Mathematical Instruments of Comparative Static Analysis: Maximum Value

Function and the Envelope Theorem

Value Function: A function that expresses the idea that the optimized value of the objective function
depends, via the optimization procedure, only on the parameters of the problem is referred to as a
value function or alternatively, an indirect objective function. In other words, the idea expressed by a
value function is that the optimal value of the objective function is a function of the parameters of the
problem. Thus, a value function is an objective function where the choice variables have been
assigned their optimal values. These optimal values are, in turn, function of the parameters (or
exogenous variables) of the problem. Thus directly, the maximum value function is a function of the
choice variables that maximise the objective function. Indirectly, the maximum value of the objective
function is a function of the parameters. This is where the name “indirect objective” function comes
from. It is assumed to be a differentiable function in classical optimisation models.

To illustrate the notion of a value function, consider the following parameterized optimization
problem

max f(x, a) = ax − x 2
x

where a is the exogenous variable or the parameter which enters into the objective function and x is
the variable.

The first-order condition for a maximum is

a
a − 2x = 0  x ∗ (a) =
2
Substituting this value into the objective function we obtain
2 a2 a2
a a 2 a
V(a) = a (2) − (2) = − =
2 4 4

The function V(a) gives the maximised value of the objective function as a function of the parameters
a, and is referred to as the maximum value function. It parameterizes solution to an optimization
problem.

In general, for the problem

max f(x, a)
x

where a = (a1 ,a2 , …………, ak ) we can write the maximum value function as

V(a) = max f(x, a) = f(x ∗ (a), a)

x

where 𝑥 ∗ (a) is the optimal solution to the maximization problem. It simply states that the optimized
value of the function is equal to the function evaluated at the optimizing value.

1
Some Economic Examples of Value Functions

Some economic examples of value function includes

1) V(p, m) = max [ u(x)px = m]

x

where V is the indirect utility function. It is a maximum value function showing the maximum
attainable utility in utility maximization.

2) π(w, p) = max [ p f(x) − w x]

x

where p is the price of output, w is the vector of factor prices, x is the vector of inputs used in the
production process, and f(x) is the production function. The function π(w, p), which gives the
maximum profit as a function of the exogenous variables w and p, is called the profit function of the
firm. It is the maximum-value function for the profit maximization model.

3) e(p, u) = min [ p x u(x) = u∗ ]

x

where e is the expenditure function. It is a minimum value function showing the minimum
expenditure needed to attain the utility level u∗ . It is also referred to as the indirect objective function
for the expenditure minimisation problem.

4) c(w, y) = min [ w x f(x) = y]

x

where c is the cost function – a minimum value function showing the minimum cost needed to
produce the output level y. It is also referred to as the indirect objective function for the constrained
cost minimization model.

The Behaviour of the Maximum Value Function under Variation of the Parameters: The
Envelope Theorem
Once a maximum value function has been obtained, the comparative static analysis simplifies to
investigating the derivative properties of the function with respect to the parameters. A mathematical
theorem which is precisely useful in dealing with such derivative properties of the value function is
known as the Envelope Theorem. The theorem basically looks at how the value of the value function
changes as the parameters of the optimisation problem change. Thus, the Envelope Theorem seeks to
investigate the behaviour of the maximum value function under variation of the parameters. For
example, once the indirect utility function is found, the derivation of comparative static results of
consumer utility maximization theory becomes a lot easier, simpler and quicker with the use of
Envelope theorem.

2
It is relevant to note in this connection that the traditional method of comparative static analysis is
comprised of using tools like Implicit Function Theorem and Jacobian matrix. The main
problems faced under the traditional are that it can be very tedious and one may end up with
complicated expressions in determinant which is not unambiguously signed. However, unlike the
traditional method, the modern method of Envelope Theorem permits easier and quicker derivation of
comparative static results.

The Envelope theorem is derived for both unconstrained and constrained problems. The
Envelope Theorem for unconstrained problem seeks to demonstrate that the maximum value function 
indirect objective function forms an envelope to a set of optimized objective functions, which are
generated by varying the parameters of the model. The Envelope Theorem for the constrained
problem, on the other hand, seeks to demonstrate that the maximum value function /indirect objective
function forms an envelope to a set of optimized Lagrange functions where for each different value of
the parameter there will typically be a different Lagrange function.

The Envelope Theorem for Unconstrained Optimisation

Consider the following one-variable-one-parameter optimisation problem

max f(x, α) ………………… (1)

x

where x is the variable and α is the parameter.

