Professional Documents
Culture Documents
1. Assumptions
Firms
Households
2. The Behaviour of the Households and Firms
Household’s Budget Constraint
Household’s Maximisation Problem
Household Behaviour
3. The Dynamics of the Economy
The Dynamics of c
The Dynamics of k
Phase Diagram
The Initial Value of c
4. Welfare
5. The Effects of a Fall in the Discount Rate
Qualitative Effects
The Rate of Adjustment and the Slope of the Saddle Path
6. The Effects of Government Purchases
Adding Government to the Model
The Effects of Permanent Government Purchases
1. Assumptions
Firms
• There are a large number of identical firms. Each has access to the production function
Y = F ( K , AL) , which satisfies the same assumptions as the Solow Model.
• The firms hire workers and rent capital in competitive factor markets, and sell their
output in a competitive output market.
• Firms take A as given; as in Solow model. A grows exogenously as rate g.
• The firms maximise profits. They are owned by the households, as any profits they earn
accrue to the households.
Households
1
The household’s utility function takes the form
∞ L(t )
(1.1) U = ∫ e − ρ t u (C (t )) dt
t =0 H
C (t )1−θ
(1.2) u (C (t )) = , θ > 0, ρ − n − (1 − θ ) g > 0
1−θ
This functional form is needed for the economy to converge to a balanced growth path. It is
known as a constant-relative-risk-aversion (or CRRA) utility. The reason for the name is that the
coefficient of relative risk aversion (which is defined as −Cu ′′(C ) / u ′(C ) ) for this utility function
is θ , and thus is independent of C.
Since there is no uncertainty in this model, the household’s attitude toward risk is not
directly relevant. But θ also determines the household’s willingness to shift consumption
between different periods: the smaller is θ , the more slowly marginal utility falls as
consumption rises, and so the more willing the household is to allow its consumption to vary
over time. If θ is close to zero, for example, utility is almost linear in C, and so the household is
willing to accept large swings in its consumption to take advantage of small difference between
its discount rate and the rate of return it gets on its saving. Specifically, one can show that the
elasticity of substitution between consumption at any two points in time is 1/ θ .
Three additional features on the instantaneous utility function are worth mentioning.
First, C 1−θ is increasing in C if θ < 1 but decreasing if θ > 1 ; dividing C 1−θ by 1 − θ thus ensures
that the marginal utility of consumption is positive regardless of the value of θ . Second, in the
special case of θ → 1 , the instantaneous utility function simplifies to ln C ; this is often a useful
case to consider. And third, the assumption that ρ − n − (1 − θ ) g > 0 ensures that lifetime utility
does not diverge: if this condition does not hold, the household can attain lifetime utility, and its
maximisation problem does not have a well-defined solution.
Firms
Firms behave relatively simple. At each point in time they employ the stocks of labour and
capital, pay them their marginal products, and sell the resulting output. Because the production
function has constant returns and the economy is competitive, firms earn zero profits.
As described in Solow Model, the marginal product of capital, dF ( K , AL) / dK , is f ′( k ),
where f (•) is the intensive form of the production function. Because markets are competitive,
2
capital earns its marginal product. And because there is no depreciation, the real rate of return on
capital equals its earnings per unit time. Thus the real interest rate at time t is
(1.3) r (t ) = f ′( k (t ))
(1.5) w(t ) = f ( k (t )) − k (t ) f ′( k (t ))
Since the household has L(t)/H members, its labour income at t is W(t)L(t)/H, and its
consumption expenditures are C(t)L(t)/H. Its initial wealth is 1/H of total wealth at time 0, or
K(0)/H. The household’s budget constraint is therefore
In many cases, it is difficult to find the integrals of this expression. But we can express the
budget constraint in terms of the limiting behaviour of the household’s capital holdings; even
when it it’s not possible to compute the integrals in (1.6), it is often possible to describe the
limiting behaviour of the economy. To see how the budget constraint can be rewritten in this
way, first bring all the terms of (1.6) over to the same side and combine the two integrals; this
gives us
K (0) ∞ L(t )
(1.7) + ∫ e − R (t ) [W (t ) − C (t ) ] dt ≥ 0
H t = 0 H
We can write the integral from t=0 to t= ∞ as a limit. Thus, (1.7) is equivalent to
3
⎡ K (0) L(t ) ⎤
e − R (t ) [W (t ) − C (t ) ]
s
(1.8) lim ⎢
s →∞ ⎣ H
+ ∫t = 0 H ⎥⎦
dt ≥ 0
K (0)
To understand (1.9), note that e R ( s ) is the contribution of the household’s initial wealth to
H
L(t )
its wealth at s. The household’s saving at t is [W (t ) − C (t )] (which may be negative);
H
e R ( s ) − R ( t ) shows how the value of that saving changes from t to s.
