UNIT 8 DIFFERENCE EQUATIONS AND

APPLICATIONS IN ECONOMIC
DYNAMICS
Structure
8.0 Objectives
8.1 Introduction
8.2 Difference Equations in Economics
8.3 Solving First Order Difference Equations
8.3.1 Behaviour of Solutions of First Order Equations
8.3.2 Economic Applications of First Order Equations
8.4 Solving Second Order Difference Equations
8.4.1 Homogeneous Equations
8.4.2 Behaviour of Solutions of Homogeneous Equations
8.4.3 Non-homogeneous Equations
8.4.4 An Economic Application of Second Order Non-homogeneous Equation
8.5 Let Us Sum Up
8.6 Key Words
8.7 Some Useful Books
8.8 Answer or Hints to Check Your Progress
8.9 Exercises

8.0 OBJECTIVES
After going through this unit you should be able to:
solve problems of economic dynamics where the time variable takes only
discrete values.

8.1 INTRODUCTION
A difference equation is used to solve the values of an unknown function
y(x) for different discrete values of x. We obtain a function y(x) such that it
satisfies the equation for all values of x. In order to understand the process of
formulation of the difference equation, you may recall the discussion on
differential equation presented in the preceding unit. See that difference and
differential equations are exactly analogous with the only difference that the
former applies when the independent variable takes only discrete values,
whereas the latter when it is continuous.

8.2 DIFFERENCE EQUATIONS IN ECONOMICS

To get an idea about how difference equations may come up in economics
consider the case where it is known that the national income y of a particular
country has been growing at a constant rate g over (say) a ten year period
starting from some base year. The rate of growth of y at any period t may be
y y t 1
represented as t . Note that this is the expression that gives the rate
y t 1
42 of growth of y at a particular point in time. In contrast, you have seen the
 dy (t ) dt Difference Equations and
expression when y was treated as a continuous variable. Treating y Applications in Economic
y Dynamics
as discrete, we find the numerator as the increment of income in current
period (period t) over the level attained in the immediately preceding period
(period t 1). The ratio of this to the income yt1 of the preceding period
gives the current rate of growth. Since we have said that rates of growth of
income, g, is constant (i.e., independent of t) over ten years interval, it can be
written as:
y t y t 1
= g, t, = 1, 2, 3, 4, , 10
y t 1
or, yt = (1 + g) yt1, t = 1, 2, , 10.. (1)
Equation (1) relates the values of the variable y at two distinct periods t and
(t 1). It is an example of a difference equation.
There is a one-period lag in the values of the relevant variable (yt and yt1).
Therefore, it is an example of a first order difference equation. The order of
a difference equation is determined by the maximum number of periods
lagged. Some examples of difference equations are given below with the
orders noted.
yt-3 3y t-4= 0 order 1.
yt = a(yt1 yt2) + 10 order 2.
log yt + 9 yt+ 7 (yt+6)3 + 6yt = 0 order 9.
t 5+1
18 yt+4 yt = 2 5 order 4.
yt +3 + ayt +1 = byt 1 + c order 4
Consider a difference equation of the following form:
y = a1yt1 + a2yt2 + . + anytn + b
where a1, a2, , an and b are constants.
This is called an nth order linear constant coefficient difference equation
(assuming an 0, otherwise the order will be less than n). It is linear because
the dependent variable y is not raised to any power and there are no product
terms, constant coefficients because a1, , an are constants and do not change
with t. This equation will be homogeneous if b = 0. If b 0, then it is non-
homogenous. In this unit, we shall work only with difference equations of this
special type of orders one and two (n = 1, 2).

