Chapter 19

Non-Linear, First-Order Difference Equations

In this chapter, we learn the method to
find out qualitative properties of non-linear
first-order autonomous difference equations.
Basically, we are interested in finding steady-
state solutions and their stability properties.
In order to so, we make use of phase dia-
gram, which basically involves plotting yt+1
against yt . In the case of non-linear difference
equation there can be more than one steady
state (multiplicity of equilibria).
The general form of non-linear first-order
autonomous difference equation is

yt+1 = f (yt ); f 0 (yt ), f 00 (yt ) 6= 0. (19.1)

The steady state solutions are given by

y = f (y). (19.2)

The existence and convergence properties of
steady-states depend on the f 0 (yt ) & f 00 (yt ).
In phase diagram, steady state values are gi-
ven by points at which f (yt ) intersects 450
line (curve depicting yt+1 = yt ).

Theorem 19.1: A steady-state equilib-
rium is locally stable if the |f 0 (y)| < 1.
It is unstable otherwise.

Theorem 19.2: A first-order difference
equation leads to oscillation in yt if the
derivative f 0 (yt ) < 0, ∀yt > 0, but m-
oves monotonically if the derivative
f 0 (yt ) > 0, ∀yt > 0.

Solow Model Revisited

Suppose that aggregate output, yt , is gi-
ven by

yt = ktα , 0 < α < 1
where kt is the capital stock at time t. The
capital stock evolves over time as follows:

kt+1 = (1 − δ)kt + It , 0 < δ < 1

where δ is the depreciation rate and It is the
investment at time t. The economy saves a
constant proportion, s, of output every pe-
riod. Thus savings, St , is given by

St = syt .
Imposing the equilibrium condition St =
It , we derive a non-linear difference equation
in capital stock, which tells us about the im-
plications of Solow growth model:

kt+1 = (1 − δ)kt + sktα .
The difference equation has two steady states
³ s ´ 1−α

k = 0, .
¡ s ¢ 1−α
k = 0 is unstable and k = δ is stable.

Stable Limit Cycles

So far we have considered the case in
which f 0 (y) does not change sign and thus
f (y) is either increasing or decreasing. Now
suppose that f 0 (y) can change its sign and
f 00 (y) < 0. Such nonlinear function produces
hill-shaped curve.
Now further suppose that f 0 (y) < −1,
then the steady state is not stable. However,
the value of yt will not diverge either to 0
or ∞. Rather it will oscillate around steady
state within finite bounds. In certain cases,
yt converges to a path that cycles back and
fourth in successive periods between two val-
ues. In this case, yt converges to stable limit
cycle of two period.