Macroeconomics I

Josep Pijoan-Mas

CEMFI, September-December 2024

 Part V

Overlapping Generations Models

Outline

1 Introduction

2 The basic OG model

3 Optimality

4 Altruism

5 Social security
Outline

1 Introduction

2 The basic OG model

3 Optimality

4 Altruism

5 Social security
 Introduction The basic model Optimality Altruism SS

Introduction

● Let’s go back to the context of the Ramsey-Cass-Koopmans model

– Consumption and savings decided by a representative dynasty
– Diminishing returns to capital accumulation

● This is a fundamental model for macroeconomics

– Used as a basis for multiple extensions dealing with such diverse questions as growth,
business cycle fluctuations, or asset pricing

● Yet, the model abstracts from two different and important issues:

1 Life cycle profiles

People behave differently at different ages during their lifetime

2 Trade between individuals in different generations

First step away from representative agent

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 1/62

 Introduction The basic model Optimality Altruism SS

1. Life cycle profiles

● People behave differently in different stages of their lives

● We observe very sharp age differences in

– Labor market outcomes (wages, hours, employment rates)
– Consumption expenditure
– Wealth holdings and portfolio composition
– Household size and household formation

⇒ Individual-level data comes with a life-cycle format. Hard to ignore.

⊳ OG models may be needed to

a) Explain age profiles of economic variables: why is age a relevant state variable?

b) Explain macroeconomic aggregates in countries experiencing a demographic transition

c) Explain demographic transitions: allow for changes in fertility and mortality

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 2/62

 Introduction The basic model Optimality Altruism SS

2. Trade between different generations

● Do the trades between today’s elderly and today’s youth matter?

● There is a wealth of question where they do. For example,

a) Some social security arrangements transfer resources between people of different ages
- Aggregate savings (and hence capital accumulation) or labor supply may be greatly affected.

b) Public debt also transfers resources between people of different ages

c) Asset prices depend on trades between generations

- Households in different stages of their careers face different amounts of labor market risk and
different investment horizons
- As a consequence, portfolio choices display life-cycle patterns.
- The trade of risk between generations becomes relevant for the market price of risky assets.

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 3/62

 Introduction The basic model Optimality Altruism SS

A comment on optimality

● An important feature of the OG models is that they give ground for state intervention:
– The competitive equilibrium may not be Pareto optimal

● Contrary to the Ramsey model, households in OG economies may save too much.

– Steady state aggregate capital may be above the golden rule ⇒ welfare theorems fail

– Why? some trades between generations are not available in the decentralized economy
(Incomplete markets)

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 4/62

Outline

1 Introduction

2 The basic OG model

3 Optimality

4 Altruism

5 Social security
 Introduction The basic model Optimality Altruism SS

The basic OG model

Samuelson (1958) and Diamond (1965)

● Discrete time
● Households
– Simplest possible structure of overlapping generations: two periods
- People born into their young age, interact with current old people
- Next period, they become old and interact with the next young generation
- After that, they pass away
t
– The number of individuals born at time t is Nt , growing at rate n: Nt = N0 (1 + n)
– They own the production factors and rent them to firms
– They buy the final good from firms

● Firms
– Representative firm characterized by a neoclassical production function
– They rent capital and labor from households to produce the homogeneous final good
– They sell the final good to households
J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 5/62
 Introduction The basic model Optimality Altruism SS

The household problem

● Preferences
– An individual who is young at time t, has preferences over consumption when young c1t and
when old c2t+1 :
u (c1t ) + βu (c2t+1 ) with β > 0
(standard properties uc > 0 and ucc < 0 apply)

● Budget constraint at youth

– Individuals work to earn labor income wt
– They decide how much to consume c1t and how much to save st
– The budget constraint is given by,
c1t + st = wt
● Budget constraint at old age
– Individuals do not take any decision
– They are retirees in a world without social security
– The budget constraint is given by,
c2t+1 = (1 + rt+1 ) st
(where rt+1 is the interest rate paid for savings held between t and t + 1)
J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 6/62
 Introduction The basic model Optimality Altruism SS

Optimization

● The optimization problem for a household born at time t:

max {u (c1t ) + βu (c2t+1 ) }

c1t ,st ,c2t+1

subject to the budget constraints:

c1t + st = wt
c2t+1 = (1 + rt+1 ) st

● We can rewrite the problem as:

max {u (wt − st ) + βu ((1 + rt+1 ) st ) }

st

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 7/62

 Introduction The basic model Optimality Altruism SS

The FOC
● The first order condition will be:

uc (c1t ) = β (1 + rt+1 ) uc (c2t+1 )

(discrete time version of the Euler equation of consumption)

⊳ The household chooses between current and future consumption in such a way that the
value of consuming an extra unit today, uc (c1t ), equals the discounted value of saving this
unit for old age and eating when old that unit and the interest rate earned.

