DYNAMIC SYSTEMS

Statics vs. Dynamics in Economic Analysis

 In statics we are interested in the properties and comparison of two
equilibrium points.
Dynamics
In dynamics, we are concerned with the path of key economic variables and
their stability over time.
Definition: A system is dynamical, if its behaviour over time is determined by
functional equations, in which variables at different points in time are
related in an essential way.
A functional equation is an equation in which the unknown is a function.

January 1, 2025
Dynamic Systems 1
 To solve an equation in the usual sense is to find the value (or values) of
the unknown which satisfies the equation.
On the other hand, to solve a functional equation means to find the
function which satisfies the functional equation identically.

Continuous vs. discrete time

Since we are interested in the behaviour of economic variables over-time,
we need to specify (assume) how we treat time: Continuous or discrete
time?
Applications of continuous time are mostly found in natural sciences.
[Examples]

2
Dynamic Systems January 1, 20
25 2
 Dynamics: Continuous time
Introduction:
Consider the functional equation: y’(x) – y(x) = 0, where y is a variable of interest
which depends on exogenous variable, x.
It is easy to confirm that the function that solves the equation is: y(x) = Ae x, since
y’(x) = Aex for any x (i.e., it satisfies the functional equation identically).

Now consider the function: y = ax + b, which gives: y’ = a. If we choose x = (a-b)/a,

we have also y =a(a – b)/a + b = a, i.e., y’ – y = 0. However, for any other value of x,
the value of the function will not be = a, therefore, the function y = ax + b does not
satisfy out functional equation identically.

Dynamic Systems January 1, 2025 3

  Let’s acknowledge that the symbol x stands for time, and we are ready to
understand the nature of dynamic systems. Using the earlier example:

y’(t) – y(t) = 0

This is a functional equation which relates the value of y at any point in time
and the value it takes at an arbitrarily close point given by y’(t).

January 1, 2025
Dynamic Systems
4
 Use of integration in dynamics (with continuous time)
To introduce the use of integration: Suppose that a variable, for example
population, grows over time at the rate:
dy/dt = y’(t) = 1/√t ( or t-1/2) (1)

We want to find the time path of this population, y(t), given that this
population is growing at the rate given by (1).
We are looking for the function y = y(t), given that we know its derivative. We
need to find the primitive function corresponding to the derivative. This is
the essence of integration.

Use of Integration January 1, 2025 5

 The function whose derivative is t-1/2 is: y(t) = 2t1/2 + c, where c is an arbitrary
constant.

In order to be able to solve for the constant, we need additional information.
In most cases we have such information; we usually know the value of the
function at some point in time, usually at time 0. This is called an initial
condition.
If, for example, we know that population at time 0 is: y(0) = 10, then:
y(0) = 2(0)1/2 + c = 10 → c = 10, and:
y(t) = 2t1/2 + 10 [or y(t) = 2t1/2 + y(0)].
Before we proceed with more examples, we can refresh the rules of integration.

January 1, 2025
Use of Integration 6
 Rules of integration
Notation: ∫f(x)dx
The outcome of integration is:
∫f(x)dx = F(x) + c, such that:
F’(x) = f(x)
Frequently used rules (which mirror those of differentiation):
(1) Integral of a constant, k:
∫kdx = kx + c
(2) Integral of a power function:
∫xndx = (xn+1)/(n+1) + c (n ≠ -1)
(3) Integral of 1/x:
∫(1/x)dx = ln(x) + c (also a sub-case of (2))

Use of Integration January 1, 2025 7

 (4) Integral of a natural exponential function:
∫ekxdx = (1/k)ekx + c
(special case: ∫exdx = ex + c)

(5) Integral of a constant times a function:

∫kf(x)dx = k∫f(x)dx

(6) Integral of a sum of 2 or more functions:

∫[f(x) ± g(x)]dx = ∫f(x)dx ± ∫g(x)dx

(7) Integral of the negative of a function:

∫-f(x)dx = - ∫f(x)dx

Use of Integration January 1, 2025 8

 Examples

1. ∫9dx = 9x + c

2. ∫x4dx = (x4+1)/(4+1) = x5/5 + c

3. ∫2x2 – x + 1)dx = 2∫x2 - ∫xdx + ∫dx

= 2(x3/3) - ½(x)2 + x + c
= 2/3(x)3 - ½(x)2 + x + c

4. ∫12x-1dx = 12∫(1/x)dx = 12 ln|x| + c

5. ∫20e-4x = 20[(-1/4)e-4x ] = -5e-4x + c

Use of Integration January 1, 2025 9

Integration by substitution
It may work in cases where it is difficult or impossible to use the basic rules.
Example (here we could still have used the simple rules):
∫20x4(x5 + 7)dx
Let another function, u = (x5 + 7)
Differentiate u: du/dx = 5x4
Solve for dx: dx = du/5x4
Substitute u for (x5 + 7) and du/5x4 for dx:
∫20x4(x5 + 7)dx = ∫20x4u(du/5x4)
= ∫4udu = 4∫udu
Integrate with respect to u:
4∫udu = 4(u2/2) = 2u2 + c

