Addis Ababa University

College of Business and

Economics
Department of Economic
Econ 3051: Mathematical Economics
Tewodros Negash (PhD)
 Tewodros Negash (PhD), Addis Ababa
Dynamic optimization University, Department of Economics

• Suppose that a firm’s output depends only on the

amount of capital it employs.
𝑄 = 𝑞(𝐾)
where 𝑸 is the firm’s output level, 𝒒 is the production
function and 𝑲 is the amount of capital employed
• If p is the price of output and 𝑹 is the rental price of 𝑲
per unit, the firm’s profit function can be written as
𝜋 𝐾 = 𝑝𝑞 𝐾 − 𝑅𝐾
• If the objective of the firm is to choose 𝑲 to maximize
current profit, the optimal amount of 𝑲 is given by the
usual first – and second – order conditions:
𝜋 ′ 𝐾 = 𝑝𝑞′ 𝐾 − 𝑅 = 0 and 𝜋 ′′ 𝐾 < 0 → 𝜋 ∗ 2
• But why should the firm care only about current profit? Why not
it take a long – term view and also care about future profits?
• The firm’s objective should be to maximize the discounted sum of
profits over a planning time running from 𝑡 = 0 to 𝑡 = 𝑇
• The firm’s profit is again assumed to depend on 𝑲, which is
influenced by investment (𝑰) and depreciation(𝛿).
𝐾 = 𝐼 𝑡 − 𝛿𝐾(𝑡)
• 𝑰(𝒕) is amount of capital purchased at time 𝒕. Once purchased
the capital lasts for a long time.
• The problem facing the firm at each point in time is therefore to
decide on how much capital to purchase
Tewodros Negash (PhD), Addis Ababa
3
University, Department of Economics
• The firm’s problem becomes a dynamic optimization problem
– Current investment affects current profit and future profits
through its effect on capital available for future production.
• The general form of a dynamic optimization problem

𝐽= 𝑒 −𝜌𝑡 𝜋 𝐾 𝑡 , 𝐼 𝑡 , 𝑡 𝑑𝑡
0
Subject to 𝐾 = 𝐼 𝑡 − 𝛿𝐾
𝐾 0 = 𝐾0 , 𝐾 𝑇 = 𝐾𝑇
• 𝜌 and 𝑒 −𝜌𝑡 are the discount rate and the continuous – time
Tewodros Negash (PhD), Addis Ababa
discount factor, respectively. University, Department of Economics

• 𝐾0 and 𝐾𝑇 are initial and terminal condition of capital stock 4

• 𝐾 = 𝐼 𝑡 − 𝛿𝐾 describes the linkage between 𝐾 and 𝐼.
– It is called state equation or equation of motion
• The firm chooses 𝑰(𝒕), which will then affect the value of 𝑲(𝒕)
via the first – order differential equation. In turn 𝑲(𝒕) will
determine the profit of the firm.
• Thus, the firm’s profit depends on the path chosen for 𝑰(𝒕).
• 𝑱 is the value of the functional to be maximized.
• 𝑰(𝒕) is the control variable whose value is directly chosen or
controlled to optimize 𝑱.
• 𝑲(𝒕) is the state variable whose value is indirectly determined
by the control variable via the state equation
Tewodros Negash (PhD), Addis Ababa
5
University, Department of Economics
Optimal control
• A method used to solving dynamic optimization problem
• It is based on the maximum principle
The Maximum Principle
• The maximum principle constitutes a set of first – order
necessary conditions that hold only on optimal paths.
• The maximum principle involves an approach similar to the
Lagrangian function and the Lagrangian multiplier variable.
• In optimal control problem, these are known as the
Hamiltonian function and costate variable
Tewodros Negash (PhD), Addis Ababa
6
University, Department of Economics
The Hamiltonian function for the dynamic optimization problem
is
𝐻 𝑥 𝑡 , 𝑦 𝑡 , 𝜆 𝑡 , 𝑡 = 𝑓[𝑥 𝑡 , 𝑦 𝑡 , 𝑡 + 𝜆 𝑡 𝑔[𝑥 𝑡 , 𝑦 𝑡 , 𝑡]
• Where 𝝀(𝒕), referred to as the costate variable, is similar to
the Lagrange multiplier in constrained optimization
problems.
• The Hamiltonian is similar to the Lagrangian function in
constraint optimization problem
• To form the Hamiltonian take the integrand (the function
under the integral sign) and add to it the equation of motion
Tewodros Negash (PhD), Addis Ababa
University, Department of Economics
(the state equation) multiplied by 𝝀(𝒕).
7
The maximum principle
• Optimal solution must satisfy the following necessary conditions
i) The control variable 𝒚(𝒕) maximizes 𝐻. That is,
𝜕𝐻
=0
𝜕𝑦
ii) The paths of 𝒙(𝒕) and 𝝀(𝒕) (state and costate variables) are
given by the solution to the following system of differential
equations:
𝜕𝐻
a) 𝜆 = −
𝜕𝑥

