new optimal bundle is
The total increase in demand for is 50 units.
Hicksian decomposition
To retain utility 50 at the new prices, the consumer minimizes expenditure subject to . The compensated bundle is
The Hicksian substitution effect on is approximately
and the income effect is approximately
Slutsky decomposition
At the new prices, the original bundle costs
With this income, the consumer chooses
The Slutsky substitution effect on is 25 units, and the remaining income effect is also 25 units.
Both decompositions explain the same total increase of 50 units. They divide it differently because preserving utility and preserving the affordability of the original bundle are different compensation rules.
Applications
Taxation and welfare analysis
A commodity tax changes relative prices as well as purchasing power. Compensated demand helps distinguish the behavioral response to the price distortion from the response to reduced economic resources.
This distinction is central to welfare economics and the analysis of deadweight loss. In a simple competitive market without pre-existing distortions, a small per-unit tax , starting from zero tax, produces an approximate deadweight loss of
when the producer price is fixed and the compensated demand slope can be treated as locally constant. More general settings require accounting for supply responses, other distortions, and distributional effects.
The substitution effect is therefore distinct from tax incidence, which concerns how a tax burden is distributed through changes in prices and incomes.
Labor supply
In the labor market, a higher wage raises the opportunity cost of leisure. Holding utility constant, the substitution effect encourages less leisure and more work.
At the same time, a higher wage can raise income. If leisure is a normal good, the income effect encourages more leisure and less work. The total labor-supply response can consequently be positive, zero, or negative.
This framework is used to analyze changes in wages and income taxes, although fixed work schedules, participation decisions, and constraints on working hours complicate the simple model.
Saving and intertemporal choice
An increase in the real interest rate makes future consumption cheaper relative to present consumption. The substitution effect favors future consumption, tending to reduce current consumption.
Its effect on observed saving also depends on income and wealth effects, including whether a household is a net borrower or saver. A higher return therefore does not imply a uniformly positive total saving response.
Production
An analogous distinction applies to firms choosing production inputs. Holding output fixed, a change in relative input prices encourages substitution toward relatively cheaper inputs. This is separated from the scale effect caused by a change in the firm’s chosen output.
For example, a higher wage may encourage a firm to use more machinery relative to labor, while the associated increase in production costs may also reduce output. The production substitution effect holds output, rather than consumer utility, constant.
Measurement and limitations
The substitution effect is not usually observable directly because ordinary price changes do not arrive with a precisely specified compensating transfer. Estimation therefore requires a demand model, information about income responses, or a research design capable of separating price and purchasing-power changes.
The distinction matters for price elasticity of demand. An uncompensated elasticity describes the response at fixed money income; a compensated elasticity describes the response at fixed utility. They answer different questions and need not have the same magnitude.
The standard framework also assumes that preferences remain stable over the comparison. Habit formation, adjustment costs, limited attention, and switching costs can make short-run observed responses differ from frictionless predictions. With perfect complements, the substitution effect may be zero; with perfect substitutes or discrete choices, demand may change discontinuously.
Finally, the compensation rule must be stated. For a substantial price change, “holding purchasing power constant” is not sufficiently precise: retaining the original utility level and retaining the ability to buy the original bundle generally produce different numerical substitution effects.