Use of Integration January 1, 2025 10

 Convert back in terms of the original function:
∫20x4(x5 + 7)dx = 2(x5 + 7)2 + c

Now consider the following example:

∫3x(x + 6)2 dx
Let u=x +6, then du/dx = 1 → dx = du
∫3x(x + 6)2 dx = ∫3xu2du = 3∫xu2du,
and we stop there because integration by substitution will not work.
However, another method will work (in this example):

Use of Integration January 1, 2025 11

 Integration by parts
It is derived by reversing the process of differentiation of a product:

d[f(x)g(x)]/dx = f(x)g’(x) + f’(x) g(x)

take the integral of the derivative:

f(x)g(x) = ∫f(x)g’(x)dx + ∫f’(x)g(x)dx

then solve algebraically for the first integral on the right-hand side:

∫f(x)g’(x)dx = f(x)g(x) - ∫f’(x)g(x)dx

Then, use the it as a formula/template

Use of Integration January 1, 2025

12
 Now, let’s go back to the previous example:
∫3x(x + 6)2 dx
Separate the integrand (i.e., what is to be integrated) into 2 parts as in the
formula. Usually we use the simpler function for f(x) and the more complex for
g’(x).
Let f(x) = 3x, g’(x) = (x + 6)2, then:
f’(x) = 3, g(x) = ∫(x + 6)2dx = 1/3(x + 6)3 + c1

Substitute for f(x), f’(x) and g(x) in the formula (note that g’(x) is not used):
∫3x(x + 6)2 dx = f(x)g(x) - ∫f’(x)g(x)dx

Use of Integration January 1, 2025

13
 = 3x[1/3(x + 6)3 + c1] – 3∫[1/3(x + 6)3 + c1]dx
= x(x + 6)3 + 3xc1 - ∫[(x + 6)3 + 3c1]dx
= x(x + 6)3 + 3xc1 – 1/4(x + 6)4 – 3xc1 + c
= x(x + 6)3 - 1/4(x + 6)4 + c
So:
∫3x(x + 6)2 dx = x(x + 6)3 - 1/4(x + 6)4 + c
Note that c1 does not appear in the final solution. We can always assume that c1
is always = 0.
Check the answer by differentiating.

Use of Integration January 1, 2025 14

 Another example:
∫5xex-9dx
Let f(x) = 5x and g’(x) = ex-9, then:
f’(x) = 5, g(x) = ∫ex-9dx = ex-9 + c1

∫5xex-9dx = f(x)g(x) - ∫f’(x)g(x)dx

= 5xex-9 + c1 - ∫(5ex-9 + c1)dx
= 5xex-9 - 5ex-9 + c

So, ∫5xex-9dx = 5xex-9 - 5ex-9 + c

Use of Integration January 1, 2025 15

 Definite integrals
The numerical value of a definite integral of a continuous function f(x)
over the interval [a to b] is:
b
∫f(x)dx, [which is a number]
a
i.e., given by the primitive function F(x) + c, evaluated at the upper
limit b, minus the same primitive evaluated at the lower limit, a.

b b
∫f(x)dx = [F(x)] = F(b) – F(a)
a a
3 3
Example: ∫(6x2 + 5)dx = [2x3 + 5x] = [2(3)3 + 5(3)] - [2(1)3 + 5(1)]= 62
1 1
Use of Integration January 1, 2025 16
 A few simple economic applications
(1) Finding a total function from the derivative function
Suppose that the slope of the total cost function (TC = f(q)) of a firm at any point
is (1/2)q, and we know that the total cost curve passes through the point (2,7).
Find the equation of the total cost function.
You are essentially given the marginal cost (MC) curve.
Then: TC= ∫(1/2)qdq = (1/2)[q2/2] = q2/4 + c
We can definitize the constant, c, if we utilize the additional information.
TC Given that when q =2, TC = 7,
TC = (2)2/4 + c = 7 → c = 6, and:
7
TC = 6 + q2/4
(=TFC + TVC)

2 q 17
Use of Integration January 1, 2025
 (2) Calculation of capital stock
This is done on the basis of a known values of investment at different points in
time (rate of change of capital stock).