The first-order condition for a maximum is

df(x,α)
=0 ………… (2)
dx

Solving (2) we obtain

x ∗ = x ∗ (α) ………… (3)

Substituting this value into the objective function we obtain

V(α) = f(x ∗ (α), α) ………… (4)

The function V(α) is called the maximum value function. What this function means is that the
maximum value of the objective function is a function of the value of x that maximises f(∙) (and this
value is a function of the parameter α) as well as of the parameter α. Thus indirectly, the maximum
value of the objective function is a function of the parameter α.

Differentiating (4) with respect to α we obtain

dV(α) ∂f(x∗ (α),a) ∂x∗ (a) ∂f(x∗ (α),a) ………… (5)

= +
dα ∂x∗ ∂α ∂α

However, from the first-order condition of optimisation (equation (2)) we have

3
 ∂f(x∗ (α),α)
=0
∂x∗

And hence the first term in equation (5) above will vanish, and we are left with

dV(α) ∂f(x∗ (α),a)

= ………… (6)
dα ∂α

This equation is the Envelope theorem result. According to this remarkable result, the total effect of a
small change in α on the value of V(α) is found by simply differentiating the direct objective function
f(x, α) partially with respect to α and thus treating x as a constant. Geometrically, equation (6)
represents a tangency position between the direct and indirect objective function.

An Analysis of the Envelope Theorem Result

To see what envelope theorem result does actually convey, we consider the following simple
statement of the theorem

dV(α) ∂f(x∗ (α),a) ∂x∗ (α) ∂f(x∗ (α),α)

= +
dα ∂x∗ ∂α ∂α

which is obtained for the problem involving one variable x and one parameter α.

Notice that, as 𝛼 changes, there are two effects: the change in a directly affects f(x(α), α) and the
change in a affects x which in turn affects f(x(α), α).But if x is chosen optimally, a small change in x
has a zero effect on f(x(α), α) so the indirect effect drops out and only the direct effect is left. This is
the essence of the envelope theorem. The theorem says that only the direct effect of a change in an
exogenous variable need to be considered, even though the exogenous variable may also enter the
maximum-value function indirectly as part of the solution to the endogenous choice variables. The
statement of the theorem is, thus, written as

dV(α) ∂f(x∗ (α),α)

dα
= | x = x∗(α)
∂α

This makes the study of the behaviour of V(α) under variation of the parameter α a lot easier. The
total effect of a small change in α on the value of V(α) is found by simply looking at the direct effect
only: the indirect effect of changing a on x will be zero because we look at optimal values of
f(x(α), α).

The Envelope Theorem for Constrained Optimisation

We derive below the Envelope Theorem for the following two-variable-one-parameter constrained
maximisation problem
Max 𝑓(𝑥, 𝑦, 𝛼)

subject. to ℎ(𝑥, 𝑦, 𝛼) = 0

4
The Lagrangian for this problem is

𝐿 = 𝑓(𝑥, 𝑦, 𝛼) +  (0 − ℎ(𝑥, 𝑦, 𝛼)) ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ (1)

The first- conditions are order

𝜕𝐿 𝜕𝑓 𝜕ℎ
= (𝑓𝑥 ) −  (ℎ ) = 0 ⋯⋯⋯⋯⋯⋯⋯⋯ (2)
𝜕𝑥 𝜕𝑥 𝜕𝑥 𝑥

𝜕𝐿 𝜕𝑓 𝜕ℎ
= (𝑓𝑦 ) − (ℎ𝑦 ) = 0 ⋯⋯⋯⋯⋯⋯⋯⋯ (3)
𝜕𝑦 𝜕𝑦 𝜕𝑥

𝜕𝐿
= − ℎ(𝑥, 𝑦, 𝛼) = 0 ⋯⋯⋯⋯⋯⋯⋯⋯ (4)
𝜕

Solving this system gives us

𝑥 = 𝑥 ∗ (𝛼); 𝑦 = 𝑦 ∗ (𝛼) ;  = ∗ (𝛼) ⋯⋯⋯⋯⋯⋯⋯⋯ (5)

Substituting the solution into the objective function we get

𝑣(𝛼) = 𝑓(𝑥 ∗ (𝛼), 𝑦 ∗ (𝛼), ∗ (𝛼)) ⋯⋯⋯⋯⋯⋯⋯⋯ (6)

