THEORY OF CONSUMER BEHAVIOR

The basis of the theory of consumer behavior rests upon the theory of demand. Theory of

demand begins with the ‘law of demand’ which states that the quantity demanded of a

commodity varies inversely with its price, ceteris paribus (keeping other factors unchanged,

e.g., taste, preference, income etc. of the consumer).

Now in the theory of consumer behavior, we assume the consumer is a rational individual who

wants to maximize her utility from a basket of consumable goods given her income unchanged.

There are two basic approaches in microeconomic theory for measuring the utility levels from

the commodity basket and they are: (1) The cardinalist approach and (2) The ordinalist

approach.

(1) THE CARDINAL UTILITY THEORY

Assumptions:

1. Rationality: The consumer is a rational individual who wants to maximize her total

utility subject to her budget constraint.

2. Cardinal Utility: The utility levels of commodities are CARDINALLY measurable

(in terms of money).

3. Diminishing marginal utility: The ‘law of diminishing marginal utility’ states that

the total utility of a commodity falls beyond certain ‘saturation’ point of consumption.

4. The total utility derived from a ‘basket of goods’ is directly related to the different

quantities consumed. Formally, Ui = f ( xi , m *, p* ) (i = 1,2,……,n)

 where Ui = Utility derived from i-th commodity,

xi= Quantity of i-th commodity,

p*= Prices of other commodities (constant)

with δUi / δxi > 0 and δ2Ui / δxi2 < 0.

The consumer’s equilibrium condition is expressed as:

MU1 / p1 = MU2 / p2 = …………… = MUn / pn in a ‘n’ – commodity world,

where MUi is the marginal utility of commodity i and pi is the price of commodity i.

(i = 1, 2, …, n)

Consumer Preferences

Consumer preferences are guided by two axioms and defined by the preference relation

‘≿’ based on rationality and they are:

(i) Completeness : For all x and y ϵ X, we have x ≿ y or y ≿ x (or both);

(By the notation x ≿ y, we mean that “x is at least good as y” or simply “x is

indifferent to y”)

(ii) Transitivity: For all x, y, and z ϵ X, if x ≿ y and y ≿ z, then x ≿ z.

In other words, if x is ‘weakly preferred’ to y and y is ‘weakly preferred’ to z,

then x is ‘weakly preferred’ to z.

The strong axiom of revealed preference theorem states that for all x, y and z ϵ X,

if we have x > y and y > z, then x > z.

Continuity

For all x. y ϵ X, the sets {x: x ≿ y} and {x: x ≾ y} are closed sets and {x: x ≻ y} and

{x: x ≺ y} are open sets. In other words, in the superset X, the indifferent choices are

closed sets and the strongly preferred choices are open sets.

Utility Functions

A function U: x → Ɍ is a utility function which represents preference relation ≿

(indifferent choices between goods) if, for all x, y ϵ X, U (x) ≥ U (y) with x ≿ y.

In a similar fashion, f (u(x)) ≥ f (u(y)) if and only if U(x) ≥ U(y).

The major assumptions related to the preferences are:

Weak monotonicity: If x ≥ y then x ≿ y.

Strong monotonicity: If x ≥ y then x > y given that x ≠ y.

Strong monotonicity obviously conforms to strong axiom of revealed preference theorem

where the preference for x will always supersede the preference for z provided that none

of the goods in the subset is an inferior good or a cause of externalities in consumption,

e.g., pollution.

Local nonsatiation

Given any x, y and ɛ ϵ X and ɛ > 0, then the Euclidian distance between x and y,

that is, |x - y| < ɛ such that y > x.

In other words, in the Euclidian set X, the preferential difference between commodity

bundles x and y must be less than the utility level of ɛ, that is, there is a probability that ɛ

may lie in a higher subset in the Euclidian set X.

Convexity

Given any x, y and z ϵ X such that x ≥ z and y ≥ z, then it follows that

tx + (1-t) y ≥ z for all 0 ≤ t ≤ 1 (t = scalar quantity).

The concept of convexity may be explained in this manner – in a finite Euclidian set X, if

two commodity bundles x and y are at least as good as commodity z, then in the finite

budget space of a consumer, the sum of respective shares of income spent on two bundles

x and y would yield the same utility from commodity bundle z.

Strict Convexity

Given x, y and z ϵ x with x ≥ z, y ≥ z and x ≠ y, then tx + (1-t)y > z for all 0 < t < 1 .

In the case of strict convexity , within the finite Euclidian set X , if two commodity

bundles x and y are at least as good as commodity bundle z, then in the finite budget

space of a consumer, the sum of respective shares of income spent on two bundles x

and y would yield more utility from commodity bundle z.

(2) Ordinal utility theory and indifference curve

The concept of utility function and the related assumptions discussed earlier jointly

form the framework of ordinal utility theory and the concept of indifference curve.

An indifference curve is a locus of points within the finite budget space of a rational

consumer in a two-commodity world which yields constant utility.

The indifference curve is negatively sloped and convex to the origin. Convex

preferences by a consumer may exhibit ‘flat spots’ on the curve but strictly convex

preferences exhibit a rotund shape.

To put it in a very simple way, the equation of the indifference curve is given by:

U¯ = f (x, y) where U = U¯ = Utility level (fixed).

By taking total differential, we have:

fx dx + fy dy = dU¯ = 0

→ dy / dx|IC = - fx / fy < 0 as fx, fy > 0.

In ordinal utility theory, a rational consumer may have a ‘two-fold’ objective:

(i) Either to maximize her utility function derived from two commodity bundles x

and y ( x, y ϵ X) subject to her budget constraint,

Or

(ii) To minimize her budget (within the finite budget space) subject to her utility

constraint derived from two commodity bundles x and y (x, y ϵ X).

For the first case, the objective function of the consumer may be expressed as:

Maximize V = f (x, y) subject to m ≥ pxx + pyy

where m= Income of the consumer,

px = Price of commodity bundle x,

py = Price of commodity bundle y.

The required Lagrangian equation would be:

ᴌ = f(x, y) + λ [m¯ - pxx – pyy ] ( λ ≥ 0 ) ( λ = Lagrangian multiplier)

The first order conditions require:

δ ᴌ / δ y = fx – λpx = 0 …. (1)

δ ᴌ / δ y = fy – λpy = 0 …. (2)

δ ᴌ / δ λ = m¯ - pxx – pyy = 0 …. (3)

From equations: (1) and (2), we have:

fx / fy = px / py

→ MRSx,y = fx / fy = px / py ………….. (4)

where MRSx,y = Marginal rate of substitution between bundles x and y.

