1516168637ECO P12 M22 E-Text
1516168637ECO P12 M22 E-Text
1516168637ECO P12 M22 E-Text
Subject ECONOMICS
TABLE OF CONTENTS
1. Learning Outcomes
2. Introduction
3. The Model
4. Representative Household
5. Representative Firm
6. The Government
7. The Balanced Growth Path
8. Linearizing around the Balanced Growth Path
9. Summary and Conclusions
1. Learning Outcomes
After studying this module, you shall be able to
2. Introduction
The stochastic growth models are typically termed as stochastic versions of the neo-
classical growth model with micro-foundations. It also provides a further insight in to the
modern macroeconomic models. While the deterministic models assume that the outcomes
are more certain if the inputs to the model are fixed, the stochastic models takes in to
account of the varying behaviour characteristics due to uncertainty. A stochastic model has
one or more stochastic element. The system having stochastic element is generally not
solved analytically. But any realistic application in macroeconomics requires models to be
stochastic. This is particularly true if we have a long-term ambition to bring
macroeconomic models to the data. The concepts and techniques involved in handling
stochastic models are no more complex than the deterministic case. Campbell (1994) has
suggested to linearize the model around the steady state to solve the stochastic models and
then solve the linearized model with the method of undetermined coefficients. Hence, in
this module first, we will discuss the baseline neo-classical growth model (with complete
markets) augmented with stochastic productivity shocks, first studied by Brock and
Mirman (1972); then we will discuss how stochastic growth models can be useful in
understanding the process of take-off from low growth and to sustained growth.
3. The Model
Brock and Mirman provided the first optimising growth model with unpredictable
(stochastic) shocks. Brock and Mirman focused on the optimal growth problem and solved
for the social planner’s maximization problem in a dynamic neoclassical environment with
uncertainty. With competitive and complete markets, the equilibrium growth path is
identical to the optimal growth path. To start with we can introduce the stochastic aggregate
productivity term in the neo-classical production function.
where z(t) denotes a stochastic aggregate productivity term showing how productive a
given combination of capital and labour will be in producing a final good of the economy.
In most of the neo-classical growth models under uncertainty, it is assumed that the
stochastic shock is a labour augmenting productivity term. So the term aggregate
production function becomes
𝐾(𝑡)
Where, 𝑘(𝑡) ≡ corresponding to the capital labour ratio. The capital stock depreciates
𝐿(𝑡)
at a constant rate ′𝛿′.
It is further assumed that the higher values of stochastic shock ‘z’ correspond to greater
productivity at all capital labour ratios i.e. if 𝑗 > 𝑗 ′ , then 𝑓(𝑘, 𝑧𝑗 ) > 𝑓(𝑘, 𝑧𝑗′ ).
Now, taking a representative household that tries to maximise its utility function 𝑢(𝑐).
4. Representative Household
The representative household supplies one unit of labour in an inelastic manner, so that
K(t) and k(t) can be used interchangeably. The consumption and saving decisions at time
‘t’ are made after realizing the stochastic shock for the same time with consumption c(t).
Hence, the objective function is
∞
𝑚𝑎𝑥𝐸0 ∑ 𝛽 𝑡 𝑢(𝑐(𝑡))
𝑡=0
The utility from current and future consumption of the representative household is defined
as
1 𝑠−𝑡
𝑈𝑡 = 𝐸𝑡 [∑∞
𝑠=𝑡 (1+𝜌) ln 𝐶𝑠 ],
𝑋𝑠+1 = 𝑋𝑠 (1 + 𝑟𝑠 ) + 𝑤𝑠 𝐿 − 𝑇𝑠 − 𝐶𝑠
Assuming that the households do not incur ever increasing debts, therefore, over the
infinitely long horizon, the present discounted value of the household’s assets must be zero
(termed as the transversality condition):
𝑠
1
lim 𝐸𝑡 [(∏ )𝑋 ] = 0
𝑠→∞ 1 + 𝑟𝑠′ 𝑠+1
𝑠′ =𝑡
1
𝑈𝑡 (𝑋𝑡 ) = max {ln 𝐶𝑡 + 1+𝜌 𝐸𝑡 [𝑈𝑡+1 (𝑋𝑡+1 )]} ……………..(1)
{𝐶𝑡 }
Subject to
𝑋𝑡+1 = 𝑋𝑡 (1 + 𝑟𝑡 ) + 𝑤𝑡 𝐿 − 𝑇𝑡 − 𝐶𝑡
1 1+𝑟𝑡+1 1
− 𝐸𝑡 [ .𝐶 ]=0 ……..(4)
𝐶𝑡 1+𝜌 𝑡+1
𝑌𝑡 = 𝐶𝑡 + 𝐼𝑡 + 𝐺𝑡 ………(5)
Therefore, we can construct a function for representative firm as well as the government
sector.