Example: If we are given that investment at time t is given by the function:

I(t) = 12t1/2
And additional information that K(0) = 25 (this is an example from the book), we
have:
K(t) = ∫12t1/2dt = 12(t3/2/1.5) + c = 8t3/2 + c
K(0) = 8(0) + c = 25 → c = 25, and:
K(t) = 8t3/2 + 25, which is the time path of capital stock.

On the other hand, if you are asked to find the amount of capital accumulated over a
particular time interval, say (1, 3) in years:
3 3
K(3) – K(1) = ∫12t1/2 dt = 8[t3/2 ] = 33.6
Use of Integration January 1, 2025 18
(3) Present value of a cash flow
Another application is finding the P.V. of a cash flow (a series of revenues
receivable, or costs incurable at various times in the future).
We recall the present value formula:
I = R/(1 + i)t , where I is the present value of a future amount R, given interest rate i.
This refers to a single future value, R (if interest is compounded once a year).
The continuous time counterpart of this:
Assume interest is compounded continuously (at every point in time):
Suppose an amount, I0, is invested at the rate of interest, i; we know that the total
amount expected t years later is:
R(t) = Ioeit
We can rearrange to get: Io = R(t)/eit = Re-it
(How much to invest now, to receive R, t years later)

Use of Integration January 1, 2025 19

  Now suppose a person expects to earn R at the end of the first period and
subsequent periods (annuity), from time 0 to t. Then he has to invest:
t
I = ∫Re-itdt
0
where I is the present value of a series of annuities given by the return expected
from an investment outlay.
i.e., if i = 0.04, and the investor wishes to receive $100 for 10 years, the amount
needed to be invested is:
10
I = ∫100e-itdt = - (100/0.04)e-0.4 + (100/0.04)eo
0
= 100/0.04[1 - e-0.4] = 2,500[1 - e-0.4] = $825

Use of Integration January 1, 2025

20
 Economic dynamics
 We can introduce economic dynamics by considering pre-Keynesian vs.
Keynesian notions of equilibrium.
 Before Keynes, the question asked was whether the equilibrium exists
and if it is unique and stable. The explanation stopped there (lacked
dynamical element). Keynesian analysis is different.

 For example: Consider a deflationary gap.

Reaching full employment equilibrium requires an increase aggregate
expenditure: Suppose there is an increase in new investment to bring total
effective demand to the level of full capacity utilization (full
employment).
 But the very fact new investment is undertaken, will change the
objective situation on which the current equilibrium is based.
 Investment will increase productive capacity at the same time it
increases total effective demand (aggregate demand).
Economic Dynamics January 1, 2025 21
 So, investment plays a special role in Keynesian economic dynamics
and acts trough 2 different channels:
1. Change in effective demand through the (simple Keynesian)
multiplier:
dy/dt = m(dI/dt) = 1/s(dI/dt) [where s is the MPS]
2. Total new investment is an addition to existing productive capacity.

Economic Dynamics
January 1, 2025 22
 Application: Domar’s growth model
The model assumes a rather rigid production function:
k = ρK (1)
where k is capacity output (full employment output), taken to be a
multiple of capital. As a result, ρ = k/K.
One implication here is that, since labour is indispensable, K and L are
combined in fixed proportion, because only then we can consider
capital alone.

By differentiating (1):
(change in productive capacity): dk/dt = ρ(dK/dt) = ρI
and:
(change in aggregate demand): dY/dt = 1/s(dI/dt)
Economic Dynamics January 1, 2025 23
 For equilibrium over time, we require: dk/dt = dY/dt (maintained over time)
There is no reason why these two effects should necessarily be compatible with the
maintenance of full capacity utilization over time. We can explore, however, the
conditions to be satisfied in order that effective demand and productive
capacity may expand in a compatible way over time.

First, total effective demand (aggregate demand) at time 0 must be equal to

productive capacity (full employment output) to begin with: Y(0) = k(0)

Secondly, dk/dt (rate of change of capacity output, i.e., from the supply side) =
dY/dt (rate of change of aggregate demand) must continuously hold over time,
i.e., ρI = 1/s(dI/dt),

Economic Dynamics January 1, 2025

24
 or after separating terms:
dI/I = ρsdt, or (1/I)dI = ρsdt. Integrating both sides:
→ ∫(1/I)dI = ∫ρsdt → ln(I) = ρst + c, or
eln(I) = e(ρst + c) → It = eρstec, where ec is a constant (call it A).
Then:
It = Aeρst (where A = I0, since I0 = Ae0 = A)
So:
It = I0eρst
The above says that, in order to maintain equilibrium in the long-run, new
investment must grow over time at a rate g = ρ*s over time). An economy
might satisfy or not satisfy this condition.
 We will be introducing a more general approach, i.e., using differential
equations (under continuous time). [Next file]

Economic Dynamics January 1, 2025

25