Advanced Microeconomics (ES30025)
Mathematics Review 2: The Lagrange Multiplier
Outline: I. Introduction
II. Duality Theory: Cobb Douglas Example III. Final Comments
I.
Introduction
The simplest method of solving a constrained maximization or constrained minimization problem is by the Lagrange multiplier method. In what follows we illustrate the method by considering, firstly, the consumer’s problems of maximizing (Cobb-Douglas) utility subject to a budget constraint, thereby yielding the consumer’s Marshallian demand function. We then, secondly, consider the consumer’s problem of minimizing expenditure subject to a (Cobb-Douglas) utility constraint, thereby yielding the consumer’s ( Hicks) Compensated demand function. We also allude to duality of the two problems and comment on the interpretation of the Lagrange Multiplier itself.
II.
Duality Theory: Cobb-Douglas Example
Marshallian Demand Functions
The Marshallian demand function is obtained by maximizing utility subject to a budget constraint. Thus: max x1,x2 Ax1bx 2 1−b st p1x1+ p2x2 =m (1) Solution: max x1,x2,λ { }L g = Ax 1 bx 2 1−b+λ m− p 1x1− p2x2
(
)
(2) ∂Lg ∂x1 =bAx1 b−1x 2 1−b−λp 1 =0 (3) ∂Lg ∂x2 =( )
1−b Ax1 b x2−b−λ p2 =0 (4) ∂Lg ∂λ =m− p1x1− p2x2 =0 (5) Dividing (3) by (4):bAx1b−1x 2 1−b 1−b
( )
Ax1bx 2 −b = λp1 λp2 ⇒ b 1−b( )
⋅xx21 = p1 p2 ⇒ x1= b 1−b( )
⋅ pp21 x2 (6) Rearrange (5): x2 = m p2 − p1 p2 x1 (7) Substitute (7) in (6): x1= b 1−b⋅ p2 p1 m p2 − p1 p2 x1 ⎛ ⎝⎜ ⎞ ⎠⎟ ⇒ x1=bm p1 (8) Thus: x2 = m p2 − p1 p2 x1 ⇒ m p2 − p1 p2 b m p1 ⎛ ⎝⎜ ⎞ ⎠⎟ x2 =( )
1−b m p2 (9)The consumer’s Marshallian demand functions are thus:
um =u x 1 m,x 2 m
(
)
=A x( )
1m b x 2 m( )
1−b ⇒ um =A bm p1 ⎛ ⎝⎜ ⎞ ⎠⎟ b 1−b( )
m p2 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ 1−b ⇒ um =Abb( )
1−b1−b m p1bp 2 1−b (11)Note that the budget constraint is satisfied:
p1x1m+ p 2x2 m= p 1 b m p1 ⎛ ⎝⎜ ⎞ ⎠⎟+p2
( )
1−b m p2 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥=m (12)Compensated Demand Function
The consumer’s Compensated demand function is obtained by minimizing expenditure subject to a utility constraint. Thus:
min x1,x2 p1x1+ p2x2 st Ax1bx 2 1−b =u (13) Solution: min x1,x2,λ { }L g = p 1x1+ p2x2 +θ u− Ax1 bx 2 1−b
(
)
(14) ∂Lg ∂x1 = p1−θbAx1 b−1x 2 1−b =0 (15) ∂Lg ∂x2 = p2−θ( )
1−b Ax1 bx 2 −b =0 (16) ∂Lg ∂θ =u −Ax1 bx 2 1−b =0 (17) Dividing (15) by (16) yields:p1 p2 = θbAx1b−1x 2 1−b θ
( )
1−b Ax1bx 2 −b ⇒ p1 p2 = b 1−b( )
⋅ xx21 ⇒ x1= b 1−b⋅ p2 p1 x2 (18) Rearrange (17): x2 = u Ax1b ⎛ ⎝ ⎜ ⎞⎠⎟ 1 1−b (19) Substitute in (19) in (18): x1= b 1−b⋅ p2 p1 u Ax1b ⎛ ⎝ ⎜ ⎞⎠⎟ 1 1−b ( ) ⇒ x1= b 1−b⋅ p2 p1 u A ⎛ ⎝⎜ ⎞ ⎠⎟ 1 1−b ( ) x1 −b 1−b ( ) ⇒ x11+ b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ = b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ p2 p1 u A ⎛ ⎝⎜ ⎞ ⎠⎟ 1 1−b ( ) ⇒ x1 1 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ = b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ p2 p1 u A ⎛ ⎝⎜ ⎞ ⎠⎟ 1 1−b ( ) (20)Raising every term by the power (1-b) yields:
x1= b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ p2 p1 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ 1−b ( ) u A ⎛ ⎝⎜ ⎞ ⎠⎟ (21)
x2 = u Ax1b ⎛ ⎝ ⎜ ⎞⎠⎟ 1 1−b ( ) = u A ⎛ ⎝⎜ ⎞ ⎠⎟ 1 1−b ( ) x1 −b 1−b ( ) ⇒ x2 = u A ⎛ ⎝⎜ ⎞ ⎠⎟ 1 1−b ( ) b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ p2 p1 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ 1−b ( ) u A ⎧ ⎨ ⎪ ⎩⎪ ⎫ ⎬ ⎪ ⎭⎪ −b 1−b ( ) = b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ p2 p1 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ 1−b ( ) ⎧ ⎨ ⎪ ⎩⎪ ⎫ ⎬ ⎪ ⎭⎪ −b 1−b ( ) u A ⎛ ⎝⎜ ⎞ ⎠⎟ ⇒ x2 = b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ p2 p1 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ −b u A ⎛ ⎝⎜ ⎞ ⎠⎟ ⇒ x2 = 1−b b ⎛ ⎝⎜ ⎞ ⎠⎟ p1 p2 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ b u A ⎛ ⎝⎜ ⎞ ⎠⎟ (22)
The consumer’s ( Hicks) Compensated demand functions are thus:
x1h = b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ p2 p1 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ 1−b ( ) u A ⎛ ⎝⎜ ⎞ ⎠⎟ x2h = 1−b b ⎛ ⎝⎜ ⎞ ⎠⎟ p1 p2 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ b u A ⎛ ⎝⎜ ⎞ ⎠⎟ (23)
Recall that the ( Hicks) Compensated demand functions measure the change in demand following a change in price holding the price of the other good and the consumer’s real income (vis. His ability to enjoy a particular level of utility) constant. Thus, the (Hicks) Compensated demand function measures the own-price substitution effect.
