IT' 9 THE OVEIRLAPPING GENE IONS

Structure
9.0 Objectives
9.1 Introduction
9.2 Structure of the Model
9.3 Dynamic Inefficiency in OverlappingGenerationsModel
1
I 9.4 Social Security

I 9.5
9.6
Let Us Sum Up
Key Words
9.7 Some Useful Books
9.8 hswer/Hints to Check Your Progress Exercises

9.0 OBJECTIVES

I After going through this Unit you should be able to exlain:

I e

e
the standard two-period overlapping generations model with production;
the concept of dynamic efficiency and examine whether the dynamic efficiency
holds for an overlapping generationseconomy; and
the role of social security system in the overlapping generations framework in
eliminatingdynamic inefficiency.

9.1 INTRODUCTION
1 In the context of intertemporal decision making on the part of the households, you
have come across the optimal growth model in the previous unit. There is another
important h e w o r k that also considers households' intertemporal decision making -
although over finite time horizon. This h e w o r k is called the overlapping generations
framework. The framework was first developed by Samuelson (1954) which has
subsequently been widely used in macro dynamics.

In the overlappinggenerationsh e w o k individualshave a finite time horizon (in the

standard case, a two-period time horizon), but the society lives forever. The term
'overlapping generations' implies that at each point of time the life-time of two
generations overlaps. To clarifl, let us take the standard case where each generation
lives exactly for two periods. Thus for the cohort of individuals who are born at the
beginning of period 't', they are alive in period 'I' (when they k e young) and in period
't+l' (when they are old). On the other hand, a new set of individuals are born at the
beginning'of period t + l , who would be alive in period '1+lYand period 't+2'
respectively. Thus between the lifetimes ofthese two successive generations, there is
an exact overlapof one period. In the following discussion we shall concentrate on the
two-periodoverlapping generations M e w o r k only.

9.2 STRUCTURE OF THE MODEL

'

32 The set ofpeoplewho are born at the beginning of period t will be called 'generation t'.
Let us now look at the activities of a representative member of generation t. Each The Overlapping
person is born with an endowment of one unit of labor. In the first period of his life, Generations Model
when he is young, he works and earns a wage income of which he consumes a part
and saves (and invests) the rest. In the next period of his life, when he is old, he does
not work anymore. He nonetheless earns an interest income on his previous period's
savings (investment). He also gets back the principal amountthat he invested, which
he consumes in the second period along with the interest earning. Thus his first and
second period budget constraintsare respectively given by:

ii) cl1= (1 + ?+IIS, 9

where c,, and c, are the first and second period consumption of the representative
member of generation t, and w, and r,+, are the wage rate at period t and the rate of
interest at period t+I respective1y.l The representative member maximizes his two-
period utility h c t i o n U(c,,, c,, ) subject to these two budgetconstraints. Noting that
C2 1
S, = , we can combine the two budget constraints by eliminating s,, which
(1 + c + l )
gives us the following single equation that represents the lfe-time budget comtraint
C2 I
of the agent: C~~+ (1 + .,+]) = w f .

MaximizingU (c,,, c,, ) subject to the life-time budget constraintgeneratesthe following

two frrst order conditions:

From these two equations we can write the optimal c,, and c,, as hlctions of w,and

C2r
r,+, .Since = ,the optimal valueof s, also becomesa functionof w, and r,,, .
(1 + rt+*)
We shall assume that all members of all generations are identical in terms of tastes and
preferences, i.e., they have similar ~tilityhctions.~
Let us now look at the overall macroeconomic picture. As we have mentioned before,
at each period there are two generations who are simultaneouslyalive. Thus at period
t, there is a set of people who belong to generation t-1(these are the people who are
currently old) and a set of people who belong to generation t (these are the people
who are currentlyyoung). The generation t people are the workers in period t, each of
' Note that though the member of generation t made his saving and investment decision at
period t, he earns the interest on that savings only in the next period. Hence :+
is the
relevant rate of interest.
Similar, but not identical. To be more precise, the time subscripts in the utility function will
be different for different generations.
Intertemporal
Decision-Ma king
whom earn a wage income w, . The generation t- I people are the interest earners who
earn an interest income on their previous period's savings (i.e., savings made in period
t-1) at the rate r, .