● The CRRA case

1/θ
c2t+1 c2t+1 1 1
= [β (1 + rt+1 ) ] ⇒ log = [ log (1 + rt+1 ) − log ]
c1t c1t θ β
´¹c¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸−c
¹¹¹¹¹¹¹ ¶ ´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶ ²
≃ 2t+1 1t ≃rt+1 ≃ρ
c1t

This shows the parallel w/ the Euler equation of the Ramsey model in continuous time

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 8/62

 Introduction The basic model Optimality Altruism SS

How does the optimal saving change as we change prices?

● Using the budget constraints, we can rewrite the Euler equation as

uc (wt − st ) = β (1 + rt+1 ) uc ((1 + rt+1 ) st )

● This defines savings as an implicit function of prices:

st = s (wt , rt+1 )

● We can characterize how st (and c1t and c2t+1 ) change w/ prices

→ Take the total differential of this equation w/ respect to the price of interest

→ This is “partial equilibrium”: change one price, keep the other fixed

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 9/62

 Introduction The basic model Optimality Altruism SS

How does the optimal saving change as we change prices?

1. Increase in labor income
● Differentiating the FOC wrt wt and st :
dst ucc (c1t )
= sw = 2
>0
dwt ucc (c1t ) + β (1 + rt+1 ) ucc (c2t+1 )

– An increase in labor income unambiguously increases savings

– It also increases consumption at old age c2t+1
dc2t+1 dst
c2t+1 = (1 + rt+1 ) st ⇒ = (1 + rt+1 )
dwt dwt
– It also increases consumption at youth c1t :
dc1t dst
c1t = wt − st ⇒ =1−
dwt dwt
dst
and notice that dwt
< 1.

⊳ Therefore, the increase in wt is spread between youth and old age

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 10/62
 Introduction The basic model Optimality Altruism SS

How does the optimal saving change as we change prices?

2. Increase in interest rate
● Differentiating the FOC wrt wt and st :

dst uc (c2t+1 ) (1 − θ2t+1 )

= sr = −β 2
≷0
drt+1 ucc (c1t ) + β (1 + rt+1 ) ucc (c2t+1 )
ucc (c2t+1 )c2t+1
where θ2t+1 ≡ − uc (c2t+1 )
> 0 is the Arrow-Pratt coefficient of relative risk aversion

● We have two counteracting forces

– Substitution effect: an increase in the interest rate makes consumption tomorrow cheaper,
which implies a shift of consumption from today to the next period (↑ st , ↓ c1t , ↑ c2t+1 ).
– Income effect: an increase in the interest rate makes the consumer richer (all income is
earned in period 1), which increases consumption today and tomorrow (↓ st , ↑ c1t , ↑ c2t+1 ).

⊳ c2t+1 increases due to both effects.

⊳ The final effect on c1t is ambiguous.
⊳ The final effect on savings is also ambiguous, and it depends on the final effect on c1t .
J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 11/62
 Introduction The basic model Optimality Altruism SS

Firm problem

● The representative firm operates a risk free technology that combines capital K and
labor L to produce output Y .

● Standard neoclassical production function Y = F (K, L)

● The firm problem will be as usual:

max {F (Kt , Lt ) − wt Lt − (rt + δ) Kt }

Kt ,Lt

● Which gives the following optimality conditions:

FK (Kt , Lt ) = rt + δ ⇒ fk (kt ) = rt + δ
FL (Kt , Lt ) = wt ⇒ f (kt ) − kt fk (kt ) = wt
1
where f (k) ≡ F (k, 1) = L F (K, L)

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 12/62

 Introduction The basic model Optimality Altruism SS

Equilibrium

Given s0 , An equilibrium in this economy will be a sequence of allocations

{c1t , c2t , st , Kt , Lt }t=1 and prices {rt+1 , wt }t=1 such that:
∞ ∞

Given s0 and the sequence of prices {rt+1 , wt }t=1 , the sequence {c1t , c2t , st }t=1
∞ ∞
1
satisfies the household problem.