Equation: (4) gives us the consumer’s equilibrium condition.

The second order condition for utility maximization requires the bordered Hessian

determinant (Ĥ) to be positive, that is,

Ĥ = fxx fxy -px

fyx fyy -py > 0 ……………………. (5)

- px - py 0

By expanding equation: (5) we have:

2 fxy fx fy - fxx py2 - fyy px2 > 0 ………………….. (6)

Finally by substituting the values of fx and fy from equation: (4) into equation: (3), we

get the Ordinary or Marshallian demand functions as x (px, m) and y (py, m).

In the second case, the objective function of the consumer may be expressed as:

Minimize Ζ = pxx + pyy subject to U = f (x, y) = U̅ (fixed).

The required Lagrangian equation would be:

Ψ = pxx + pyy + λ [U̅ - f (x, y)]

The required first order conditions require:

δΨ / δx = px – λfx = 0 ……………………………. (7)

δΨ / δy = py – λfy = 0 ……………………………. (8)

δΨ / δλ = U̅ - f (x, y) = 0 ………………………… (9)

Equations: (7) and (8) give us the similar equilibrium condition as shown in equation:

(4) as shown in equation: (4), i.e.,

MRSx,y = fx / fy = px / py.

However the demand functions derived for x and y would be somehow different from

Marshallian demand functions and they are known as Compensated or Hicksian

demand functions.

Illustration:

Let us take a Cobb - Douglas utility function as:

U = xα y 1 - α (α < 1)

The consumer’s equilibrium condition requires:

MRSx,y = px / py

Thus MRSx,y = fx / fy = α xα – 1 y1 – α / (1 – α) xα y –α = (α / 1 – α) * (y / x)

Hence (α / 1 – α) * (y / x) = px / py gives us:

x = (α / 1 – α) (pyy / px) and

y = (α / 1 – α) (pxx / py).

By substituting the values of x and y in the budget equation, we get:

m̅ = pxx / α

→ x* = Optimal Ordinary or Marshallian demand function for x = α m̅ / px,

and y* = Optimal Ordinary or Marshallian demand function for y = (1 – α) m̅ / py.

Thus x* = x (px, m) and y* = y (py, m).

Similarly, if we try to derive the Compensated or Hicksian demand function, we

get:

px / py = (α / 1 – α) * (y / x)

→ pxx = (α / 1 – α) pyy and pyy = (1 – α / α) pxx

Hence y = (α / 1 – α) * (pxx / py)

By substituting the value of y in the utility constraint equation, we get:

U̅ = xα [(α / 1 – α) * (pxx / py)]1 – α

→ U̅ = [(α / 1 – α) * (px / py)]1 – α x

→ x* = [(1 – α) ∕ α]1 – α (py / px)1 – α U̅ = k (py / px)1 – α U̅

where k = [(1 – α) ∕ α]1 – α = constant, and

x* = Optimal Compensated or Hicksian demand function for x = x (px, py, U̅ ).

Similarly, we derive:

y* = (1 – α / α)α (px / py)α U̅ = kʹ (px / py)α U̅ = kʹ (px / py)α U̅

where kʹ = (α ∕ 1 – α)α = constant, and

y* = Optimal Compensated or Hicksian demand function for y = y (px, py, U̅ ).

 Good 2

Maximizes utility

Minimizes expenditure

Good 1

Figure 1

Corner solutions for indifference curve approach

Good 2

IC

Good 1

Figure 2 (a): When the consumer consumes good 1 only.

 Good 2

IC

Good 1

Figure 2 (b): When the consumer consumes good 2 only.

Indirect utility

Properties of the indirect utility function:

(1) If v (p, m) is the indirect utility function, then it is nonincreasing in p, that is,

if pʹ > p, then v (pʹ, m) ≤ v (p, m). Similarly, v (p, m) is nondecreasing in m;

(2) v (p, m) is homogeneous of degree zero in (p, m) space;

(3) v (p, m) is quasiconvex in p, that is, {p: v (p, m) ≤ k} is a convex set for all

values of k;

(4) v (p, m) is continuous at all p >> 0, m > 0 (monotonic increase in p).

Properties of the expenditure function:

(1) If e (p, U) is an expenditure function, it is nondecreasing in p,

(2) e (p, U) is homogeneous of degree 1 in p,

(3) e (p, U) is concave in p,

(4) e (p, U) is continuous in p for p >> 0 (monotonic increase in p),

(5) If g (p, U) is the expenditure – minimizing bundle necessary to achieve utility U at prices

pi, then gi (p, U) = δ e (p, U) / δ pi where i = 1, 2, …, n assuming pi > 0.

Roy’s Identity

If x (p, m) is the Marshallian demand function, then xi (p, m) = - δ v (p, m)/ δ pi ∕ δ v (p, m)/ δ pi

for i = 1, 2, ….., n with pi, m > 0.

Proof:

The indirect utility function is given by v (p, m) ≡ U (x (p, m)) ………………. (10)

Now by differentiating equation: (10) with respect to price pi, we get:

δ v (p, m) ∕ δ pi = ∑i = 1n δ U (x) ∕ δ xi / δ xi ∕ δ pi

Again we know that x (p, m) is the Marshallian demand function which satisfies the first order

condition for utility maximization. The first order condition for utility maximization requires:

δ v (p, m) ∕ δ pi = λ ∑i = 1n pi (δ xi ∕ δ pi) …………………………… (11)

The demand functions also satisfy the budget constraint p x (p, m) ≡ m. Thus by differentiating

the budget equation with respect to price pi, we get:

xi (p, m) + ∑i = 1n pi (δ xi ∕ δ pi) = 0

→ ∑i = 1n pi (δ xi ∕ δ pi) = - xi (p, m) ………………………….. (12)

By substituting equation: (12) in equation: (11) we get:

δ v (p, m) ∕ δ pi = - λ xi (p, m) ………………………. (13)

Now by differentiating equation: (10) with respect to m, we get:

δ v (p, m) ∕ δm = λ ∑i = 1n pi (δ xi ∕ δm) ……………….. (14)

Now the idea of budget constraint gives us:

∑i = 1n pi (δ xi ∕ δm) = 1

Thus we can rewrite equation: (14) as:

δ v (p, m) ∕ δm = λ ……………………… (15)

By substituting equation: (15) into equation: (13), we get:

δ v (p, m) ∕ δ pi = [δ v (p, m) ∕ δm] * xi (p, m)

→ xi (p, m) = δ v (p, m) ∕ δ pi ∕ δ v (p, m) ∕ δm (Proved)

The term ‘λ’ is simply the marginal utility of money.