Suppose the population in successive generations grows at a constant rate n. Thus

if L, denotes the number of people belonging to generation t, and L,-, denotes the
-,
number of people belonging to generation t-1,then L, = (1+ n) L, . Total population
in the economy at period t is given by L, + L,-,.

The production side of the economy is like any standard neoclassical growth model I

that you have seen before. A single final commodityis produced which is used both as
aconsumption good as well as an investment good. Technology for final comlnodity
production is given by a neoclassical production function: Y, = F(K,,L,) , where Fis 1

continuous, concave, exhibits constant returns to scale (CRS) with respect to its two
factors- capital (4and labour (L). As in the Ramsey model, the CRS property of the
production Eunction implies that per capita output Qcan be written as the h c t i o n of
the capital-labourratio (k) in the following way:

Moreover the marginal productsof capital and labour can also written as the following
hctions of the capital-labour ratio:

The continuityand concavitypropertiesof F(L, K) ensue thatf (k) is also continuous

and concave. Additionally we assume that the Inada conditions hold: f (0) = 0;

In a market economy with pedect competition, the wage rate and the rate of interest
are equated with the respective marginal products of labour and capital.
Thus w, = f(k,) - k,f '(kt) and r, = f '(kt).

For the economy as a whole, savings-investment equality (which generates goods

m e t equilibrium for the aggregate economy)tells us that Ct + I, = Y; = F(K,,L,) .
Noting that investment is nothing but the augmentationof the capital stock (assuming
no depreciation), i.e., I, = K,,, - K,,and also noting that due to the property of CRS
= w,L,+ r,K,,wecan write theabove s a v h p i n v equality
F(L,,Kt) ~ cbndition
as: C, + Kt+, - Kt = w, L, + r,K,. Noticethat C,denotes the aggregate consumption
at period t, which includes consumption by the old group and consumption by the
young p u p . There are L,-,
number of old people who are alive at period t and each
ofthem consume an amount c,,-, .3On the other hand, there are :L, number of people
-.
3
C,,, is the 2nd period cmsumption.ofthe mpmmtniive member of generation t-I.
at period t belonging to the young generation, each of them consumingan amount c,, The Overlapping
Generations Model
Hence C, = L,-,c2,-,+ L,c,,.
At this p i n t it is important to recall that each young person is born with m endowment
of one unit of labour. There is no bequest; therefore the young people do not o h any
capital stock at period t (capital stock being the only asset in this economy). Thus the
entire capital stockp owned by the older generation.And since the older generation is
going to die at the end of this period, they consume their entire interest earning plus the
capital stock (which, in the one good world, is directly consumable). Hence
L,-,c2,-, = K, + r, K, . Again, each young person earns a wage w,, of which he
consumes c,, and saves s,. Thus L,c,, = L,(w, - s, ) . Using all these information, we
can write the savings-investment equality condition as:

Simplifying, we get the goods market equilibriumcondition for the aggregate economy
as: Kt+,= L,s, .Recall thats, isafunctionofw,and r,+,. w, andr,,, are intumfunctions
of k, and k,+, respectively. Thus s, can be written as a function ofk, and kt+,
:
s(w(k,), r(k,+,)) . This allows us to write the goods market equilibrium condition in
terms ofthe capital labour ratio as follows:

This last line above represents the basic dynamicequation ofthis overlappinggenerations
model which specifiesthe relationship between the capital-labour ratio of today and
the capital-labourratio oftomorrow. Tracing this dynamic equation will tell us how the
capital-labour ratio of the economy changesover time. It is easy to see that this quation
is a first order non-linear difference equation in k. To characterizethe solution path we
shall use the phase diagram technique. The phase diagram plots k,,, on the vertical
axis and k, on the horizontal axis. The intersection ofthe kt+,line with the 45" line
denotes the steady state. Let us now determine the slope of this line. The slope is given
dk,+I
by the derivative -. As you can see, the LHS of the dynamic equation is also a
dk,
h c t i o n of kt,,. Hence total differentiating both sides,

dw dr
Noting that --- = -kfW(k)and - = f"(k)we can write the above equation as:
dk dk
Intertemporal
Notice that as yet we do not know the signs of s, and s, ; hence we cannot say
Decision-Making
whether the kt+,line is positively sloping or negatively sloping. It is easy to see that
under the assumption that consumption in both periods are normal goods (i.e., both
c, and c, increases withan increase in w), 0 < s, < 1. The sign of s, however is
ambiguous. Since an increase in r implies that the relative price of future consumption
in terms of current consumption falls. Hence due to substitution effect current
consumption should fall, which means that with unchanged wage rate, savings would
rise. However, a fall in relative price will also be associated with a positive income
e f f i ton current consumption. Thus whether current consumption increases due to an
i n c m in r depends cruciallyon whether the income effect dominates the substitution
effect. If the income effect dominates the substitution effect, thenc, rises and
consequently s, < 0 . Onthe other hand, ifthe substitutioneffect dominatesthe income
effect, then c, fils and consequently s, > 0 . We shall assume here that the latter holds,
i.e., s, > 0. Under this assumption the kt+,line is positively sloping. We still do not
know the curvature of this line. Depending on the c d a t u r e multiple equilibria (i.e.,
multiple steady states) are possible,as shown in Fig. 9.1.

k, , line

Fig. 9.1 : Multiple Steady State

Local stability of a steady state depends on whether the k,,, line intersects the 45" line
from above or fiom below. In Fig. 9.1 we find that k',where tlle kt+,line intersects the
45" line from above, is a locally stable equilibrium. On the other hand, k*', where
the kt+,line intersects the 45" line fiom below, is a locally unstable equilibrium.

9.3 DYNAMIC INEFFICIENCY IN OVER LAPPING

GENERATIONS MODEL
In the following discussion we shall assume that the parametricconditions are such that
there exists a unique equilibrium which is locally stable.

Even when the steady state is unique and stable, in an overlappinggeneration h e w o r k

36 the equilibrium can still be characterized by dynamic inefficiency. This is a serious
I
shortcoming of the overlapping generations framework in comparison to the optimal The Overlapping
growth fkamework. Before we elaborate this point firther it is important to know what Generations Model
we mean by dynamic efficiency or inefficiency.

The concept of dynamic efficiency is closely related to the concept ofPareto efficiency.
To understand the concept consider a situation where we are comparing between
various possible steady states or equilibrium points. Each ofthese steady states is
characterized by a different capital-labour ratio and correspondingsteady state leve 1s
of per capita consumption for the old and the young. Since utiIity depends on the
consumption during youth and during old-age, each of these steady states is therefore
associated with a different level of steady state utility. Now among all these steady
states, the one which provides maximum steadystate value of utility is called the 'golden
rule' point and the corresponding capital-labourratio is called the 'golden rule' capital
labour ratio. This point is the best possible steady state point which providesmaximum
life-time utility to each person. One can show that steady state utility is maximized at
the point where f '(k) = n . Thus golden rule capital-labour ratio is defined by k, such
that .f '(k,) = n . Fig. 9.2 depicts the golden rule capital labour ratio as the point
which maximizessteady state utility.