The expression in (1.9) is e R ( s ) times the expression in brackets in (1.8). Thus we can
write the budget constraint simply as
K ( s)
(1.10) lim e − R ( s ) ≥0
s →∞ H
Expressed in this form, the budget constraint states that the present value of the household’s
asset cannot be negative in the limit. Equation (1.10) is known as the no-Ponzi-game condition.
• The representative household wants to maximise its lifetime utility subject to its budget
constraint.
• As in the Solow model, it is easier to work with variables normalised by the quantity of
the effective labour. To do this, we need to express both the objective function and the
budget constraint in terms of consumption and labour income per unit of effective labour.
We start with the objective function. Define c(t) to be consumption per unit of effective labour.
Thus C(t), consumption per worker, equals A(t)c(t). The household’s instantaneous utility, (1.2),
is therefore
(1.11) =
1−θ 1−θ
1−θ
⎡⎣ A(0)e gt ⎤⎦ c(t )1−θ
=
1−θ
c(t )1−θ
= A(0)1−θ e(1−θ ) gt
1−θ
4
Substituting (1.11) and the fact that L(t ) = L(0)e nt into the household’s objective function, (1.1)-
(1.2), yields
∞ C (t )1−θ L(t )
(1.12) U = ∫ e− ρt dt
t =0 1−θ H
∞ c(t )1−θ
= B ∫ e− β t dt
t =0 1−θ
A(t ) L (t ) equals A(0) L(0)e( n + g )t . Substituting this fact into (1.13) and dividing both sides by
A(0) L (0) / H yields
∞ ∞
(1.14) ∫t =0
e − R (t ) c(t )e( n + g )t dt ≤ k (0) + ∫ e− R (t ) w(t )e( n + g )t dt
t =0
Finally, because K(s) is proportional to k ( s )e( n + g ) s , we can rewrite the no-Ponzi-game version of
the budget constraint, (1.10) as
(1.15) lim e − R (t ) e( n + g ) s k ( s ) ≥ 0
s →∞
Household Behaviour
5
∞ ∞
∫ t =0
e − R ( t ) c(t )e( n + g )t dt ≤ k (0) + ∫ e − R ( t ) w(t )e( n + g ) t dt
t =0
⎣ ∫t =0 t =0 ⎥⎦
t
= ln λ − ∫ r (τ )dτ + (n + g )t
τ =0
t
where we use R (t ) = ∫ r (τ )dτ .
τ =0
c&(t )
(1.19) −β − θ = −r (t ) + (n + g )
c(t )
c&(t ) r (t ) − n − g − β
(1.20) =
c(t ) θ
= r (t ) − ρ −θ g
θ
where we use β = ρ − n − (1 − θ ) g .
Euler equation
Since C(t) (consumption per worker)=c(t)A(t). Thus, C ‘s grows rate is: growth rate of c plus the
growth rate of A. Which is [ r (t ) − ρ ] / θ
The Dynamics of c
(Since r (t ) = f ′(k (t ))
c&(t ) f ′( k (t )) − ρ − θ g
(1.23) =
c(t ) θ
∗
Thus c& =0 if f ′(k ) = ρ + θ g . Let k be this level of k .