The money metric utility functions

The money metric utility function may be defined as the utility function which yields the

consumer same or more utility at price p for a particular bundle of good z at minimum expenses

compared to purchasing a commodity bundle x at the same price p.

It is otherwise known as minimum income function.

Formally, the optimization problem may be expressed as:

Minimize pz subject to U (z) ≥ U (x).

Good 2

m (p, x) ∕ p2 z

Good 1
m (p, x) ∕ p1

Figure 3

Money metric direct utility function

The money metric direct utility function may be expressed as:

m (p, x) ≡ e (p, U (x))

Now when x = x̅ (fixed), then U (x) = U (x̅ ) (fixed).

Thus with fixed prices, the direct money metric utility function m (p, x) is simply a

monotonic increasing utility function.

The money metric indirect utility function may be defined as the utility level derived

from two different prices p and pʹ with a given income m.

Formally, ɸ (p, pʹ, m) = e (p, v (pʹ, m)).

Good 2

ɸ (p, pʹ, m) ∕ p2 Optimal bundle at price p and income ɸ (p, pʹ, m)

Optimum bundle at price pʹ and income m

(pʹ, m) Good 1
Figure 4

Illustration:

Let a Cobb – Douglas utility function be:

U (x1, x2) = x1α x21 – α (α < 1)

Since the monotonic transformation of the above utility function does not change the

preferences of the consumer, hence it may be written as:

U (x1, x2) = α log x1 + (1 – α) x2.

The objective function would be to:

Maximize U (x1, x2) subject to m = p1 x1 + p2 x2.

The necessary Lagrangian equation would be:

Ɣ = α log x1 + (1 – α) x2 + λ [m̅ - p1 x1 - p2 x2]

The required first order conditions are:

α ∕ x1 = λ p1 and 1 – α ∕x2 = λ p2

→ α / p1 x1 = 1 – α ∕ p2 x2

Now by substituting p2 x2 = (1 – α / α) p1 x1 in the budget equation, we get:

m = p1 x1 ∕ α → x1* = x1 (p1, m) = α * m ∕ p1.

Similarly, x2* = x2 (p2, m) = (1 – α) * m ∕ p2.

The indirect utility function may be derived by substituting the Marshallian or

ordinary demand functions in the original utility function like below:

v (p1, p2 , m) = α log (α * m ∕ p1) + (1 – α) log [(1 – α) * m ∕ p2]

= α [log α + log m – log p1] + (1 – α) [log (1 – α) + log m – log p2]

= k + log m - α log p1 – (1 – α) log p2

 where k = α log α + (1 – α) log (1 – α) = constant.

Thus v (p1, p2 , m) = log m - α log p1 – (1 – α) log p2 ……………… (I)

(The constant term ‘k’ is omitted in the functional form)

Different elasticities of demand

Own price elasticity of demand

The own price elasticity of demand of a commodity may be defined as the ratio of

proportionate change in quantity demanded of a commodity X1 to the proportionate

change in its price p1. In other words, the own price elasticity of demand measures the

percentage change in consumption of X1 due to 1 percent change in its price.

Formally, it may be expressed as:

e11 = δ x1 ∕ x1 ∕ δ p1 ∕ p1 = (p1 / x1) * (δ x1 / δ p1)

Goods with very large values of elasticities (|e11| > 1) are called luxuries and goods

with very small values of elasticities (|e11| < 1) are called necessities.

Cross price elasticity of demand

The cross price elasticity of demand of a commodity X2 may be defined as the ratio of

proportionate change in its quantity demanded to the proportionate change in the price

of commodity X1, that is, p1 (for a two – commodity world).

In other words, the cross price elasticity of demand of a commodity X2 measures the

percentage change in its consumption due to 1 percent change in the price of

commodity X1.

Formally, it may be expressed as:

ε21 = (p1 / x2) * (δ x2 / δ p1)

The definition of cross price elasticity of demand may be extended to a Euclidian set

X as below:

εij = (pj / xi) * (δ xi / δ pj)

(i = 1, 2,…, m; j = 1, 2, ….., n; i ≠ j)

Cournot’s Aggregation Condition

In a two – commodity world, let the budget equation of a rational consumer be:

p1 x1 + p2 x2 = m …………….. (i)

By taking total differential of equation: (i), we get:

p1 dx1 + x1 dp1 + p2 dx2 + x2 dp2 = dm ……….. (ii)

Now, in the short run, we assume dp2 = dm = 0 and thus equation: (ii) reduces to:

p1 dx1 + x1 dp1 + p2 dx2 = 0

→ x1 dp1[1 + (p1 dx1 ∕ x1 dp1)] + (p2x2 / p1) dp1 [p1 dx2 ∕ x2 dp1] = 0

→ x1 [1 + (p1 dx1 ∕ x1 dp1)] + (p2x2 / p1) [p1 dx2 ∕ x2 dp1] = 0

→ x1 [1 + e11] + (p2x2 / p1) ε21 = 0

→ (1 + e11) + (p2x2 / p1x1) ε21 = 0

→ e11 + α2 ε21 = -1 …………………… (iii)

where α2 = (p2x2 / p1x1) = Ratio of relative shares of good 2 and 1.

Equation: (iii) is known as the Cournot’s aggregation condition.

Income elasticity of demand

The income elasticity of demand for a commodity X may be defined as the ratio of

proportionate change in quantity demanded of good X to proportionate change in the

income m of the consumer.

In other words, the income elasticity of demand measures the percentage change in

the consumption of a commodity X due to 1 percent change in the income m of the

consumer.

Formally, it may be expressed as:

η = m δx1 ∕ x1 δm

For normal good, η > 0; and for inferior good, η < 0.