𝜕𝐻
b) 𝑥 =
𝜕𝜆

iii) The boundary condition

The two boundary conditions used to solve the system of
differential equations are given by
𝑥 0 = 𝑥0 , 𝑥(𝑇) = 𝑥𝑇
Tewodros Negash (PhD), Addis Ababa
8
University, Department of Economics
Example 1
Solve the following problem
1
𝑀𝑎𝑥 0 𝑥 − 𝑦 2 𝑑𝑡
Subject to 𝑥 = 𝑦
𝑥 0 =2 𝑥 𝑇 𝑓𝑟𝑒𝑒 - free end point
Solution:
A. Form the Hamiltonian function: 𝐻 = 𝑥 − 𝑦 2 + 𝜆𝑦
B. Apply the maximum principle
𝜕𝐻 𝜆(𝑡)
1. = −2𝑦 + 𝜆 = 0 this gives 𝑦= (1)
𝜕𝑦 2
𝜕𝐻
2. (a) 𝜆 = − = −1 (2)
𝜕𝑥
𝜕𝐻 𝜆(𝑡)
(b) 𝑥 = =𝑦 →𝑥= (3)
𝜕𝜆 2
C. Obtain the boundary conditions
𝑥 0 =2 𝜆 𝑇 = 𝜆 1 = 0 - Transversality condition
for free end point
Tewodros Negash (PhD), Addis Ababa
9
University, Department of Economics
Example 2
2
2. 𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒 0
6𝑥 − 4𝑦 2 𝑑𝑡
𝑆𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 𝑥 = 16𝑦
𝑥 0 = 24 𝑥 2 = 408 - fixed end point
A. The Hamiltonian is 𝐻 = 6𝑥 − 4𝑦 2 + 𝜆(16𝑦)
B. The necessary conditions are
𝜕𝐻
1. = −8𝑦 + 16𝜆 = 0 → 𝑦 = 2𝜆 (1)
𝜕𝑦

𝜕𝐻
2. (a) 𝜆 = − = −6 (2)
𝜕𝑥

𝜕𝐻
(b) 𝑥 = = 16𝑦 (3)
𝜕𝜆

C. Obtain the boundary conditions

𝑥 0 =2 𝑥 2 = 408 Tewodros Negash (PhD), Addis Ababa
University, Department of Economics 10
Exercise:
• Solve the following problems
4
1. 𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒 0
8𝑥 − 10𝑦 2 𝑑𝑡
𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 𝑥 = 24𝑦
𝑥 0 =7 𝑥 4 𝑓𝑟𝑒𝑒
3
2. 𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒 0
4𝑥 − 5𝑦 2 𝑑𝑡
𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 𝑥 = 8𝑦
𝑥 0 =2 𝑥 3 = 117.2
Tewodros Negash (PhD), Addis Ababa
11
University, Department of Economics