5. Representative Firm
On the production side, let us take a representative firm, which has a production function
of Cobb-Douglas type and assuming that the aggregate productivity term is labour
augmenting the aggregate production function becomes:
‘Y’ is aggregate output, ‘K’ is the aggregate capital stock, ‘L’ is the aggregate labour
supply and ‘A’ is a technology parameter. The subscript ‘t’ denotes the time period.
The aggregate capital stock depends on aggregate investment ‘I’ and the depreciation rate
′𝛿′:
As discussed before, here the productivity parameter ‘A’ follows a stochastic path with
trend growth ‘g’:
ln 𝐴𝑡 = ln 𝐴∗𝑡 + 𝐴̂𝑡
𝐴̂𝑡 = ∅𝐴 𝐴̂𝑡−1 +∈𝐴,𝑡 with |∅𝐴 | < 1 …….(8)
∗ ∗
𝐴𝑡 = 𝐴𝑡−1 (1 + 𝑔)
Here, the stochastic shock ∈𝐴,𝑡 is independent and identically distributed (i.i.d.) with mean
zero; variables evaluated on the balanced growth path are denoted by a *; and loglinear
deviations from the balanced growth path are denoted by a ^ (so that 𝐴̂ = ln 𝐴 − ln 𝐴∗ ).
With equilibrium in goods and factor market, taking the current and future prices as given,
the firm hires labour and capital to maximise its current value. For this purpose, the firm
hires labour until the marginal product of labour is equal to the wages:
𝑌
(1 − 𝛼) 𝑡 = 𝑤𝑡 , ………(9)
𝐿 𝑡
And it hires the capital stock until the marginal cost of investment, which is assumed to be
‘1’ here is equal to the sum of expected discounted marginal product of capital at time t+1
and the discounted value of the extra capital stock which is left after depreciation at time
t+1. This is represented by the following equation:
1 𝑌 1−𝛿
1 = 𝐸𝑡 [1+𝑟 𝛼 𝐾𝑡+1 ] + 𝐸𝑡 [1+𝑟 ] ……..(10)
𝑡+1 𝑡+1 𝑡+1
The government expenditure also follows a stochastic path and is expressed in the similar
way as in case of productivity in representative firm.
ln 𝐺𝑡 = ln 𝐺𝑡∗ + 𝐺̂𝑡
𝐺̂𝑡 = ∅𝐺 𝐺̂𝑡−1 +∈𝐺,𝑡 with |∅𝐺 | < 1 …….(11)
∗ ∗
𝐺𝑡 = 𝐺𝑡−1 (1 + 𝑔)
Here, the stochastic shock ∈𝐺,𝑡 is independent and identically distributed (i.i.d.) with mean
zero. ∈𝐴 𝑎𝑛𝑑 ∈𝐺 are exogenous factors which represent the respective shocks in
technology and government expenditure. Both ∈𝐴 𝑎𝑛𝑑 ∈𝐺 are uncorrelated at all leads
and lags.
On the balanced growth path, all the shocks and the deviations have zero value.
∈𝐴,𝑡 = 𝐴̂𝑡 =∈𝐺,𝑡 = 𝐺̂𝑡 = 0
With this assumption, the model assumes the form of neoclassical growth model, for which
the solution is given as follows:
If consumption ‘C’ grows are rate ‘g’, then the consumption level at the balanced growth
path can be derived from equation (5).