Engel’s Aggregation Condition

We now return to the two – commodity world budget equation as:

p1 x1 + p2 x2 = m

By assuming p1 and p2 unchanged in the short run, and by taking the total differential

of the equation above, the following equation is derived as:

p10 dx1 + p20 dx2 = dm

→ (p10 x1 ∕ m) * (m dx1 ∕ x1 dm) + (p20 x2 ∕ m) * (m dx2 ∕ x2 dm) = 1

→ β1η1 + β2η2 = 1 …………………………. (iv)

where β1 = (p10 x1 ∕ m) = Relative share of good X1 in income m,

β2 = (p20 x2 ∕ m) = Relative share of good X2 in income m,

η1 = Income elasticity of demand for good X1, and

η2 = Income elasticity of demand for good X2.

Equation: (iv) is known as the Engel’s aggregation condition.

Engel’s aggregation condition also implies that in a two – commodity world, both the

goods cannot be inferior simultaneously.

Now by adding equation: (iii) and equation: (iv) and rearranging, we get:

e11 + α2 ε21 + β1η1 + β2η2 = 0

→ α1 e11 + α2 ε21 + β1η1 + β2η2 = 0 ………………….. (v)

where α1 = 1, α2 = p2x2 / p1x1, β1 = p10 x1 ∕ m, and β1 = p20 x2 ∕ m are all scalars.

From equation: (v), it may be stated that in a finite set X and within a closed sub set

(pi, xi) with i = (1, 2); the own price elasticity of demand, the cross price elasticity of

demand and the income elasticity of demand are linearly dependent.

Income consumption curve

Income consumption curve may be defined as a locus of utility maximizing

commodity bundles with respect to changes in the income levels of a consumer.

It is otherwise known as Income Expansion Path [Figure 5 (a)].

Good 2

Income consumption curve

Good 1
Figure 5 (a)
The income consumption curve is a straight line through the origin if the income

elasticity of demand is 1.

Proof:

Let the budget equation of the consumer in a two – commodity world be:

p1̅ x1 + p2̅ x2 = m (where p1̅ = p2̅ = fixed)

Hence p1̅ dx1 ∕ dm = 1 → dx1 ∕ dm = 1 ∕ p1̅

→ (m ∕ x1) * (dx1 ∕ dm) = η1 = m ∕ p1̅ x1

Similarly, (m ∕ x2) * (dx2 ∕ dm) = η2 = m ∕ p2̅ x2

Now η1 = η2 = 1gives us: m = p1̅ x1 → x1* = (1 ∕ p1̅) * m and

m = p2̅ x2 → x2* = (1 ∕ p2̅ ) * m → which are equations of straight lines through the

origin.
Good 2

Income Consumption Curve

Good 1

Figure 5 (b)

Income consumption curve for good 2 (luxurious good)

Good 2

Good 1

Figure 5 (c)

Income consumption curve for good 2 (inferior good)

Now let us repeat the previous budget equation with expenditure on good 1 fixed and

try to evaluate the impact of price change on good 2.

Thus we may write:

p1̅ x1̅ + p2 x2 = m (p1̅ x1̅ = Expenditure on good 1 = Fixed)

By taking total differential of the equation above, we get:

p2 dx2 + x2 dp2 = dm

→ p2 dx2 [1 + (x2 dp2 ∕ p2 dx2)] = dm

→ 1 + [1 ∕ (p2 dx2 ∕ x2 dp2)] = dm ∕ p2 dx2

→ 1 + [1 ∕ e2] = dm ∕ p2 dx2

[Since p2 dx2 ∕ x2 dp2 = e2 = Price elasticity of demand for good 2]

→ 1 + [1 ∕ e2] = m ∕ p2 x2 [1 ∕ (m dx2 ∕ x2dm)]

→ 1 + (1 ∕ e2) = m ∕ p2 x2 [1 ∕ η2]

[Since η2 = m dx2 ∕ x2dm = Income elasticity of demand for good 2]

→ 1 ∕ e2 = m ∕ p2 x2 η2 – 1

→ (m - p2 x2 η2) ∕ p2 x2 η2 = 1 ∕ e2

→ p1̅ x1̅ + p2 x2 - p2 x2 η2 ∕ p2 x2 η2 = 1 ∕ e2 [Since m = p1̅ x1̅ + p2 x2]

→ (p1̅ x1̅ ∕ p2 x2 η2) + (1 - η2) ∕ η2 = 1 ∕ e2

→ k ∕ η2 + 1 ∕ η2 – 1 = 1 ∕ e2 [Where k = p1̅ x1̅ ∕ p2 x2 = constant]

→ η2 = (1 + k) e2 ∕ 1 + e2 → η2 = (kʹ e2) ∕ 1 + e2 ……………………….. (vi)

[Where kʹ = (1 + k) = (p1̅ x1̅ + p2 x2) ∕ p2 x2 = constant]

Equation: (vi) establishes the relationship between the income elasticity of demand

and the own price elasticity of demand for a particular commodity.

A Giffen good is an ultra inferior good because it helps to decrease the real income of

a consumer to a great extent.

Engel curve

Engel curve may be defined as a locus of optimum bundles of consumption of a

particular commodity with respect to changed income levels of a consumer.

Income

Engel curve

Good 1

Figure 5 (d)
Price consumption curve

Price consumption curve is a locus of optimum bundles of consumption of a particular

commodity with respect to changes in its price levels keeping the price of the other

commodity unchanged.

Good 2

Price consumption curve

Good 1

Figure 5 (e)

Price consumption curve in case of normal good

 Good 2

Price consumption curve

Good 1

Figure 5 (f)

Price consumption curve in case of inferior good

Types of utility functions

Separable and additive utility functions

Utility functions are assumed to be strictly quasi – concave, increasing and differentiable

in nature.

A utility function which is strongly separable may be expressed as:

U = F [∑i = 1n fi (qi)] with δU ∕ δF, δU ∕ δfi > 0.

 A utility function which is weakly separable may be expressed as:

U = F [f1 (q1, q2, ……, qk) + fn – 1 (qk + 1, ………….., qn)]

 A utility function which is strongly additive may be expressed as:

U = ∑i = 1n fi (qi)

 A utility function which is weakly additive may be expressed as:

U = f1 (q1, q2, ……, qk) + fn – 1 (qk + 1, ………….., qn)

Homogeneous, Homothetic and Leontief utility functions

A utility function is homogeneous of degree n if

f (tq1, …………, tqn) = tn f (q1, q2, ……, qn)

 When n = 1, there is a constant elasticity of substitution between the goods in a

two – commodity world.