2 Given a pair of prices {rt , wt }, the pair of allocations {Kt , Lt } solve the firm problem
for every t.

3 The labor market clears: Lt = Nt ∀t

4 The asset market clears: Kt+1 = Nt st ∀t

5 And the aggregate resource constraint is satisfied (which happens due to Walras law):

Yt = C1t + C2t + It

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 13/62

 Introduction The basic model Optimality Altruism SS

Verify the aggregate resource constraint

Let’s check that the aggregate resource constraint holds

● Summing up the individual budget constraints we have:

Nt (c1t + st ) + Nt−1 c2t = Nt wt + Nt−1 st−1 (1 + rt )

● Using equilibrium conditions (2) and (4),

C1t + C2t + Kt+1 = [Yt − Kt fk (kt )] + Kt (1 + fk (kt ) − δ)

● This yields,
C1t + C2t + Kt+1 = Yt + (1 − δ) Kt

● Or also,
Yt − C1t − C2t = (Kt+1 − Kt ) + δKt
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶ ´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
St It

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 14/62

 Introduction The basic model Optimality Altruism SS

Comment on capital markets

● The asset market equilibrium condition

Kt+1 = Nt st ∀t

states that next period’s productive capital is equal to the supply of assets by the
young.

● What happens with the undepreciated capital?

● Let’s look again at the budget constraints of retirees:

Nt−1 c2t = Nt−1 st−1 (1 + rt )

= Kt (1 + FK (Kt , Nt ) − δ)
= FK (Kt , Nt ) Kt + (1 − δ) Kt

→ Undepreciated capital (and its income) is eaten up by the elderly

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 15/62

 Introduction The basic model Optimality Altruism SS

Law of motion for aggregate capital

Now, let’s look for a law of motion for aggregate capital.

● Take the optimality condition of household behavior and substitute prices for their
values in equilibrium:

st = s (wt , rt+1 ) ⇒ st = s (f (kt ) − kt fk (kt ) , fk (kt+1 ) − δ)

● Then, the market clearing condition for capital will give us:

s (f (kt ) − kt fk (kt ) , fk (kt+1 ) − δ)

Kt+1 = Nt st ⇒ kt+1 =
1+n

⊳ This equilibrium equation provides the law of motion for aggregate capital
– Note that, w/o specific funcitonal forms for u(c) and f (k), it is a fairly complicated object

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 16/62

 Introduction The basic model Optimality Altruism SS

Steady state equilibrium

● A steady state in this economy will be an equilibrium such that per worker capital is
constant over time, kt+1 = kt .

● Therefore, steady state capital per worker k ∗ will be defined by:

s (f (k ∗ ) − k ∗ fk (k ∗ ) , fk (k ∗ ) − δ)
k∗ = (1)
1+n

– One equation in one unknown

● Notice that the function s (⋅, ⋅) depends on the parameter β and on the parameters
that characterize the utility function.

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 17/62

 Introduction The basic model Optimality Altruism SS

Properties of the steady state

● Questions:
1 Which is the shape of the law of motion for capital?
2 Is there a unique steady state equilibrium?
3 Is (are) the steady state(s) stable?

– graph 3.1 BF – and – graph 3.2 BF –

● Therefore:

1 If dkt+1 /dkt > 0 we will have non-oscillating trajectories

2 If ∣dkt+1 /dkt ∣ < 1 we will have convergent paths

● The slope of the equilibrium law of motion for capital depends on the functional
forms of the utility function and the production function:
dkt+1 −sw kt fkk (kt )
=
dkt 1 + n − sr fkk (kt+1 )

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 18/62

 Introduction The basic model Optimality Altruism SS

Properties of the steady state

● For a steady state to be (locally) non-oscillatory we require:

dkt+1 −sw k ∗ fkk (k ∗ )

∣ = >0
dkt k∗ 1 + n − sr fkk (k ∗ )

– The numerator is positive. Then, we require:

1 + n − sr fkk (k∗ ) > 0 ⇒ sr >

1+n
fkk
²
−

– Hence,
→ A sufficient condition is that the income effect does not dominate (sr ≥ 0)
1+n
→ If the income effect dominates, sr < 0, and is large enough, sr < fkk
, we will have a negative slope
(⇒ oscillatory trajectories).

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 19/62

 Introduction The basic model Optimality Altruism SS

Properties of the steady state

● For a steady state to be (locally) stable (in addition to non-oscillatory) we require:

dkt+1
1> ∣ >0
dkt k∗

– Which can be rewritten as

dkt+1 1+n
∣ >0 ⇒ sr >
dkt k∗ fkk
+ sw k∗
dkt+1 dkt+1 1+n
∣ > 0 and ∣ <1 ⇒ sr >
dkt k∗ dkt k∗ fkk
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹¶
+ or −

→ A necessary condition is that sr is not too small, that is to say, the income effect is not too large.