Differentiating 𝑣(𝛼) with respect to 𝛼, we get

𝑑𝑣 𝜕𝑓 𝜕𝑥 ∗ 𝜕𝑓 𝜕𝑦 ∗ 𝜕𝑓
= + +
𝑑𝛼 𝜕𝑥 𝜕𝛼 𝜕𝑦 𝜕𝛼 𝜕𝛼

𝑑𝑣 𝜕𝑥 ∗ 𝜕𝑦 ∗ 𝜕𝑓
∴ = 𝑓𝑥 + 𝑓𝑦 + ⋯⋯⋯⋯⋯⋯⋯⋯ (7)
𝑑𝛼 𝜕𝛼 𝜕𝛼 𝜕𝛼

Substituting (5) into (2) and (3) we obtain

𝑓𝑥 (𝑥 ∗ (𝛼), 𝑦 ∗ (𝛼), ∗ (𝛼)) = ∗ (𝛼)ℎ𝑥 (𝑥 ∗ (𝛼), 𝑦 ∗ (𝛼), ∗ (𝛼)) ⋯⋯⋯⋯⋯⋯⋯⋯ (8)

∗
𝑓𝑦 (𝑥 ∗ (𝛼), 𝑦 ∗ (𝛼), ∗ (𝛼)) = ∗ (𝛼)ℎ𝑦 (𝑥 ∗ , (𝛼), 𝑦 ∗ (𝛼) (𝛼)) ⋯⋯⋯⋯⋯⋯ (9)

Substituting (8) and (9) into (7) we obtain

𝑑𝑣 𝜕𝑥 ∗ 𝜕𝑦 ∗ 𝜕𝑓
=  𝛼 ℎ𝑥 (𝑥 𝛼 , 𝑦 𝛼 ,  𝛼 )
∗
( ) ∗
( ) ∗( ) ∗( )
+  𝛼 ℎ𝑦 (𝑥 𝛼 , 𝑦 𝛼 ,  𝛼 )
∗
( ) ∗
( ) ∗( ) ∗( )
+
𝑑𝛼 𝜕𝛼 𝜕𝛼 𝜕𝛼

𝜕𝑥 ∗ 𝜕𝑦 ∗ 𝜕𝑓
= ∗ (𝛼) [ℎ𝑥 (∙) + ℎ𝑦 (∙) ]+ ⋯⋯⋯⋯⋯⋯ (10)
𝜕𝛼 𝜕𝛼 𝜕𝛼

5
If we substitute the solutions to 𝑥 and 𝑦 into the constraint we get

ℎ(𝑥 ∗ (𝛼), 𝑦 ∗ (𝛼), ∗ (𝛼)) = 0 ⋯⋯⋯⋯⋯⋯ (11)

Differentiating with respect to 𝛼, we get

𝜕𝑥 ∗ 𝜕𝑦 ∗ 𝜕ℎ
ℎ𝑥 + ℎ𝑦 + =0
𝜕𝛼 𝜕𝛼 𝜕𝛼

𝜕𝑥 ∗ 𝜕𝑦 ∗ 𝜕ℎ
∴ ℎ𝑥 + ℎ𝑦 =− ⋯⋯⋯⋯⋯⋯ (12)
𝜕𝛼 𝜕𝛼 𝜕𝛼
Substituting (12) into (10) gives

𝑑𝑣 𝜕ℎ 𝜕𝑓
= −∗ (𝛼 ) + ⋯⋯⋯⋯⋯⋯ (13)
𝑑𝛼 𝜕𝛼 𝜕𝛼

From the Lagrangian function calculated at the optimum, we have

𝜕𝐿∗ 𝜕𝑓 𝜕ℎ 𝜕ℎ 𝜕𝑓
𝜕𝑥
=
𝜕𝛼
− ∗ (𝛼)
𝜕𝛼
= −∗ (𝛼) 𝜕𝛼 + 𝜕𝛼 ⋯⋯⋯⋯ (14)

Comparing (13) and (14) we have

𝑑𝑣 𝜕𝐿∗
= ⋯⋯⋯⋯ (15)
𝑑𝛼 𝜕𝑥

This is the Envelope Theorem result for the constrained maximisation problem. The result
demonstrates that the partial derivative of the maximum value function with respect to a parameter,
such as 𝛼, is the same as the partial derivative of the Lagrangian function with respect the same
parameter.

Some Useful Envelope Theorem Results

Envelope theorem has wide variety of applications in economics that encompass constrained as well
as unconstrained optimization problems. We seek to illustrate one such example from unconstrained
optimization problems.

Profit functions and Hotelling’s Lemma

Consider the following one-input-one-output profit maximisation problem of a firm

max  (x) = p1 f(x) − p2 x ………………………………. (1)

6
where p1 is the price of output, p2 is the price of input, x is the quantity of input used in the
production process, and f(x) is the production function.