 A function g (x) is termed to be a positive monotonic transformation if

g (x) > g (y) for x > y.

A utility function is homothetic in nature if and only if f (q) = g (h (q)) where h (∙) is

homogeneous of degree 1 and g (∙) is a monotonic function.

Good 2 f (q) = y

2q
yy
f (q) = 2y

q 2qʹ
q
qʹ

Good 1

Figure 6 (a)

Homogeneous utility function

 Good 2

2q

2qʹ f (q) ≠ 2y

qʹ f (q) = 2y

Good 1

Figure 6 (b)

Homothetic utility function

Leontief utility function

The Leontief utility function may be expressed as:

U = min [x1 ∕ α, x2 ∕ β] with α, β > 0.

 It is an L – shaped utility / production function,

 The commodities / inputs in the Leontief utility / production function are

extremely rigid in consumption / production,

 The commodities / inputs are complements to each other.

 Good 2

Good 1

Figure 7

Leontief utility function

Elasticity of substitution

The elasticity of substitution may be defined as the ratio of proportionate change in the input

ratio to the proportionate change in the ratio of Marginal rate of technical substitution (MRTS) of

two inputs of production (for a production function). It may be also defined as the ratio of

proportionate change in the consumption ratio of two goods to the ratio of proportionate change

in the ratio of Marginal rate of substitution (MRS) between the two goods (for a utility

function).

In case of production function:

σ = Elasticity of substitution between labor and capital

= d (K ∕ L) ∕ (K ∕ L) ∕ d (fL ∕ fK) ∕ (fL ∕ fK)

= (fL ∕ fK) ∕ (K ∕ L) * d (K ∕ L) ∕ d(fL ∕ fK)

 where L = Labor, K = Capital, fL = Marginal productivity of labor, and

fK = Marginal productivity of capital.

In case of utility function:

σ = Elasticity of substitution between Good 1 and Good 2

= d (x2 ∕ x1) ∕ (x2 ∕ x1) ∕ d (f1 ∕ f2) ∕ (f1 ∕ f2)

= (f1 ∕ f2) ∕ (x2 ∕ x1) * d (x2 ∕ x1) ∕ d (f1 ∕ f2)

where x1 = Quantity of good 1, x2 = Quantity of good 2,

f1 = Marginal utility of good 1, and f2 = Marginal utility of good 2.

We know d (x2 ∕ x1) = x1 dx2 – x2 dx1 ∕ x12 and

d (f1 ∕ f2) = δ ∕ δ x1 * (f1 ∕ f2) dx1 + δ ∕ δ x2 * (f1 ∕ f2) dx2

Also dx2 = - (f1 ∕ f2) dx1 [Since f1 dx1 + f2 dx2 = 0]

Thus σ = f1 (f1 x1 + f2 x2) ∕ f2 x1 x2 [f1 δ (f1 ∕ f2) ∕ δ x2 – f2 δ (f1 ∕ f2) ∕ δ x1]

= f1 f2 (f1 x1 + f2 x2) ∕ x1 x2Ɗ

where Ɗ = 2f12 f1 f2 - f12f22 – f22f11 > 0

[Ɗ = The bordered Hessian determinant]

Since we know that σ = f1 f2 (f1 x1 + f2 x2) ∕ x1 x2Ɗ; hence we consider three different cases with

three different utility functions:

First case

Let the utility function be of Cobb – Douglas form: U (x1, x2) = x1α x21 – α (α < 1)

Thus f1 = δ U ∕ δ x1 = α x1α - 1 x21 – α and f2 = δ U ∕ δ x2 = α x1α x2 – α.

We know Ɗ = 2f12 f1 f2 - f12f22 – f22f11

So f11 = α (α – 1) x1α - 2 x21 – α, f12 = α (1 – α) x1α - 1 x2 – α and f22 = α (α - 1) x1α x2 – (1 + α).

By substituting the derived values of f11, f12 and f22 in the formula of elasticity of substitution, we

get σ = 1.

So for a utility function in the Cobb – Douglas form, the elasticity of substitution is 1 and hence

the utility function takes the shape of a straight line through the origin.

Proof:

σ = 1 gives us: (f1 ∕ f2) ∕ (x2 ∕ x1) * d (x2 ∕ x1) ∕ d (f1 ∕ f2) = 1

Now let f1 ∕ f2 = F and x2 ∕ x1 = X and hence the above equation may be rewritten as:

(F ∕ X) * dX ∕ dF = 1

→ dX ∕ X = dF ∕ F

By integrating both sides, we get:

log X = log cF [c = Constant of integration]

→ X = cF → This is the equation of a straight line through the origin.

Second case

Utility function in the linear equation form: U = x1 + x2

Here f1 = f2 = 1, hence f1 ∕ f2 = 1 and d (f1 ∕ f2) = 0.

So in this case, σ → ∞ and the utility function is horizontal to the X – axis and infinitely elastic.

Third case

Utility function in the product form: U = x1 x2 where x2 = k x1 (k ≠ 0).

Thus x2 ∕ x1 = k and d (x2 ∕ x1) = 0.

Here σ = 0 and the utility function is parallel to the Y – axis and perfectly inelastic.

Good 2

U (x1, x2)

Good 1

Case 1

The utility function is a straight line through the origin (σ = 1)

Good 2

U (x1, x2)

Good 1

Case 2

The utility function is infinitely elastic (σ = ∞)

Good 2

U (x1, x2)

Good 1

Case 3

The utility function is perfectly inelastic (σ = 0)

Compensating variation of income method for isolation of Income Effect (IE)

and Substitution Effect (SE)

The Price Effect (PE) on a particular commodity (due to a change in its price) may be

decomposed into two parts:

(a) Income Effect (IE), and

(b) Substitution Effect (SE).

Formally, PE = IE + SE.

The substitution effect may be defined as the change in quantity demanded of a particular

commodity due to a change in its price after adjusting the income of the consumer in such a way

so that her real purchasing power is unchanged.