8.3 SOLVING FIRST ORDER DIFFERENCE

EQUATIONS
In solving a difference equation, we find a time path y(t) from a given
initial condition. As pointed out above, a first order difference equation
takes the form
yt = f (t , yt 1 ) for all t.
We can solve such an equation by successive calculation,also called
recursive method , taking the initial value of y (say y0) as given. Thus,
y1 = F (1, y0)
y2= F(2, y1) = f(2, f(1, y0))
and so on. 43
Integral Calculus and Note that given any value y0, there exists a unique solution path y1, y2, ....
Economic Dynamics
However, resorting to calculation of the solution through such a method
doesn't tell us much about the properties of the solution. We should have a
general formula, which exists if the form of f is simple.
Let us start with a first-order linear difference equation with constant
coefficient. It takes the form
yt = ayt 1 + bt where bt for t = 1, ... are constants.
When the recursive method is used, you will see a pattern as follows:
t
yt = ay0 + a t k bk ..(2)
k =1

and such an equation has a unique solution path.To check that we

get the unique solution from the above formulation, verify that it
satisfies the original equation.
Since we have
t 1

ayt 1 + bt = a a t 1 y0 + a t k bk + bt
k =1
t 1
= a t y0 + a t k bk + bt
k =1

t
= a t y0 + a t k bk
k =1

= yt ,
so that the solution obtained is correct.
Taking the equation (2), we can examine the special case of
bk = b for all k = 1..
We have
t 1
yt = aty0 + b a t 1
j =0

Making use of the result of geomtric series summation, the term

t 1

a
j =0
t 1
may be expanded as 1 + a + a2 + ... + at1 to give

1 + a + a2 + ... + an1 = (1an)/(1a).

if a 1. Thus we have
yt = aty0 + b(1 at)/(1 a)
if a 1.
For any given value y0, the unique solution of the difference equation
yt = ayt1 + b,
where a 1, is
yt = at(y0 b/(1 a)) + b/(1 a).

44
Equilibrium or Stationary Value Difference Equations and
Applications in Economic
For a given value y0, the value of yt changes with t. But there may be some Dynamics
value of y0 for which yt doesn't change. Such a solution exists if
y* = b/(1 a)
and yt is constant, equal to b/(1 a).
We call y* the equilibrium value of y and rewrite the solution as
yt = at(y0 y*) + y*.
Example: Solve yt +1 = yt + , ....(3)
where and are constants.
Look for a stationary or equilibrium value of yt over time which
can be repeated for any t consistently satisfying the above equation.
May be you consider y as an equilibrium value of yt such that
y =y +

or , y =
1
To understand the above example, we need to remember the
dynamic multiplier.
Write
Ct = yt 1 + ..(4)

Let the investment be fixed at I for every t so that we have

Yt = Ct + I t = Ct + I
= Yt 1 + ( + I )
= Yt 1 + '

where ' = + I
Use the above relation (4) we have
Yt +1 = Yt + '

If an equilibrium income Y is found, the solution can be written as

Y = Y + '
' +I
= =
1 1
1
Note that is the Keynesian multiplier.
1
It is important to remember that we have solved equation (3) for the stationary
level of yt i.e., y . There is no guarantee that the actual path of y converges to
y . In case yt approaches y , then

( yt y ) 0.
If these values of yt and y t+1 hold, we can write

gt = yt y ..(5) 45
Integral Calculus and Since yt and y satisfy (3), we have
Economic Dynamics
yt +1 = yt + and
y = y +
Thus,
yt +1 y = ( y t y ) .

From (5),
gt = yt y
or, g t+1 = yt +1 y
or , g t+1 = yt +1 y
or , gt +1 = gt ..................(6)
Since
gt +1 = gt
gt = gt 1
.
.
.
.
g1 = g0
Substituting backward,
g & +1 = 2 gt 1 = 3 gt 2 ..................
we get
gt +1 = t 1g 0
or g t = t g 0 for t = 0, 1, 2 ..............
Thus, any difference equation of the form yt = yt 1 has a
solution yt = t y0 , where y0 is the value of y at some chosen initial point.
General Solution
Suppose we intend to solve the equation
yt +1 + ayt = c................ (7)

Its general solution will be consisting of particular solution ( y p ) and

complementary function ( yc ) , i.e., y g = y p + yc . In this approach, the
y p component represents the inter-temporal equilibrium level of y while that
of yc gives the deviations if the time path from that equilibrium. The solution
is called general solution due to the presence of an arbitrary constant. In order
to get a definite solution, we need an initial condition.
Let us work with complementary function. From (7), we get its reduced from
as
yt +1 + ayt = 0 ..(8)
46
It is seen above that yt = t yo is a solution to the difference equation. In that Difference Equations and
Applications in Economic
case we have yt +1 = a t +1 y0 as well. We modify this and rewrite Dynamics

yt = Abt and y t+1 = Abt +1.