Derivation of the indirect profit function

The first-order condition for a maximum profit is

df(x)
p1 − p2 = 0 ………………………………… (2)
dx

which gives the solution

x = x (p1 , p2 ) ………………………………………….. (3)

Substituting this solution into the objective function gives

 (p1 , p2 ) = p1 f(x(p1 , p2 )) − p2 x (p1 , p2 ) …………………… (4)

where  (p1 , p2 ) is the maximum value of the profit function. This function is called the indirect
profit function (or the optimal profit function). It gives maximum profit as a function of the
exogenous variables p1 and p2 .

Derivation of Hotelling’s Lemma

By the Envelope theorem, we obtain from (4)

d (p1 ,p2 )
= − x x = x (p , p ) = f(x(p1 , p2 )) …………………………………… (5)
dp1 1 2

d (p1 ,p2 )
= f(x) = − x (p1 , p2 ) .………………………………. (6)
dp2 x = x (p1 , p2 )

Equation (5) tells us that the partial derivative of the indirect objective function  (p1 , p2 ) with
respect to p1 is simply the partial derivative of the objective function with respect to p1 , evaluated at
the optimal choice x = x (p1 , p2 ). This we refer to as the profit −maximising supply of the firm at the
price vector (p1 , p2 ). It tells us how much output the profit –maximising fill will supply when prices
are p1 for output and p2 for inputs.

Equation (6) tells us that the partial derivative of the indirect objective function  (p1 , p2 ) with
respect to p2 is simply the partial derivative of the objective function with respect to p2 , evaluated at
the optimal choice x = x (p1 , p2 ). This we refer to as the profit −maximising input (or factor) demand
function of the firm at the price vector (p1 , p2 ). The minus sign comes from the fact we are increasing
the price of the input− so profit must decrease.

7
Equations (5) and (6) are collectively known as Hotelling’s Lemma. They state that the partial
derivatives of the optimal profit function yield the firm’s output-supply function and the input-
demand functions, respectively. Notice that Equation (5) is simply the statement that if the price of
output rises by a marginal unit, the firm will earn additional profit leaving apart possible changes in
production. Also notice that Equation (5) is simply the statement that if the price of the input rises by
one percent and the firm uses x units of the input, then its profit will decrease by x units, even taking
into consideration possible changes in production.

What does Hotelling’s Lemma actually convey?

To see what Hotelling’s Lemma does actually convey, we look at equation (5). When the price of an
input increases by a small amount there will be two effects. First, there is a direct effect: because of
the price increase the firm will make more profits, even if it continues to produce the same level of
output. But secondly, there will be an indirect effect: the increase in the output price will induce the
firm to change its level of output by a small amount. However, the change in profits resulting from
any change in output must be zero since the firm has already chosen a profit−maximising production
plan. Hence, the impact of the indirect effect is zero, and we are left with the direct effect. To see this
result more formally, we differentiate the optimal profit function (4) with respect to p1 to obtain

∂ (p1 ,p2 ) ∂f(x (p1,p2)) ∂x (p1 ,p2 ) ∂x (p1 ,p2 )

= f(x(p1 , p2 )) + p1 ∂x
−p2
∂p1 ∂p1 ∂p1

∂f(x (p1 ,p2 )) ∂x (p ,p )

= f(x(p1 , p2 )) + [p1 ∂x
− p2 ] ∂p1 2 ……… (7)
1

The first term on the right-hand hide represents the direct effect while the second-term the indirect
effect. Now by virtue of the first-order condition at the optimum we have

∂f(x (p1,p2))
p1 ∂x
− p2 = 0

Substituting this condition into (7), we obtain

∂ (p1 ,p2 )
= f(x(p1 , p2 ))
∂p1

This explains why indirect effect is zero. In general, the basic advantage of using the Envelope
Theorem approach is that only the direct effect of a change in an exogenous variable is to be
considered. There is no need to consider the indirect effect as the choice variables are chosen
optimally.

8
The property of monotonicity of the indirect profit function

The derivative properties of the optimized profit function can be written as

d (p1 ,p2 )
=−x = f(x(p1 , p2 ))
dp1 x = x (p1 , p2 )

d (p1 ,p2 )
= f(x) x = x (p1 , p2 ) = − x (p1 , p2 )
dp2

where p1 denotes the price of output and p2 the price of input. The above equations say that partial
derivative of  (p) with respect to price i, where i = 1, 2 will be positive if good I is an output and
negative if good I is an input. This essentially implies that the profit function is a monotonic function
of the prices. Thus, Hotelling’s Lemma yields an important property of the profit function, namely,
the property of monotonocity of the profit function with regard to prices.