– From now one, we will work with the OG model under the constraint that this condition holds

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 20/62

 Introduction The basic model Optimality Altruism SS

Intuition
The equilibrium relationship,

kt+1 (1 + n) = s( f (kt ) − kt fk (kt ), fk (kt+1 ) − δ )

´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶ ´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
wt rt+1

establishes that, if we increase kt :

● the market wage wt increases and since sw > 0 households want to save more and it
seems that we need kt+1 to increase in order to restore the equilibrium in the capital
markets.

● But, if kt+1 increases, firms must agree on it and therefore they will require rt+1 to fall

● With a fall in rt+1 households modify their savings behavior

– w/ sr > 0 they correct down their savings
– w/ sr < 0 they correct them up

● Then, the final effect on kt+1 depends on the sign and size of sr .
J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 21/62
Outline

1 Introduction

2 The basic OG model

3 Optimality

4 Altruism

5 Social security
 Introduction The basic model Optimality Altruism SS

Optimality

⊳ Are the decentralized allocations Pareto optima?

– A non optimal allocation would be one in which we can improve the welfare of at least one
individual without hurting anybody else

● Optimality in the Solow model

– The steady state that maximizes consumption is known as the golden rule and is
characterized by:
fk (kg∗ ) = n + δ ⇔ r = n

– If the steady state capital in the decentralized allocation happens to be above the golden
rule we are in a situation known as dynamic inefficiency and a Pareto improvement is
possible.

● In the Ramsey and OG models we have the same notion of golden rule steady state
(because the aggregate resource constraint and the production function are the same)

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 22/62

 Introduction The basic model Optimality Altruism SS

The golden rule steady state

● Let’s revise this argument in the context of the OG model.

– Take the aggregate resource constraint of the economy

F (Kt , Nt ) + (1 − δ) Kt = C1t + C2t + Kt+1 ⇒
C1t + C2t = F (Kt , Nt ) − [Kt+1 − (1 − δ) Kt ]

– Write it in per worker terms

ct = f (kt ) − [kt+1 (1 + n) − (1 − δ) kt ]
C1t +C2t
where we have defined ct ≡ Nt
.

– Then, in steady state (kt+1 = kt ) we have

c∗ = f (k∗ ) − (n + δ) k∗

– Aggregate consumption per worker is maximized at the kg∗ such that

fk (kg∗ ) = n + δ

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 23/62

 Introduction The basic model Optimality Altruism SS

Is it possible that households over-accumulate resources?

● We need to compare k ∗ — defined by equation (1) — with kg∗
– We are in a dynamically inefficient situation if
k∗ > kg∗ ⇔ fk (k∗ ) < fk (kg∗ )

● A simple case. Assume: u (c) = log (c) and f (k) = k α

– Then, we can show that:

1

k∗ = ( ∗
1−α β 1−α 1+n 1+β
) and fk (k ) = α
1+n 1+β 1−α β

– Therefore we are in a dynamically inefficient situation if

fk (k∗ ) < fk (kg∗ ) ⇔ α

1+n 1+β
<n+δ
1−α β

– Or in a more meaningful way:

−1
(n + δ) (1 − α)
β>( − 1)
α (1 + n)
J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 24/62
 Introduction The basic model Optimality Altruism SS

Over-accumulation and numbers

● That is to say, if households are patient enough (β is large enough), they may reach a
situation in which there is over-accumulation of capital.

● Note that in the Ramsey model (∞-lived agents) the TVC would prevent this situation
from happening
– We do not have a TVC here

● How plausible is this situation?

– Assume that the period length is 30 years and that the annual rate of population growth is
1%, depreciation rate 8% and capital share 28%.

– This inequality implies that if households annual subjective rate of discount is smaller than
1.16%, then we are in the non optimal case.

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 25/62

 Introduction The basic model Optimality Altruism SS

So ... why does this happen?

● Let’s start at k in a dynamic inefficiency area (fk < δ + n ⇔ r < n)

(For clarity, you can assume n = 0)

– Ramsey model
- If individuals ↓ s in every period, this allows them to ↑ c in every period
∗
(Indeed, if n = 0 the saving technology has a negative return at k > kg )

- Hence, k will diminish and the steady state will be at some k∗ < kg∗

– OG model
- if young ↓ s in every period, they ↑ c as young in every period but they ↓ c as old in every period
- The reduction of saving when young implies having less to eat as old
- A steady state k∗ > kg∗ is possible

● Is there any way that agents in the OG economy can trade and so benefit from the
available Pareto improvement?