1 + 𝑟∗ ∗
𝐶𝑠∗ (1 + 𝑔) = 𝐶
1+𝜌 𝑠
Here, 1+r* indicates the gross real rate of return. By rearranging, we get,
1 + 𝑟 ∗ = (1 + 𝑔)(1 + 𝜌)
at r*, rate of return at the balanced growth path, the first order condition shown in equation
(10) becomes
𝑌∗
𝛼 𝐾𝑡+1
∗ = 𝑟∗ + 𝛿 …..(13)
𝑡+1
In equation (6)
𝑌𝑡 = 𝐾𝑡∝ (𝐴𝑡 𝐿𝑡 )1−∝
∝ (𝐴 1−∝
For given value of L, 𝑌𝑡+1 = 𝐾𝑡+1 𝑡+1 𝐿)
∗ ∗∝ (𝐴 1−∝
At balanced growth path 𝑌𝑡+1 = 𝐾𝑡+1 𝑡+1 𝐿)
1 𝑌 1−𝛿 ∗
𝜕 {1+𝑟 ∝ 𝐾𝑡+1 + 1+𝑟 }
𝑡+1 𝑡+1 𝑡+1
𝜑1 = ( )
𝜕 ln 𝑌𝑡+1
1 𝑌 1−𝛿 ∗
𝜕{ ∝ 𝑡+1 + } 𝜕𝑌𝑡+1 1 𝑌 ∗ 1 1 ∗
1+𝑟𝑡+1 𝐾𝑡+1 1+𝑟𝑡+1
=( ) =(1+𝑟 ∝ 𝐾𝑡+1 ) = (1+𝑟 ∝𝐾 𝑌𝑡+1 )
𝜕𝑌𝑡+1 𝜕 ln 𝑌𝑡+1 𝑡+1 𝑡+1 𝑡+1 𝑡+1
∗ ∗ ∗
and since, ∝ 𝑌𝑡+1 = (𝑟 + 𝛿)𝐾𝑡+1
𝑟 ∗ +𝛿
𝜑1 =1+𝑟 ∗
Similarly,
1 𝑌 1−𝛿 ∗
𝜕 {1+𝑟 ∝ 𝐾𝑡+1 + 1+𝑟 }
𝑡+1 𝑡+1 𝑡+1
𝜑2 = ( )
𝜕 ln 𝐾𝑡+1
1 𝑌 1−𝛿 ∗
𝜕 {1+𝑟 ∝ 𝐾𝑡+1 + 1+𝑟 } 𝜕𝐾 1 𝑌𝑡+1
∗
𝑡+1 𝑡+1 𝑡+1 𝑡+1
=( ) = −( ∝ 2 𝐾𝑡+1 )
𝜕𝐾𝑡+1 𝜕 ln 𝐾𝑡+1 1 + 𝑟𝑡+1 𝐾𝑡+1
∗ ∗
and since, ∝ 𝑌𝑡+1 = (𝑟 ∗ + 𝛿)𝐾𝑡+1
𝑟 ∗ +𝛿
𝜑2 =− 1+𝑟 ∗
finally,
∗
1 𝑌𝑡+1
𝜑3 = − ( [∝ + 1 − 𝛿])
(1 + 𝑟𝑡+1 )2 𝐾𝑡+1
1
𝜑3 =− 1+𝑟 ∗
ECONOMICS Paper 12: Economics of Growth and Development - I
Module 22: Stochastic Growth Models
____________________________________________________________________________________________________
∗
In equation (24), (ln 𝑌𝑡+1 − ln 𝑌𝑡+1 ) = 𝑌̂𝑡+1 ; and (ln 𝐾𝑡+1 − ln 𝐾𝑡+1
∗
)=𝐾 ̂𝑡+1
∗ ∗
𝑟 +𝛿 𝑟 +𝛿 𝑟𝑡+1 − 𝑟 ∗
𝑍𝑡+1 = 1 + 𝑌̂ − ̂ −
𝐾
1 + 𝑟 ∗ 𝑡+1 1 + 𝑟 ∗ 𝑡+1 1 + 𝑟∗
Therefore, equation (23) becomes
𝑟 𝑟 ∗ +𝛿
𝑡+1
𝐸𝑡 [1+𝑟 ∗
]= [𝐸𝑡 (𝑌̂𝑡+1 ) − 𝐸𝑡 (𝐾
̂𝑡+1 )] ……(25)
1+𝑟 ∗
1+𝑟𝑡+1 𝐶𝑡
Further, in a representative household, in equation (23), 𝑍𝑡+1 = .𝐶
1+𝜌 𝑡+1
∗
again taking Taylor’s first-order approximation in which ln 𝐶𝑡+1 = ln 𝐶𝑡+1 , ln 𝐶𝑡 = ln 𝐶𝑡∗ and
𝑟𝑡+1 = 𝑟 ∗ .