Substitution of these into (8) gives

Abt +1 + aAbt = 0
or , Abt ( b + a ) = 0

or , (b + a ) = 0
or , b = a
We must have b = a in the trial solution such that the complementary
solution can be written as
yc = Abt = A ( a ) .
t

Particular solution needs to be recasted such that it is in agreement with the

general solution. Consider the simplest value of y. If yt has an equilibrium
value k such that it remains constant overtime, we have yt = k as well as
yt +1 = k . Substitution of these values to the trial solution gives
k + ak = c
c
or , k =
1+a
Since the value, k, satisfies the equation, the particular solution can be written
as
c
yp = k = for a 0
1+ a

In case a = 1, however, the particular solution is not defined and some other
solution of (7) must be searched for.

Substituting k into (7), we get

k ( t + 1) + akt = c.
c
or , k = = C and y p = Ct . .
t+1+a t

The general solution can now be written in one of the following forms:

47
Integral Calculus and C
ytt = A ( a ) + if a 1
t
Economic Dynamics
1+ a
or , y t = A ( a ) + Ct =A+C t if a = 1.
t

Notice that the solution above still remains indeterminate. This is due to the
presence of arbitrary constant A. We have to take the help of initial condition
( yt = yo ) for eliminating it. Thus, taking t =0, we have
C
y0 = A +
1+ a
C
or , A = y9
1+ a

The definite solution therefore, becomes

C C
yt = y0 ( a ) + for a 1
t

1+ a 1+ a
or , yt = yo + Ct for a = 1

8.3.1 Behaviour of Solutions of First Order Equations

The solution of a difference equation gives an expression for the relevant
variable as an explicit function of time. In other words, a time path of the
variable is obtained. To investigate the nature of this time path of a solution of
the first order equation, we write the solution for a 1.
The behavior of the solution path depends on the value of a.
a < 1
yt converges to y* and the solution is stable. There are two
subcases:
0<a<1,
Monotonic convergence.
1 < a < 0
Damped oscillations.
a > 1
Divergence:
a>1
Explosion.
a < 1
Explosive oscillations.
To understand these features see Figure 8.1.

48
 Difference Equations and
t at Applications in Economic
a Dynamics

B B

t t

(1) a>1 (2) 0<a<1
at at

B B

t t

(3) 1<a<0
(4) a < -1
at

(5) a = 1

Fig. 8.18.1
Figure
In short,
a > 1 time path explodes (diverges)

a < 1 time path converges

a > 0 time path non-oscillating

a < 0 time path oscillating.
Thus, the condition for stability is a < 1.

The different cases are shown in Figure 8.1.

49
Integral Calculus and
Economic Dynamics
8.3.2 Economic Applications of First Order Equations
We consider three applications of the type of equations discussed in the
previous section. The first is an analysis one sector Harrod-Domar model
while the second is of price dynamics. The last one deals with the amortisation
problem of hire purchase of consumer durables.
a) Harrod-Domar One Sector Model
An economy produces one good Q with capital K through a production
function Qt = bKt, where b = constant productivity of capital. Accumulation of
capital between t and t+1 is given by
It = Kt+1 Kt, where It = investment in t.
Saving St = sQt.
Equilibrium level of income is determined at the equality of savings and
investment. So,
S t = It
or, sQt = Kt+1 Kt.
Since Qt = bKt, we have sbKt = Kt1 Kt
or, Kt+1 = (1 + sb)Kt, a homogeneous first order linear difference equation.
Therefore, solution to this equation is given by
Kt = (1 + sb)tK0.
1
Since b is productivity of capital in the model, we write = capital output
b
ratio = v (say).
t
s
Now Kt = 1 + K0 and
v
t
s
Qt = 1 + Q0
v
s
Remember that = warranted rate of growth and constituted by two basic
v
parameters s and v. We can find out the output growth rate given s and v.
b) The Cobweb Model
The essential feature of this model is that production or supply responds to
price with a one-period lag. This type of lagged supply response is often
observed for agricultural products.
We assume: 1) The market demand and supply functions are linear and do not
change over time, 2) demand in any period t responds to price prevailing in
the same period t, but supply in t depends on price that prevailed in the last
period, (t 1) and 3) the market is competitive in the sense that the price that
prevails in each period is the price that equates demand and supply. Thus, the
model can be set out as consisting of the following equations.
Dt = a b Pt; a, b > 0
St = Pt 1; , > 0, < a
Dt = St for all t.