Dynamically inefficient region

k,

Fig. 9.2: Golden Rulecapital Labour Ratio

Now consider all the points which lie to the right of kg. Here the capital-labour ratio
is more than optimal. In other words people are over-saving and over-investing. If
instead of saving, people consume a part of their savings in the first period of their life,
then the capital-labour ratio will fall to k, and at the same time their life time utility will
increase. Thus all the points lying to the right ofk, are Pareto inferior to k, : one can
improvethe cutrent consumptionwithout reducing future co~?sumption and thus improve
total lie-time utility.Thesepoints are called dynamicallyinefficientpoints. Now consider
all the points to the left of k , . Here also the utility level is less than k, ;hence by
moving to kg one can improve the steady state value of utility. However such amove
is not costless anymore. If one wants to move fiom a point to the left ofk, to kg,then
he has to save more. In other words he has to forgo some a m o u t of currant
consumption. Thus his current utility wouldfall, even though he would be better off in
Intertemporal
Decision-Making
the future (once he reaches k , ). Since such a move from the left to right involves a
current utility loss, we cannot say for sure whether these points are better or worse
then k, .All these points are Pareto efficient or dynamically efficient.

We have seen that in the optimal growth framework the steady state is defined
by f '(k) = n + p . Thus the steady state point in that framework is always to the left
of the golden rule point and is therefore dynamically efficient. In the case of the
overlapping generations h e w o r k however dynamic efficiency ofthe equilibrium point
cannot be In fact under very reasonable parametric values the steady
state could be dynamically inefficient.

To see how, consider the following example where we assume a specific utility function
and aspecific production function. Let the utility function ofthe representative member
of generation t be U(c,,,c2,) E log c,, + log c,, .Also let the per worker production
h c t i o n be f ( k t )z Ak; , 0 < a < 1. Both the log utility function and the Cobb-
Douglas production h c t i o n are known to be well-behaved which satis@all the standard
neoclassical properties.

With the log utility b c t i o n as specified above, one caneasily verifLthat the first order
conditionsfor individual's utility maximization exerciseare given by:

These two first order conditions generate a savings function which is given by:
1
S, =- w,,Also, with Cobb-Douglas production function w,= (1 - a ) A k P . Thus
2
1
the basic dynamicequation in this example is given by: k,+, = -(1- a ) A k P .At
(1 + n)
steady state k, = k,,, = k*. Hence puttingk' at the LHS and RHS of the above
1 /(]-a)

equation, we can solve for the steady state value as: k' -

Let us now comparethis steady state value of capital-labourratio with the conresponding
golden rule capital labour ratio. Note that with the Cobb-Douglasproductionfimction,
the golden rule capital labour d o in this example is defined by: A a(k)"-' = n. Solving,

we getk, as: kg = . Thus whenever a: > -, n k* > kg,i.e.,

the steady state will be dynamically inefficient. For example, the equilibrium will be
dynamidyinefficientife = 1/ 4 and n = 1.

An obvious question that arises here is: why is it that the steady state could be
dynamicallyinefficient in the overlapping genedons b e w o r k , while such possibility The Overlapping
is ruled out in the optimal growth framework? The answer lies in the fact that in the Cenerstions Model
overlappinggenerations individualsare selfish (no bequest). Since they do not have to
share the benefits of their invesbnent withtheir successive genemtions(who are growing
at the rate n), when they consider the possible future return to their investment, the
return is not net of the population growth. TO 'put it differently, the relevant return for
them is not (f '(k) - n) but just f '(k) . Hence they would be interested in investing as
long as this returnlis positive, even if it falls short of n. This is the reason for their
possible over-saving.

Check Your Progress 1

1) Derive the basic dynamic equation of the standard two period overlapping
generations model with production. Explain the intuition behind this equation.

2) What is dynamic efficiency? Is the steady state under the overlapping generations
structure necessarily dynamically eficient? Give an example to elaborate your
answer.