The income effect may be defined as the change in the consumption level of a particular

commodity due to the change in the real income of the consumer. The substitution effect and

income effect are both negative in nature.

Figures 8 (a), 8 (b) and 8 (c) explain the compensating variation of income method for

isolation of substitution and income effect for normal, inferior, and Giffen good respectively by

Hicksian method.
 Good 2

2
A

e3

e2 e1

B1 D B Good 1

Figure 8 (a)

Let the consumer initially consumes at the tangency point e1 (tangent between the budget line

AB and indifference curve 1). Now if the price of good 1 increases, the initial budget line AB

will shift leftward to AB1 and the consumer is forced to curtail her consumption of good 1 and

now she consumes at tangency point e2 (tangent between the shifted budget line AB1 and

indifference curve 2). Now if she desires to maintain her initial utility level from good 1

unchanged, she will have to consume at tangency point e3 (tangent between an imaginary or

compensated budget line CD drawn parallel to AB1 and indifference curve 1) where her initial

utility level will be restored as she will be on her initial indifference curve once again.

The movement from e1 to e3 is the substitution effect and the movement from e2 to e3 is the

income effect. The entire range of movement from e1 to e2 is the price effect.
In case of normal good [Figure 8 (a)], the stronger substitution effect outweighs the weaker

income effect.

Good 2

e3

e1

e2

Good 1
B1 D B

Figure 8 (b)

According to figure: 8 (b), the movement from e1 to e3 is the substitution effect and the

movement from e2 to e3 is the income effect. The entire range of movement from e1 to e2 is the

price effect.

In case of inferior good [Figure 8 (b)], the stronger income effect outweighs the weaker

substitution effect.
 Good 2

C Price Consumption Curve

B1 D B Good 1

Figure 8 (c)

In case of a Giffen good, an extremely strong income effect outweighs the weak substitution

effect so that the Price Consumption Curve (red dashed line) bends backwards [Figure 8 (c)].

For a Giffen good, the real income of the consumer falls due to continual increase in the price of

a commodity (Good 1 in figure).

The Slutsky Equation

The Slutsky equation helps us to segregate the substitution effect and income effect with respect

to changes in price and income respectively.

Formally, it may be expressed as:

δ xi (p, m) / δ pi = δ hi (p, v (p, m)) ∕ δ pi – [δ xi (p, m) / δm * xi (p, m)]

 where δ xi (p, m) / δ pi = Price effect,

δ hi (p, v (p, m)) ∕ δ pi = Substitution effect, and

δ xi (p, m) / δm * xi (p, m) = Income effect.

Proof:

Let x* be the utility maximizing quantity at optimal price – quantity combination (p*, m*) and let

U* = U (x*).

Hence we can write: hi (p, U*) ≡ xi (p, e (p, U*)) where e (p, U*) = Indirect utility function.

Now by differentiating the above identity with respect to price pi and estimate the result at

optimal values of p* and m*, we get:

δ hi (p*, U*) ∕ δ pi = δ xi (p*, m*) ∕ δ pi + [δ xi (p*, m*) ∕ δm] * [δ e (p*, U*) ∕ δ pi]

Now by rearranging we get:

δ xi (p*, m*) ∕ δ pi = δ hi (p*, U*) ∕ δ pi - [δ xi (p*, m*) ∕ δm] * [δ e (p*, U*) ∕ δ pi]

→ δ xi (p*, m*) ∕ δ pi = δ hi (p*, U*) ∕ δ pi - [δ xi (p*, m*) ∕ δm] * xi*

Since xi = e (p, U (p, m)), hence δ xi ∕ δ pi = δ e (p*, U*) ∕ δ pi = xi* (Proved).

The substitution effect is derived by keeping the prices of other commodities unchanged and

the income effect is derived by keeping the utility level of the consumer unchanged.
The theory of revealed preference

The theory of revealed preference attempts to rationalize a consumer’s preference pattern within

a given ‘basket’ of goods with varying utility levels.

First, let us assume there are commodity bundles xi available at prices pi (i = 1, 2,…, n).

Now for a commodity bundle x, if U (xi) ≥ U (x), then pi xi ≥ pi x condition states that the

commodity bundle xi is strictly revealed preferred to commodity bundle x and formally it may

be expressed as xi PD x.

Now if we have a preferential sequence based on the revealed preference of commodity bundles

available such that xi RD xj, xj RD xk, …… , xn RD x, then in this case xi would be revealed

preferred to x and it would be formally expressed as xi R x (The symbol RD stands for directly

revealed preferred. Hence xi RD xj should be read as bundle xi is directly revealed preferred to

bundle xj).

Generalized Axiom of Revealed Preference (GARP)

If xi is revealed preferred to xj, then xj cannot be strictly directly revealed preferred to xi.

In other words, for a price vector pi, pi xj ≤ pi xi.

Weak Axiom of Revealed Preference (WARP)

If xi RD xj and xi ≠ xj, then xj is not directly revealed preferred to xi.

Strong Axiom of Revealed Preference (SARP)

If xi is revealed preferred to xj, xj is revealed preferred to xk, xk is revealed preferred to xn, then

xn must never be revealed preferred to xi (xi ≠ xj ≠ xk ≠ ….. ≠ xn).

The SARP ensures the transitivity axiom of revealed preference, but it is stronger than the usual

transitivity axiom.

Afriat’s theorem

Let (pi, xi) for i = 1, 2, ….., n be a finite vector space for price – quantity combinations.

Then the following conditions hold:

1. There exists a locally nonsatiated utility function that will rationalize the consumer’s

preferences,

2. The price – quantity combination that maximizes the utility level of the consumer will

also satisfy the GARP,

3. There exists positive vector combinations (Ui, λi) for i = 1, 2, …., n; that will satisfy the

Afriat inequality:

Uj ≤ Ui + λi pi (xj - xi) for all i, j (i ≠ j).

In other words, (Ui – Uj) + λi pi (xj - xi) ≥ 0.

The consumer will always find herself in a nonsatiated situation when Ui ≠ Uj and xj ≠ xi.
 Thus the condition for nonsatiation may be written as:

|Ui – Uj| + λi pi |xj - xi| > 0.

When Ui = Uj and xj = xi, the above inequality becomes zero (since λi, pi ≠ 0) and the

consumer reaches her satiation level,

4. There exists a locally nonsatiated, continuous, concave and monotonic utility function

that will rationalize the consumer’s preferences.