50
The first equation gives us the simple demand curve in period t. The second Difference Equations and
Applications in Economic
displays the lag in supply. Supply in t, St, is determined by prices of the Dynamics
immediately preceding period, Pt 1. The last equation is the condition of
market clearing in each period. The three equations together yield a first order
constant coefficient non-homogeneous difference equation in price.
a
Pt = P t 1 + (9)
b b
With a, b , known, a specification of the initial price P0 allows us to solve
the equation as:
a a
t

Pt = P0 + (10)
b+ b b+
From our previous discussion, it is clear that the behaviour of P over time

depends crucially on the term .
b
Since this term is negative (b, > 0) the time path will always be oscillatory.
a
Let us denote the constant by P*.
b+
Then

>1 Price diverges
b

=1 Price oscillates uniformly
b

<1 Price converges to P*.
b
Only in the last case (Pt approaches P* as t increases), the system is stable.
1
Thus, the condition for stability is < 1. Since graphically is the
b
1
slope of the supply curve and that of the demand curve in absolute
b
value, the stability condition states that the slope of the supply curve must be
steeper than the absolute value of the slope of the demand curve.
a
At this point, we pause to note the significance of the value . This is
b+
the constant value of price that is a solution of the equation (9). To check,
substitute Pt = Pt 1 = P* (a constant) in (9).
a
P* = P* +
b b
a
or, P* = .
b+

51
Integral Calculus and a
Economic Dynamics Thus, Pt = is a solution of (9). This type of constant solution is called
b+
Stationary solution. The price P* may be called the equilibrium price
because it equates demand and supply and stays unchanged over time.
Example: We want to investigate the behaviour of price in a market with the
demand and supply functions:
Dt = 86 0.8 Pt
St = 10 + 0.2 Pt 1
Assuming market clearing in each period (Dt = St) we have
(0.8) Pt = 0.2 Pt 1 96
or, Pt = (0.25)Pt 1 + 120
The solution is
120 120
Pt = P0 (0.25) +
t

1 + 0.25 1 + 0.25
= (P0 96) (0.25)t + 96
Since -0.25 = 0.25 < 1, the time path of P is oscillating but converges. The
market is stable and with the passage of time price approaches the equilibrium
value 96.
c) The Amortisation Problem
We are all familiar with the practice of hire purchase or purchase by
instalments of consumer durables like refrigerators, cars or T.V. sets. The
buyer pays a part of the price at the time of purchase (the down payment) and
pays the rest in monthly or annual instalments over a specified period.
Because the payments are spread over a period of time, an interest cost is
included in the value of instalments. Amortisation is the term associated with
this method of repaying an initial debt plus interest charges by a series of
payments of equal magnitude at equal intervals.
Let the value of the article purchased by V and P the down payment. Then the
initial debt of the buyer is D0 = V P. The contract states that the debt, D0 is
to be paid off over T periods. The rate of interest is r (100 r%). The question
we are interested in is: how is the magnitude of periodic instalment to be
determined?
Let us denote the value of the instalment (still unknown) by B. This value
stays constant over time. The outstanding debt Dt at the end of the period t
obeys the equation
Dt = (1 + r)Dt 1 B ..(11)
This simply says that to find the outstanding debt at the end of the tth period
you take the debt outstanding at the end of the previous ((t 1)th) period Dt 1,
add the interest charge on it, rDt 1, but subtract the payment B made in that
period. Given the initial debt of D0 the solution of (11) is
B B
Dt = D0 + (1 + r)t + . (12)
r r
The value of B is to be selected so that the debt disappears at the end of period
T, that is, Dt = 0. From (12) we get