9.4 SOCIAL SECLTFUTY

Given tlut the overlapping generations equilibriunlcould be characterized by dynamic
inefficiency,one can ask ifthere is any role for the government here. In other words,
does there exist any government policy which could check the over-saving by the
households? Typically the government could influence the savings decision through
some kind of social security programmes, What is a social security programme? A
social security programme is meant to provide some income to individuals afler
retirement. There are usually two types of social security programmes which are in
vogue - a fi~llyfunded system and a pay-as-you-go system,
Ihtertemporal In a fully funded system a stipulated amount is taken by the government from each
Decision-Making young person at period t as social security contribution. The same amount is invested
by the government and returned with interest to the old people at period t + l . Since
the young at period tare the old at period t+l, in a fully h d e d system the set of
people who made the contributionstoday and the set of people who will receive the
corresponding benefit tomorrow are the same. If dl is the contribution made by each
young person at period t, and b,,,is the benefit received by each old person at period

In an unfunded pay-as-you-go system, a stipulated amount is taken by the government

from each young person at period t and the entire amount is immediately distributed to
the old people at period t. Inother words, the pay-as-you-go system involves a transfer
of funds fiom the current young to the current youth. Note that populatioil is growing
at the rate n; therefore at eachpoint of time there are n more people belonging to the
c m n t l y young group compared to the currently old group. Thus if d, is the contribution
,
made by each young person at period t, and b, is the9benefitreceived by each old
person at period t,then b, = (1 + n)d,.

The type of social security system has important implications for the savings decisions
of the young. Note that in a filly funded system the government is doing part of the
savings on behalf of the individuals. The representative individual is effectively saving
an amount equal to (st + d,) and in the next period earning an interest income equal to
(1 + r,,, )(sf + d,) . Since each person knows that this is the amount that he would
receive in the next period, in his own savings decision, the individual will optimally
adjust (cut back) his own savings so that his total effectivesavings renlains the sameas
in the pre-social security economy. Thus a fidly funded social security system has no
impact on the total savingsand capital accumulation. Ifthe economy was in a dynamically
inefficient steady state in the pre-social security economy, it will remain so even after
inlroducinga l l l y h d e d social security system.

In contrast, in apay-as-you-go social security system, the relevant rate of return on

government securities for an individual is n. On the other hand, the rate of interest on
capital accumulationis r. Thus a s long as r<n, it pays to save less and contribute more
in the form of social security. Hence if the economy was in a dynamically inefficient
steady state in the pre-social security economy (which implies that r is indeed less than
n), introducing apay-as-you-go type of social security system will reduce private
savings and capital accumulation. As a result the economy will move to the dynamically
efficient region.

Check Your Progress 2

I ) Doesanytype of social securitysystem necessarily elinzinatethe dynamic inefficiency
problem in the overlapping generations framework? Elaborate your answer with
the two types of social security systems.
 The Overlapping
Generations Model

9.5 LET US S U M UP
In this unit, we discussed at the standard overlapping generations framework with
production. We have derived the basic dynamic equation which basically states that
the tomorrow's capital stock is equal to the savings by the young generations today.
We have derived the steady state conditions. It has been shown that the steady state
under the OLG fixmework may not be dynamically efficient. The government can play
an important role here to ensure dynamic efficiency by introducing a social security
system. Zlowever the type of the social security system is important: apay-as-you-go
type of social security system is effective in eliminating inefficiencywhile a l l I y h d e d
social security system is not effective.

Dynamic Efficiency The situation where an increase in future

consumption implies a fall in current
consumption so that future utility cannot be
increased without reducingcurrent utility.

Overlapping Generations A h e w o r k where agents have finite lifetime

and at every point of time more than one
generation of agents coexist.

Social Security System A provision for retirement benefit such that

an agent receives certain m o n e w benefit
from the government when he is old.

1 9.7 SOME USEFUL BOOKS

(
I
Blanchard, Olivier and Stanley Fischer,1989, Lectures on Macroeconomics, MIT
Press, chapter 2.

9.8 ANSWEWHINTS TO CHECKYOUR

PROGRESS EXERCISES
Check Your Progress 1
I

1) Read Section 9.2 and Mite your answer.

2) Read Section 9.3 and write your answer.

1
I
Check Your Progress 2

1) Read Section 9.4 and write your answer.