The Substitution Effect is negative

Proof:

Let a consumer consumes a bundle x1 at price p1 and she is indifferent between bundles x1 and x2

at price vector p1 in a finite set of price – quantity vectors (pi, xi) for i = 1, 2, …., n.

Now as per assumption, she is indifferent between x1 and x2 and decides to purchase x1, then we

can write:

p1 x1 ≤ p1 x2 …………………….. (I)

Again, if she decides to purchase bundle x2 at price p2, then it implies that x1 must not be cheaper

than the bundle x2 at price p2, which gives us:

p2 x2 ≤ p2 x1 ……………………… (II)

From equation: (I) we have:

p1 x1 - p1 x2 ≤ 0

→ p1 (x1 – x2) ≤ 0,
 → - p1 (x2 – x1) ≤ 0 ………………….. (III)

From equation: (II) we have:

p2 x2 – p2 x1 ≤ 0

→ p2 (x2 – x1) ≤ 0 ……………………... (IV)

Now by adding equation: (III) and (IV), we get:

p2 (x2 – x1) - p1 (x2 – x1) ≤ 0

→ (p2 – p1) (x2 – x1) ≤ 0 ……………….. (V)

The inequality shown in equation: (V) will be equal to zero if and only if p1 = p2 (since | x2 – x1|

≠ 0). But this is not possible as the price vectors p1 and p2 are different in the (p, x) vector space.

Thus we may write (p2 – p1) (x2 – x1) < 0 – which implies that an effect of a price change on the

consumption level of a consumer is negative – which is nothing but the substitution effect.

Hence it is proved that the substitution effect is negative.

Shephard’s lemma

Let the expenditure function of a consumer be m (p1, p2, …., pn, U*) which conforms to

minimization of the expenditure function subject to the utility constraint.

Formally, it may be expressed as:

m (pi, U*) = ∑i = 1n pi xi (pi, U*) (i = 1, 2, ….., n)

Shephard’s lemma states that the partial derivative of the expenditure function with respect to the

price of i – th commodity will be equal to the compensated demand function of the i – th

consumer.

Proof:

We have: m (pi, U*) = ∑i = 1n pi xi (pi, U*)

Thus δm ∕ δpi = ∑i = 1n xi (pi, U*) + ∑i = 1n pi * {δ xi (pi, U*) ∕ δ pi}

Since the compensated demand functions are derived by minimizing expenditure functions

subject to the utility constraints, hence in the finite (p, x, m) vector space, change of a single

price vector pi won’t affect the remaining pn – 1 price vectors. Thus under this assumption, the

term δ xi (pi, U*) ∕ δ pi = 0 and the second term on the right hand side of the above equation

vanishes.

Hence the above equation reduces to: δm ∕ δpi = xi (pi, U*) – which eventually proves the

Shephard’s lemma (Proved).

Consumer’s Surplus

Equivalent Variation (EV) of income method

Equivalent Variation (EV) of income method uses the current prices as the base to measure the

change of income required (at current prices) so that her change in wealth would be equivalent

between two time periods in case of an impact on her overall welfare. It may be formally defined

in two ways:
 (i) Let U0 = U0 (p0, w) and U1 = U1 (p1, w) with e (p0, U0) = e (p1, U1) = w;

where U0 (p0, w) = Utility at the base price p0 and welfare level (w),

U1 (p1, w) = Utility at the new price p1 and welfare level (w),

e (p0, U0) = Expenditure function / Indirect utility function at the base price p0 and

utility level U0,

e (p1, U1) = Expenditure function / Indirect utility function at the new price p1 and

utility level U1.

Hence as per definition, EV (p0, p1, w) = e (p0, U1) - e (p0, U0) = e (p0, U1) – w ………… (A)

(ii) EV (p0, p1, x0, x1, m) = (p0 x1) m = m1 – (p0 x0) m = m0

where x0 = Current consumption,

x1 = New consumption,

m0 = Current income, and

m1 = New income.

Now for the time period t ϵ [0, 1], the Equivalent Variation (EV) Index may be expressed as:

EV (p, x, m) = [(p0 x1) m = m1 ∕ (p0 x0) m = m0] * 100 ……………………. (B)

Hence if this concept is applied in a finite (p, x, m) space with good i (i = 1, 2, …., n), then

equation: (B) may be summed up and rewritten as:

EV (p, x, m) = [∑i = 1n (pi0 xi1) m = mi1 ∕ (pi0 xi0) m = mi0] * 100 …………. (C)
Equation: (C) is simply the Laspeyre’s Quantity Index (LQI) for measuring the change of

welfare for the consumer.

Compensating Variation (CV) of income method

Compensating Variation (CV) of income method uses the new prices as the base to measure the

change of income required (at new prices) so that her change in expenditure on commodities

would keep her utility level unchanged between two time periods in case of an impact on her

overall welfare. It may be formally defined in two ways:

(i) CV (p0, p1, w) = e (p1, U1) - e (p1, U0) = w - e (p1, U0) ………………… (D)

(ii) CV (p0, p1, x0, x1, m) = (p1 x1) m = m1 – (p1 x0) m = m0.

Now for the time period t ϵ [0, 1], the Compensating Variation (CV) Index may be expressed as:

CV (p, x, m) = [(p1 x1) m = m1 ∕ (p1 x0) m = m0] * 100 ……………………. (E)

Hence if this concept is applied in a finite (p, x, m) space with good i (i = 1, 2, …., n), then

equation: (E) may be summed up and rewritten as:

CV (p, x, m) = [∑i = 1n (pi1 xi1) m = mi1 ∕ (pi1 xi0) m = mi0] * 100 …………. (F)

Equation: (F) is simply the Paasche’s Quantity Index (PQI) for measuring the change of

welfare for the consumer.