52
 B B Difference Equations and
D0 + (1 + r) + = 0
t
Applications in Economic
r r Dynamics

rD0
or, B =
1 (1 + r ) T
Thus, we have the exact relationship between the magnitude of the periodic
payment and the rate of interest, the magnitude of the initial debt and the time
horizon of the contract. The expression:
1 (1 + r ) T

r
is referred to as the amortisation factor and value of this factor has been
extensively tabulated for different values of r and T.
Check Your Progress 1
1) What is a difference equation? Distinguish it from a differential equation.





2) Discuss the nature of the following time paths
t
1
i) yt = 3t + 1 (ii) yt = 5 + 3
10





3) Suppose you find the following the path of y.
yt = Aat + B; A < 0, B > 0.
Draw the different cases of the behaviour of yt for different values of a.
4) Solve the following equations:
1
i) yt +1 yt = 6 for y0 = 1
3
ii) yt +1 yt = 3 for y0 = 5





53
Integral Calculus and
Economic Dynamics 8.4 SOLVING SECOND ORDER DIFFERENCE
EQUATIONS
A general second-order difference equation which we have already
mentioned at begniging outset of this unit takes the form
yt+2 = f (t, yt, yt+1).
Just as in the case of first-order equation, a second-order equation will
have a unique solution and can be derived by successive (recursive)
calculation. We will show that given y0 and y1 there exists a uniquely
determined value of yt for all t 2. Note that for a second-order equation
we need two starting values, y0 and y1, in place of one taken in the first
order counterpart.
8.4.1 Homogeneous Equations
Consider the following second order constant coefficient equation
yt + 2 + ayt +1 + byt = 0 . (13)
We need to find two solutions of the equation above.
If we make a guess that the solution takes the form ut = mt
In order for ut to be a solution, we must have
mt(m2 + am + b) = 0
or, if m 0,
m2 + am + b = 0.
This is called the characteristic (or auxiliary) equation of the difference
equation and its solutions are
(1/2)a ((1/4)a2 b).
8.4.2 Behaviour of Solutions of Homogeneous Equations
Looking at the component ((1/4)a2 b), we distinguish three cases:
i) Distinct real roots
If a2 > 4b, the characteristic equation has distinct real roots, and the
general solution of the homogeneous equation is
Amt1 + Bm2t ,
where m1 and m2 are the two roots.
ii) Repeated real root
If a2 = 4b, then the characteristic equation has a single root, and the
general solution of the homogeneous equation is
(A + Bt)mt,
where m = (1/2)a is the root.
ii) Complex roots
If a2 < 4b, then the characteristic equation has complex roots, and the
general solution of the homogeneous equation is
Art cos(t + ),

54
where A and are constants, r = b, and cos = a/(2b), or, alternatively, Difference Equations and
Applications in Economic
C1rt cos(t) + C2rt sin(t), Dynamics