Hence,
∗
𝑍𝑡+1 = 1 + 𝜑1 (ln 𝐶𝑡+1 − ln 𝐶𝑡+1 ) + 𝜑2 (ln 𝐶𝑡 − ln 𝐶𝑡∗ ) + 𝜑3 (𝑟𝑡+1 − 𝑟 ∗ )
Here,
1+𝑟𝑡+1 𝐶𝑡 ∗
𝜕{ .𝐶 }
1+𝜌 𝑡+1
𝜑1 = ( )
𝜕 ln 𝐶𝑡+1
1+𝑟𝑡+1 𝐶𝑡 ∗
𝜕{ . } 𝜕𝐶𝑡+1 1+𝑟𝑡+1 𝐶 ∗
1+𝜌 𝐶𝑡+1
=( ) =− ( . 𝐶 2𝑡 𝐶𝑡+1 ) = −1
𝜕𝐶𝑡+1 𝜕 ln 𝐶𝑡+1 1+𝜌 𝑡+1
Similarly,
1+𝑟𝑡+1 𝐶𝑡 ∗
𝜕{ .𝐶 }
1+𝜌 𝑡+1
𝜑2 = ( )
𝜕 ln 𝐶𝑡
1+𝑟𝑡+1 𝐶𝑡 ∗
𝜕{ .𝐶 } 𝜕𝐶 1 + 𝑟𝑡+1 1 ∗
1+𝜌 𝑡+1 𝑡
=( ) =( . 𝐶) =1
𝜕𝐶𝑡 𝜕 ln 𝐶𝑡 1 + 𝜌 𝐶𝑡+1 𝑡
finally,
1+𝑟𝑡+1 𝐶𝑡 ∗ 1+𝑟 𝐶 ∗
𝑡+1
𝜕{ .𝐶 } 1 𝐶𝑡 ∗ . 𝑡
1+𝜌 𝑡+1 1+𝜌 𝐶𝑡+1
𝜑3 = ( ) = ( . ) =( )
𝜕𝑟𝑡+1 1 + 𝜌 𝐶𝑡+1 1 + 𝑟𝑡+1
1
𝜑3 = 1+𝑟 ∗
𝑟 −𝑟 ∗
𝐶̂𝑡 = 𝐸𝑡 (𝐶̂𝑡+1 ) − 𝐸𝑡 [ 𝑡+1 ] ……(26)
1+𝑟 ∗
The equilibrium condition in the goods market is represented by equation (5)
𝑌𝑡 = 𝐶𝑡 + 𝐼𝑡 + 𝐺𝑡
or,
ln 𝑌𝑡 − ln 𝑌𝑡∗ = ln(𝐶𝑡 + 𝐼𝑡 + 𝐺𝑡 ) − ln 𝑌𝑡∗
As, ln 𝑌𝑡 − ln 𝑌𝑡∗ = 𝑌̂𝑡 , ln 𝐶𝑡 − ln 𝐶𝑡∗ = 𝐶̂𝑡 , ln 𝐼𝑡 − ln 𝐼𝑡∗ = 𝐼̂𝑡 and ln 𝐺𝑡 − ln 𝐺𝑡∗ = 𝐺̂𝑡
We have,
̂ ̂ ̂
𝑌𝑡 = 𝜑1 𝐶𝑡 + 𝜑2 𝐼𝑡 + 𝜑3 𝐺𝑡 ̂ …….(27)
Here,
∗
𝜕 ln{𝐶𝑡 + 𝐼𝑡 + 𝐺𝑡 }
𝜑1 = ( )
𝜕 ln 𝐶𝑡
∗ ∗
𝜕 ln{𝐶𝑡 + 𝐼𝑡 + 𝐺𝑡 } 𝜕𝐶𝑡 1
=( ) =( . 𝐶𝑡 )
𝜕𝐶𝑡 𝜕 ln 𝐶𝑡 𝐶𝑡 + 𝐼𝑡 + 𝐺𝑡
𝐶𝑡∗
𝜑1 =
𝑌𝑡∗
Similarly,
∗
𝜕 ln{𝐶𝑡 + 𝐼𝑡 + 𝐺𝑡 }
𝜑2 = ( )
𝜕 ln 𝐼𝑡
∗ ∗
𝜕 ln{𝐶𝑡 + 𝐼𝑡 + 𝐺𝑡 } 𝜕𝐼𝑡 1
=( ) =( .𝐼 )
𝜕𝐼𝑡 𝜕 ln 𝐼𝑡 𝐶𝑡 + 𝐼𝑡 + 𝐺𝑡 𝑡
𝐼𝑡∗
𝜑2 =
𝑌𝑡∗
finally,
𝐺𝑡∗
𝜑3 =
𝑌𝑡∗
Hence, equation (27) becomes
𝐶𝑡∗ 𝐼𝑡∗ 𝐺𝑡∗
𝑌̂𝑡 = 𝐶̂ + ̂
𝐼 + 𝐺̂
𝑌𝑡∗1 𝑡 𝑌𝑡∗ 𝑡 𝑌𝑡∗ 𝑡
All the values of income, consumption, investment and government expenditure on balanced
growth path can be easily calculated through the investment and consumption functions of
representative firm and the representative household, assuming that the wages and the rate
of interest are market clearing. All the variables are endogenously determined. The solution
can also be found if there is some technology shock in the economy which further enables
us to trace out the effects of technology shock in to the infinite future. Same is true for shocks
in government expenditure. But once the economy has passed through these shocks, all the
state variables eventually converge back to their steady state values. We can see this
diagrammatically in figure 1 and 2. We can see that in figure 1, the technology jumps in