 Price of good 1

h1 (p1, p*, U0) h1 (p1, p*, U1)

p10

p11 x1 (p1, p*, w)

Quantity of good 1

Figure 9 (a)

Consumer’s Surplus under the Equivalent Variation (EV) of income method (colored

region)

Price of good 1

h1 (p1, p*, U0) h1 (p1, p*, U1)

p10

p11 x1 (p1, p*, w)

Quantity of good 1

Figure 9 (b)

Consumer’s Surplus under the Compensating Variation (CV) of income method (colored

region)
Note:

[p* = (p2̅, p3̅, ………, pn - 1̅) = Set of constant prices other than p1 in the finite space]

Consumer’s Surplus

The difference between what a consumer actually pays for consuming a bundle of goods and

what she is willing to pay (rather than not purchasing the bundle of goods) is known as the

consumer’s surplus. A simple diagrammatic presentation is shown in figure 9 (c) below.

Price of good 1

A Consumer’s surplus

p0 B

x0
Quantity of Good 1

Figure 9 (c)

From figure 9 (c), it may be inferred that the triangular area Ap0B refers to the consumer’s

surplus. If price p > p0, then the area under consumer’s surplus will decrease and on the other

hand, the area under producer’s surplus will increase.

Formally, it may be expressed as:

Consumer’s Surplus (CS) = ∫0 x0 g (x) dx – p0 x0

= G (x0) – R0

where g (x) = Inverse demand function of the consumer,

G (x0) = Definite integral of ∫0 x0 g (x) dx, and

R0 = p0 x0 = Revenue of the producer.

The deadweight loss from commodity taxation

Let the government imposes a tax ‘t’ per unit on the consumption of good 1. The post – tax price

of good 1 now may be written as:

p11 = p10 + t

where p11 = Post – tax price of good 1,

p10 = Pre – tax price of good 1.

The gross revenue earned from taxation would be:

T = t * x1 (p1, w)

We know that EV (p0, p1, w) = e (p0, U1) - e (p0, U0).

Hence the Deadweight Loss = e (p1, U1) - e (p0, U0) – T.

In other words, the difference [w – T - e (p0, U1)] is known as the Deadweight Loss of

commodity taxation.
Formally, the deadweight loss may be defined as the economic event which leads to a net loss of

wealth of the consumer due to commodity taxation. In absence of taxation, the consumer might

have been on a higher indifference curve at the base price of the commodity.

Formally, the deadweight loss (DWL) in terms of Equivalent Variation (EV) of income

method may be expressed as:

DWLEV = e (p1, U1) - e (p0, U0) – T

= ∫ p10 p10 + t h1 (p1, p*, U1) dp1 - t h1 (p10 + t, p*, U1)

= ∫ p10 p10 + t [h1 (p1, p*, U1) d p1 - h1 (p10 + t, p*, U1)] dp1

where T = t * x1 (p1, w) = t * h1 (p1, U1), and

h1 (p1, U1) = δ e (p1, U1) ∕ δ p1.

On the other hand, the deadweight loss (DWL) in terms of Compensating Variation (CV) of

income method may be expressed as:

DWLCV = e (p1, U0) - e (p0, U0) – T = e (p1, U1) - e (p0, U0) – t * h1 (p1, U0)

= ∫ p10 p10 + t h1 (p1, p*, U0) dp1 - t h1 (p10 + t, p*, U0)

= ∫ p10 p10 + t [h1 (p1, p*, U0) - h1 (p10 + t, p*, U0)] dp1

where h1 (p1, U0) = δ e (p1, U0) ∕ δ p1.

 Price of good 1

h1 (p1, p*, U1) h1 (p1, p*, U0)

p10 + t

p10 x1 (p1, p*, w)

Quantity of good 1

Figure 10 (a)

The deadweight loss has occurred due to commodity taxation on good 1

based on U1 (colored region).

 Price of good 1

h1 (p1, p*, U1) h1 (p1, p*, U0)

p10 + t

x1 (p1, p*, w)

p10

Quantity of good 1

Figure 10 (b)

The deadweight loss has occurred due to commodity taxation on good 1

based on U0 (colored region).

Inverse demand function

In order to find the inverse demand function, let us assume:

m = Income = ∑i = 1n pi xi = 1 (i = 1, 2, …., n)

The conventional first order condition for utility maximization requires:

δ U (x) ∕ δ xi - λ pi = 0 ………………………… (G)

and m̅ = ∑i = 1n pi xi = 1.
Now by multiplying equation: (G) by xi, we get:

(δ U (x) ∕ δ xi) * xi - λ pi xi = 0

→ (δ U (x) ∕ δ xi) * xi = λ pi xi ……………………….. (H)

Now by summing up both sides of equation: (H), we get:

∑i = 1n (δ U (x) ∕ δ xi) * xi = λ ∑i = 1n pi xi

→ ∑i = 1n (δ U (x) ∕ δ xi) * xi = λ [Since ∑i = 1n pi xi = 1]

Hence by substituting the value of λ in equation: (G), we get:

δ U (x) ∕ δ xi - pi ∑i = 1n (δ U (x) ∕ δ xi) * xi = 0

→ pi = δ U (x) ∕ δ xi ∕ ∑i = 1n (δ U (x) ∕ δ xi) * xi ………… (I)

Equation: (I) is the inverse demand function for commodity xi.

Discontinuous demand functions

We have already known that the demand functions have to be strictly convex and continuous in

their entire range to suffice to the conditions of utility maximization for a consumer.

In other words, for any three commodity bundles x, y and z ϵ X and t being a scalar, the

condition tx + (1 – t) y > z (0 < t < 1) has to be met in order to maintain the continuity of the

demand curve. However, when the above condition is not met, then the preferences of the

consumer become nonconvex and hence the demand curve becomes discontinuous in nature

(Figure 11).
 Price

p1*

x1* x1 Quantity

Figure 11

References

Bilas, Richard A. (1971) Microeconomic Theory (2nd ed.), New York, NY: McGraw – Hill, Inc.

Gravelle, Hugh, and Rees, R. (1992) Microeconomics (2nd ed.), London: Pearson.

Henderson, James M., and Quandt, R. E. (1980) Microeconomic Theory: A Mathematical

Approach (3rd Ed.), New York, NY: McGraw – Hill, Inc.

Koutsoyiannis, A. (1979) Modern Microeconomics (2nd ed.), London: ELBS / MacMillan.

Mas – Colell, Andreu, Whinston, M. D., and Green, J. R. (2003) Microeconomic Theory, New

York, NY: Oxford University Press.

Varian, Hal R. (1992) Microeconomic Analysis (3rd ed.), New York, NY: W. W. Norton &

Company, Inc.