where C1 = A cos and C2 = A sin (using the formula that cos(x+y) =

(cos x)(cos y) (sin x)(sin y).
When the characteristic equation has complex root, the solution oscillates.
Art is the amplitude (which depends on the initial conditions) at time t, and
r is growth factor. /2 is the frequency of the oscillations and is the
phase (which depends on the initial conditions).
If r < 1 then the oscillations are damped; if r > 1 then they are
explosive.
Stability
We say that a system of differential equations is stable if its long-run
behavior is not sensitive to the initial conditions.
Consider the second-order equation
yt + 2 + ayt +1 + byt = ct .
Write the general solution as
yt = Aut + Bvt + ut*,
where A and B are determined by the initial conditions.
This solution is stable if the first two terms approach 0 as t , for all
values of A and B. In this case, for any initial conditions, the solution of the
equation approaches the particular solution ut*. If the first two terms
approach zero for all A and B, then ut and vt must approach zero. You can
take A = 1 and B = 0 to see that ut approaches zero. On the other hand, take
take A = 0 and B = 1 to see that vt approaches 0. A necessary and sufficient
condition for this to be so is that the moduli of the roots of the characteristic
equation be both less than 1. Note that the modulus of a complex number
+ i is +(2 + 2), which is the absolute value of number if the number is
real.
There are two cases:
If the characteristic equation has complex roots then the modulus of
each root is b (the roots are i, where = a/2 and = (b
(1/4)a2)). So for stability need b < 1.
If the characteristic equation has real roots then the modulus of each
root is its absolute value. So for stability we need the absolute values of
each root to be less than 1, or a/2 + (a2/4 b) < 1 and a/2
(a2/4 b) < 1.
8.4.3 Non-homogeneous Equations
To find the general solution of the original equation
yt+2 + ayt+1 + byt = ct
we need to find one of its solutions. Suppose that b 0.
The form of a solution depends on ct.
Suppose that ct = c for all t. Then yt = C is a solution if C = c/(1 + a + b) and
if 1 + a + b 0;
55
Integral Calculus and if 1 + a + b = 0 then try yt = Ct; if that does not yield a solution, we have to
Economic Dynamics
try yt = Ct2.
8.4.4 An Economic Application of Second Order Non-
homogeneous Equation
We discuss now an economic example of a second order non-homogeneous
equation. This is Samuelsons model of interaction between the multiplier and
the accelerator. Consider the following macro-economic equations:
Ct = C0 + cYt 1, 0 < c < 1.
It = I0 + v(Ct Ct1); v > 0.
Yt = Ct + It
The symbols Y, C, I stand for national income, consumption and investment
respectively. The first equation is the consumption function with a one-period
lag; the second is the investment function of the accelerator type. C0 + I0 are
the levels of autonomous consumption and investment. The marginal
propensity to consume c and the accelerator coefficient v are assumed to be
constant. The final equation is the condition of macro balance. The three
equations together generate the following difference equation in Y
Yt c(1 + v) Yt 1 + cvYt 2 = C0 + I0 (15)
The characteristic equation for the homogeneous part is
m2 c(1 + v) m + cv = 0
The roots are
1
m1, m2 = (c(1 + v) c 2 (1 + v) 2 4cv ) (16)
2
Both m1 and m2 are positive because from the theory of quadratic equations
we know m1 + m2 = c(1 + v) > 0 and m1m2 = cv > 0. Since c(1 + v) cv 1,
C + I0
the particular solution is 0 . Three types of solution are possible
1 c
depending on the values of c and v.
1) c2(1 + v)2 > 4cv
or, c(1+v)2 > 4v, the roots are real and distinct.
Here,
C0 + I 0
Yt = A1 m1t + A2 m 2t + ; A1, A2 0 and constants
1 c
1
2) c(1+v)2 = 4v, the roots are real and equal with value c(1+v). In this
2
case
c(1 + v) 2 C + I0
Yt = (A1 + A2t) ( )+ 0 ;
2 1 c
3) c(1+v)2 < 4v, the roots are complex. From (16) we see that the roots are
of the form (a ib) with
1
a= c(1+v)
2

56
 1 Difference Equations and
b= 4cv c 2 (1 + v) 2 Applications in Economic
2 Dynamics

( c 2 (1 + v) 2 4cv = i 2 (4cv c 2 (1 + v) 2 ) = i 4cv c 2 (1 + v) 2 = ib)

The modulus of the roots

r= a2 + b2 = cv
The solution is
C0 + I 0
Yt = ( cv )2 (A1 cos(t) + A2 sin (t)) +
1 c
4cv c 2 (1 + v) 2
where = tan1
c(1 + v)