quarter 1, resulting in to increase in investment, capital stock, income as well as consumption
by the representative firm and the household. In case of households ‘C’ increases less than
‘Y’ and in case of firms ‘I’ increases more than ‘Y’, leading to an increase in capital stock
‘K’. The real wages (w) follow the time path of income (Y) while the expected rate of return
is high as the economy experiences the technology shock but as its effects fade out, the
expected rate of interest falls and becomes negative after a few quarters. Eventually, all
variables converge back to their steady state values.
Similarly, we can also see the effect of shock in government expenditure in figure 2. In this
figure we can see that after moving on the balanced growth path, the economy experiences
a shock in government expenditure in quarter 1. This leads to the increase in expected returns
[E(r)] but as the shock in government expenditure fades away, the E(r) declines. This has
also led to an increase in investment but in order to keep ‘Y’ equal to ‘C+I+G’, the ‘C’
remains at low level initially. As usual, the real wages follow the same pattern as experienced
in case of ‘Y’ and ultimately every variable of the economy resumes the steady state growth
path.
9. Summary
To sum up, we can say that the stochastic models take in to account of the varying
behaviour characteristics due to uncertainty. A stochastic model has one or more stochastic
element. In theory most of the growth models are deterministic in nature. But any realistic
application in macroeconomics requires models to be stochastic. These models have many
macro-economic implications and play a wider role in policy making. We have also
observed how stochastic models can significantly enrich the analysis of economic growth
and economic development. In particular, this model showed how a simple extension of
our standard models can generate an equilibrium path even though, the shocks are
introduced in to the model.