Scribd
Upload
9 views3 pages

A Very Short Introduction to Dynamic Optimisation

Uploaded by

levigagne0
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
9 views3 pages

A Very Short Introduction to Dynamic Optimisation

Uploaded by

levigagne0
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

A Very Short Introduction to Dynamic Optimisation

Mario Alloza∗

1 Introduction
Economics is often interested in the behaviour of individuals or agents. In the models we will
study, these agents are assumed to behave rationally, that is, taking decisions that optimise
their utility (in the case of households) or profits (in the case of firms). Optimisation implies
that agents maximise their utility/profits subject to the restrictions they face. When this
optimisation process spans more than one period, we call it Dynamic Optimisation. In this
short tutorial, we will introduce some tools that allow us to analyse the agent’s behaviour.

2 Dynamic Optimisation
We will analyse the case of an economic agent (we will focus on a household) who optimises a
concave 1 utility function U which depends (positively) on consumption, ct and (negatively)
on the number of hours worked, nt : U (ct , nt ).2 We assume that the agent lives from period 0
to period T , so she will take decisions on how much consume or work in each period t during
her lifetime. However, the agent discounts future utility by a factor 0 < β < 1 so that her
present-value utility becomes:
∞
X
V0 = β t U (ct , nt ) (1)
t=0

The agent faces a budget constraint (holding ∀t = 0 . . . ∞) which imposes a trade-off


between consuming or working:

ct + kt+1 ≤ wt nt + (1 + rt )kt (2)


where kt are savings that allow the agent to move consumption over time, wt is the real
wage, and rt an interest rate paid over the savings.
We will make two further observations:
∗
PhD candidate, Department of Economics, UCL. E-mail: [email protected]
1
Do you know why concavity of this function is relevant?
2
Alternatively, we could also say that the individual obtains positive utility from consumption and leisure
(lt ): U (ct , lt ). If we normalise the available time during the day to 1, there is a direct relationship between the
time we spend consuming leisure and the time spend working: nt = 1 − lt (i.e. taking a decision on nt implies
a decision on lt ). We can then write U (ct , 1 − nt ). Notice that if want to take the derivative of this expression
with respect we can apply the chain rule: ∂U (c∂n t ,1−nt )
t
= −Ul (ct , lt ).

1
1. In the optimum, the budget constraint holds with equality: ct + kt+1 = wt nt + (1 + rt )kt ,
that is, no resources are left un-used. Otherwise, we would be facing an optimisation
problem with inequality constraints which would require the use of the Kuhn-Tucker
conditions.

2. We assume that there is no uncertainty, so future values of the variables (e.g. ct+1 , wt+1 ,
kt+2 ...) for t + 1 to ∞ are known. The alternative would be to recognise the presence of
uncertainty and set up a stochastic dynamic optimisation problem, which would involve
expectations over unknown objects.

2.1 The Method of Lagrange Multipliers


To maximise Equation 1 subject to Equation 2 we use the method of Lagrange Multipliers.
This procedure involves setting up the Lagrangian function
∞
X
L(c, n, k, λ) = V0 + λt [wt nt + (1 + rt )kt − ct − kt+1 ] (3)
t=0

where λt are called the Lagrange Multipliers. In our case: 3

T
X
L= {β t U (ct , nt ) + λt [wt nt + (1 + rt )kt − ct − kt+1 ]} (5)
t=0

The first order conditions for t = 0 . . . ∞ are obtained by taking derivatives with respect
to the arguments:

∂L
= β t Uc − λ t = 0 (6)
∂ct
∂L
= β t U n + λ t wt = 0 (7)
∂nt
∂L
= −λt + λt+1 (1 + rt+1 ) = 0 (8)
∂kt+1
∂L
= wt nt + (1 + rt )kt − ct − kt+1 = 0 (9)
∂λt

where Uc,t = ∂U (c t ,nt )


∂ct and Un,t = ∂U ∂n
(ct ,nt )
t
. Note that the above conditions hold for
t = 0 . . . ∞, so in each period t, these conditions are the rules that households will use to
behave optimally.
3
Notice that in this case the Lagrange Multipliers are implicitly including a time discount factor. We can
make this explicit and then write:
T
X
L= β t {U (ct , nt ) + λt [wt nt + (1 + rt )kt − ct − kt+1 ]} (4)
t=0

The optimality conditions resulting from this problem will have the same interpretation.

2
2.2 Interpretation of First Order Conditions
Combining Equations 6 and 7 we obtain an intratemporal restriction relating consumption
and leisure decisions:

Un,t
= −wt (10)
Uc,t
Particularly, the left-hand side of Equation 10 is the Marginal Rate of Substitution of
consumption and hours worked (the slope of the indifference curve), which is equated to the
ratio of prices of hours worked (wt ) and consumption (normalised to 1). More importantly,
Equation 10 defines the labour supply of the agent.
On the other hand, by combining Equation 8 in period t and t + 1 we obtain the Euler
Equation which relates the utility of consumption today and consumption tomorrow:

Uc,t = βUc,t+1 (1 + rt+1 ) (11)

This condition tells us how the agent should allocate consumption intertemporally. Par-
ticularly, it says that the marginal utility of consumption today must be equalised to the
discounted future marginal utility of consumption tomorrow; note that when the agent de-
clines consuming one unit of the good today, she obtains (1 + rt+1 ) to be consumed tomorrow.

3 Readings
A more extensive analysis of dynamic optimisation can be found in the appendixes of the
following books:

• A. Mas-Colell, M.D. Whinston and J.R. Green: Microeconomic Theory, Oxford Univer-
sity Press (1995).

• M. Wickens: Macroeconomic theory: A dynamic General Equilibrium Approach, Prince-


ton University Press (2009)

You might also like