In this case, we have a cyclical time path of national income Y. If cv < 1,
then ( cv )t will tend to zero as t increases and Yt will approach the value
C0 + I 0
.
1 c
Thus, the condition for stability (damped oscillations in Y) is cv < 1, that is,
the product of the marginal propensity to consume and the accelerator
coefficient should be less than unity.
Check Your Progress 2
1) Solve the following difference equations and determine whether the
solution paths are convergent or divergent, oscillating or not.
a) yt+2 + 3yt+1 (7/4)yt = 9.
b) yt+2 2yt+1 + 2yt = 1.
c) yt+2 yt+1 + (1/4)yt = 2.
d) yt+2 + 2yt+1 + yt = 92t.
e) yt+2 3yt+1 + 2yt = 35t + sin((1/2)t).





2) Find the roots of the equation
yt = ayt 1 + byt 2
Examine when the roots are
1) real, unequal
2) real, equal
3) complex
What is the auxiliary or the characteristic equation of the equation above?
What are the final forms of general solution of the equation in each case?
57
Integral Calculus and
Economic Dynamics




3) Find the solutions of the equations:
a) yt + 4yt 2 = 0, y = 12, 11 at t = 0, 1 respectively.
4
b) yt = 2yt 2 yt 2 , y = 0, 1 at t = 0, 1 respectively.
3

8.5 LET US SUM UP

In continuation with the theme on solving economic problems in a dynamic
set up, the present unit took up time as a discrete independent variable and
examined tool of simple difference equations. In the process, we considered
the solutions of first and second order linear difference equations covering
homogeneous and non-homogeneous cases. To see the applications of these
equations to economic problems, the time path of adjustment of macro-
economic variable national income in case of the simple Keynesian
multiplier model and Samuelsons multiplier-accelerator interaction model
were discussed. We also examined the time path of adjustment of the price
variable and looked into the conditions of dynamic stability of the different
systems i.e., whether over time, the economic variables price or national
income converge to a stable equilibrium. Finally, the conditions when the
systems become dynamically explosive i.e., the variables move further and
further away from the equilibrium value were examined.

8.6 KEY WORDS

Amortisation: It is the term associated with the method of repaying an initial
debt plus interest charges by a series of payments of equal magnitude at equal
intervals.
Cobweb Model: A model where production or supply responds to price with
a one-period lag. This model is often used to analyse the demand-supply
mechanism for markets of agricultural commodities.
Constant Coefficient Difference Equation: A difference equation has
constant coefficient if the coefficients ais associated with the y values are
constant and do not change over time.
Difference Equation: A difference equation is an equation involving the
values of an unknown function y(x) for different values of x. The independent
variable time in problems of economic dynamics takes only discrete
values. The form of the equation is, yt = a1yt1 + a2yt2 + + anyn1 + b, where

58
a1, a2, ., an and b are constants, is an example of an n-th order linear, Difference Equations and
Applications in Economic
constant coefficient, difference equation. Dynamics
Homogeneous Difference Equation: A difference equation is homogeneous
if the constant term b is zero.
Linear Difference Equation: A difference equation is linear if (i) the
dependent variable y is not raised to any power and (ii) there are no product
terms.
Non-homogeneous Difference Equation: A difference equation is non-
homogeneous if the constant term, b, is non-zero.
Order of a Difference Equation: It is determined by the maximum number
of periods lagged.

8.7 SOME USEFUL BOOKS

Allen, R.G.D., 1959, Mathematical Economics (Second Edition) St. Martins
Press Inc., New York.
Baumol, W.J., 1974, Economic Dynamics (Second Edition) Macmillan, New
York. Chapters 9, 10 and 11.
Chiang, Alpha C. 1984, Fundamental Methods of Mathematical Economics
(Third Edition): Mc-Graw Hill International Edition, New Delhi.