When k > k ∗ → f ′(k ) < p + θ g , and → c& <0; when k < k ∗ , c& >0.
6
c •
c=0
• •
c>0 c<0
k∗ k
FIGURE 1.1 The dynamics of c
The Dynamics of k
•
(1.24) k (t ) = f ( k (t )) − c(t ) − (n + g )k (t )
As in Solow model:
•
For a given k, when k = 0 , c = f (k ) − (n + g )k ;
This c ↑ until f ′(k ) = (n + g ) (the golden rule level of k) and c ↓.
•
When c >c in k = 0 , k ↓; and vice versa.
7
•
c k <0
•
k =0
•
k >0
•
∗
In Figure 1.3: k (the level of k that implies c = 0 )<the golden-rule level of k (the value of k
•
associated with the peak of the k =0 locus).
Why?
8
•
c c=0
∗
k k
FIGURE 1.3 The dynamics of c and k
9
•
c=0
c
A
E
E
•
k =0
C
F
D k
k∗
FIGURE 1.4 The behaviour of c and k for various initial values of c
But: there should be budget constraint and requirement that the economy’s capital stock not be
negative.
c ↑ and k<0.
Impossible. When output is zero when k is zero, c must drop to zero. This means that households
are not satisfying their inter-temporal optimisation condition (1.23).
− R (t ) ( n + g ) s
B) If starting below F: recall (1.15): lim e e k (s) ≥ 0 .
s →∞
10
The Saddle Path
•
c=0
E
•
k =0
k∗ k
FIGURE 1.5 The saddle path
Qualitative Effects
ρ enters the equation for c& but not the one for k& . Thus, if ρ changes, only c& = 0 shifts.
Recall (1.23): c&(t ) / c(t ) = [ f ′(k (t ) − ρ − θ g ] / θ .
11
•
c c=0
E*
E
•
k =0
∗ ∗
kOLD k NEW k
FIGURE 1.6 The effects of a fall in the discount rate
dc& dc&
(1.25) c& ⎡⎣ k − k ∗ ⎤⎦ + ⎡⎣c − c∗ ⎤⎦
dk dc
dk& dk&
(1.26) k& [ k − k ∗] + ⎡⎣c − c∗ ⎤⎦
dk dc
c% = c − c∗ and k% = k − k .
∗
Let
c∗ k∗ &
Since and are const, then c&% = c& and k% = k& .
dc& % dc&
(1.27) c&% = k + c%
dk dc
12
Since c& = {[ f ′( k ) − ρ − θ g ] / θ } (1.23).
f ′′(k ∗ )c∗ %
c&% k
(1.29)
θ
Since k& = f (k ) − c − (n + g )k .
&
(1.30) k% = ⎡⎣ f ′(k ∗ ) − (n + g )k% ⎤⎦ − c%
= ⎡⎣( ρ + θ g ) − (n + g )k% ⎤⎦ − c%
= β k% − c%
&
k% c%
(1.32) β− %
k% k
Let µ = c&% / c% . Equation (1.31) implies
c% f ′′(k ∗ )c∗ 1
=
(1.33)
k% θ µ
&
From (1.32), the condition that c&% / c% = k% / k% is thus
f ′′( k ∗ )c∗ 1
(1.34) µ=β−
θ µ
f ′′( k ∗ )c∗
(1.35) µ − βµ +
2
=0
θ
This is a quadratic equation in µ . The solutions are
1/ 2
β + ⎡⎣ β 2 − 4 f ′′(k ∗ )c∗ / θ ⎤⎦
(1.36) µ=
2
13
Let µ1 and µ2 denote these two values of µ .
If µ >0, then c% and k% are growing away from (k ∗ , c∗ ) . Thus when (k, c)
→ (k
∗
, c∗ ) , then µ <0.
Since f ′′(•) <0, then c% and k% are oppositely linked to µ . Thus AA is positively sloped
•
c c=0
A
B k =0
k∗ k
14
7. The Effects of Government Purchases
.
c c=0
•
•
E* k =0
15