Note that the utility constraint is satisfied:
A x
( )
1h b x 2 h( )
1−b = A b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ p2 p1 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ 1−b ( ) u A ⎛ ⎝⎜ ⎞ ⎠⎟ ⎧ ⎨ ⎪ ⎩⎪ ⎫ ⎬ ⎪ ⎭⎪ b 1−b b ⎛ ⎝⎜ ⎞ ⎠⎟ p1 p2 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ b u A ⎛ ⎝⎜ ⎞ ⎠⎟ ⎧ ⎨ ⎪ ⎩⎪ ⎫ ⎬ ⎪ ⎭⎪ 1−b ⇒ Ax1bx 2 1−b =u (24)Note that minimized expenditure is given by:
mh = p 1x1 h+ p 2x2 h ⇒ mh = p 1 b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ p2 p1 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ 1−b ( ) u A ⎛ ⎝⎜ ⎞ ⎠⎟+ p2 1−b b ⎛ ⎝⎜ ⎞ ⎠⎟ p1 p2 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ b u A ⎛ ⎝⎜ ⎞ ⎠⎟ (25) Thus:
mh = p 1 bp 2 1−b u A ⎛ ⎝⎜ ⎞ ⎠⎟ b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ 1−b ( ) + 1−b b ⎛ ⎝⎜ ⎞ ⎠⎟ b ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⇒ mh = p 1 bp 2 1−b u A ⎛ ⎝⎜ ⎞ ⎠⎟ b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ −b 1+ b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⇒ mh = p 1 bp 2 1−b u A ⎛ ⎝⎜ ⎞ ⎠⎟ b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ −b 1 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⇒ mh = p 1 bp 2 1−b u A ⎛ ⎝⎜ ⎞ ⎠⎟ 1 b ⎛ ⎝⎜ ⎞ ⎠⎟ b 1 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ 1−b ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ (26)
III. Final Comments
Interpretation of the Lagrange Multiplier
The Lagrange multiplier measures the marginal effect of a relaxation in the constraint on the consumer’s objective function. Thus, the Lagrange multiplier in the first problem, λin the first problem measures the marginal effect on utility of a relaxation in the budget constraint and may therefore be interpreted as the consumer’s marginal utility of income. The Lagrange multiplier in the second problem, θ, measures the marginal effect on expenditure on a relaxation in the utility constraint.
Second-Order Conditions
Note that we have not explored the second-order conditions to confirm that we have indeed found the utility maximizing and cost minimizing choices for the consumer. We will generally assume that second-order conditions are satisfied in Lagrange multiplier constrained optimization problems.
Duality
Note the duality of the problem. Set the constraint level of utility u in the consumer’s (Hicks) Compensated demand function (23) equal to the Marshallian maximized utility um in
x1= b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ p2 p1 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ 1−b ( ) um A ⎛ ⎝⎜ ⎞ ⎠⎟ ⇒ x1= b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ p2 p1 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ 1−b ( ) 1 A ⎛ ⎝⎜ ⎞ ⎠⎟ Ab b
( )
1−b 1−b m p1bp 2 1−b ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⇒ x1=b1−b+b( )
1−b −( )1−b+( )1−b p 2 1−b ( )−( )1−b p1−( )1−b−bm ⇒ x1=bm p1 (27)This is the level of demand determined by the Marshallian demand function (10). A similar result occurs if we set the budget constraint in the consumer’s Marshallian demand function (10) equal to the (Hicks) Compensated minimized expenditure, mh in Equation (26):
x1=bm h p1 = b p1 p1 bp 2 1−b u A ⎛ ⎝⎜ ⎞ ⎠⎟ 1 b ⎛ ⎝⎜ ⎞ ⎠⎟ b 1 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ 1−b ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎧ ⎨ ⎪ ⎩⎪ ⎫ ⎬ ⎪ ⎭⎪ ⇒ x1= b 1 b ⎛ ⎝⎜ ⎞ ⎠⎟ b 1 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ 1−b ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ p2 p1 ⎛ ⎝⎜ ⎞ ⎠⎟ 1−b u A ⎛ ⎝⎜ ⎞ ⎠⎟ ⇒ x1= b1−b 1 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ 1−b ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ p2 p1 ⎛ ⎝⎜ ⎞ ⎠⎟ 1−b u A ⎛ ⎝⎜ ⎞ ⎠⎟ ⇒ x1= b 1−b ⎛ ⎝⎜ ⎞ ⎠⎟ p2 p1 ⎛ ⎝⎜ ⎞ ⎠⎟ ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ 1−b u A ⎛ ⎝⎜ ⎞ ⎠⎟ (28)