8.8 ANSWER OR HINTS TO CHECK YOUR

PROGRESS
Check Your Progress 1
1) See Section 8.1
2) i) Non-oscillatory; divergent
ii) Oscillatory; convergent
3) Fig. 8.1 shows the time path for A > 0. Draw the corresponding figures
for A < 0.
t
1
t
1
4) i) yt = 8 + 9 ; ii) yt = 2 + 4
3 4
Check Your Progress 2
1)

7
t t
1
a) A1 + A2 + 4 . Nonconvergent oscillations.
2 2

( 2 ) A cos 4 t + A sin 4 t + 1 .Nonconvergent oscillation.

t
b) 1 2

t t
1 1
c) A1 + A2t + 8. Convergent, non-oscillating.
2 2
d) The characteristic equation is m 2 + 2m + 1 = (m + 1) 2 = 0 , which has
a double root of 1. So the general solution of the homogeneous
equation is yt = ( C1 + C2t ) (1)t . A particular solution is obtained

59
Integral Calculus and
Economic Dynamics
by inserting ut * = A2t , which yields A = 1. So the general solution of
the inhomogeneous equation is yt = ( C1 + C2t )( 1) + 2t.
t

e) By using the method of undetermined coefficients the constants A,

B, and C in the particular solution

u* = A5t + B cos t + C sin t , we obtain A = 1/4, B = 3/10,
2 2
and C = 1/10. So the general solution to the equation is
1 3 1
yt = C1 + C2 2t + 5t + cos t + sin t .
4 10 2 10 2
2) See Section 8.4 and answer.
Note that in the text yt = ayt 1 + byt 1
Here you have a slightly changed equation.
t 11 t
3) a) yt = 2t 12 cos + sin .
2 2 4

2 3
t

b) yt = sin(t )

3 2 sin
1
where = tan-1
3

8.9 EXERCISES
1) Investigate the behaviour of price in a market, i.e., the stability of a
system with demand and supply functions:
a) Dt = 86 0.8 Pt
St = 10 + 0.8 Pt 1
b) Dt = 86 0.8 Pt
2) What is amortisation? Derive the exact relationship between the
magnitude of the periodic payment B and the rate of interest r, the
magnitude of the initial debt D0 and the time horizon of the contract T.
3) Establish the stability condition of Samuelsons multiplier-accelerator
interaction model.
t
4
4) Find the time path represented by the equation yt = 2 + 9 .
5
5) Find the solution of the equation yt +1 + 14 yt = 5 for y0 = 2 .
5) The demand and supply for cobweb model is given as
Qdt = 19 6 Pt and Qst = 6 Pt 1 5 . Find the intertemporal equilibrium
price and comment on the stability of the equilibrium.
Answer or Hints to Exercises
1) Note that for the Cobweb model
Dt = a b P t a, b > 0
St = + Pt 1 , > 0, < 0
60
 Dt = St for all t. Difference Equations and
Applications in Economic
A specification of the initial price P0 allows us to solve the equation. Dynamics

a
Pt = Pt 1 +
b b

a a
t

as Pt = P0 +
b + b b +

a
t

or, Pt = ( )
P0 P + P * ; P * =
*

b b+

If 0 as t , Pt P* , the equilibrium value.

b

This is possible if and only if <1
b
i.e., < b.
a) = 0.8, b = 0.8. Hence = b.
1 1
This results in uniform oscillation, as = .
b
or, the slope of the supply curve = the absolute slope of the demand curve.
b) b = 0.8, = 0.9

Hence =1
b
or, > b.
1 1
or, < .
b
i.e., the slope of the supply curve is less than the absolute value of the slope
of the demand curve.
Hence, price diverges further and further away from the equilibrium and you
come across an explosive and oscillatory situation.
c) B = 0.9, = 0.8

< 1 implies damped oscillation and the system is stable.
b
You should draw diagrams in each case and satisfy yourself.
2) See Example (b) in Sub-section 8.3.4.
3) Sub-section 8.4.4 and answer.
4 4
4) Since b = < 0, the time path is oscillatory. As b = < 1 , the
5 5
oscillation is damped and it converges to equilibrium level of 9.
yt = 2 ( 14 ) + 4.
t
5)

6) p = 2 ; discuss on uniform oscillation.

61