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 26/62

 Introduction The basic model Optimality Altruism SS

A trade that dominates saving

● Imagine the following transfer system:
– when young, every member of a given cohort at time t gives one unit of income to the
contemporaneous old cohort
– in exchange, when old at t + 1, the young guy gets “the same deal”

⊳ This implies that each member of the old cohort receives Nt /Nt−1 = 1 + n units of
consumption.

● If instead of this transfer system, the young cohorts save with the available
technology, they obtain 1 + r units of consumption when old.

● Whenever r < n (⇒ dynamic inefficiency) the intergenerational transfer system

dominates private savings.

⊳ However, this type of trade cannot occur in the decentralized economy.

– This breaks the welfare theorems: markets are not complete

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 27/62

 Introduction The basic model Optimality Altruism SS

Solutions

● PAYG social security system → forced transfers from young to old

● Public Debt → crowds out private capital (Diamond, AER 1965)

● Bubbles → A fundamentally worthless asset crowds out capital (Tirole, ECTA 1985)

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 28/62

Outline

1 Introduction

2 The basic OG model

3 Optimality

4 Altruism

5 Social security
 Introduction The basic model Optimality Altruism SS

Altruism

● So far we have considered individuals that only care about themselves

– This logically brings the optimality condition of no bequests.

● But bequests do take place in real life, and they are quantitatively important.

● They can be either

– accidental
– intended

● Among intended bequests, we can see them:

– as pure altruism towards offspring
– as a way to manipulate children’s actions
Barczyk, Kredler (REStud 2018)

⊳ We are now going to introduce a well-defined notion of altruism into the OG model
J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 29/62
 Introduction The basic model Optimality Altruism SS

Objective function
● Let’s consider that parents care about the welfare of their kids

● The objective function of a household will be:

Ut = u (c1t ) + β u (c2t+1 ) + γ (1 + n) Ut+1
(where γ is the degree of altruism)

So, an individual cares about his consumption today and tomorrow plus the welfare
that his kids will attain.

● Substituting recursively, we get:

∞
i
Ut = ∑ γ i (1 + n) [u (c1t+i ) + β u (c2t+1+i )]
i=0
where we need γ (1 + n) < 1 for finite utility

● This shows that caring about the children who cares about their own children makes
the current generation care about all future mankind
(albeit with a potentially different weight on future generations)
J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 30/62
 Introduction The basic model Optimality Altruism SS

Optimization problem
● The budget constraints of the household problem are:

c1t + st = wt + bt
c2t+1 + (1 + n) bt+1 = (1 + rt+1 ) st

(where bt is the amount of bequest per young person)

● We impose a non-negativity constraint on bequests: bt+1 ≥ 0

● Then, starting at some t, the optimization problem of the parents is to choose the
optimal paths of savings and bequests
⎧
⎪ ⎫
⎪
⎪∞ i i ⎪
max ∞ ⎨ ∑ γ (1 + n) [u( wt+i + bt+i − st+i ) + β u( (1 + rt+1+i ) st+i − (1 + n) bt+1+i )]⎬
{st+i ,bt+1+i }i=0 ⎪
⎪ ´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶ ´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶ ⎪
⎪
⎩ i=0 c1t+i c2t+1+i
⎭
s.t. bt+1+i ≥ 0 ∀i, and bt given

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 31/62

 Introduction The basic model Optimality Altruism SS

First Order Conditions

● Lagrangian:
∞
i
L = ∑ γ i (1 + n) [u( wt+i + bt+i − st+i ) +
i=0 ´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
c1t+i

β u( (1 + rt+1+i ) st+i − (1 + n) bt+1+i ) + βλt+i bt+1+i ]

´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
c2t+1+i

(where λt+i are the Lagrange multipliers of the inequality constraints)

● Then, the optimality conditions (Khun-Tucker) are:

∂L
= 0 ⇒ uc (c1t+i ) = β (1 + rt+1+i ) uc (c2t+1+i )
∂st+i
∂L β
= 0 ⇒ βuc (c2t+1+i ) = γuc (c1t+1+i ) + λt+i
∂bt+1+i 1+n

λt+i bt+1+i = 0, bt+1+i ≥ 0, λt+i ≥ 0

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 32/62

 Introduction The basic model Optimality Altruism SS

First Order Conditions

Interpretation

⊳ The first condition we already know it. The household allocates resources between
youth and old age such that the marginal value of a consumption unit is equalized
across periods.

⊳ The second condition is new. It says that the marginal value of a consumption unit
during the individual old age has to be equal to the marginal value of that unit if left
to the offspring.
– If the inequality constraint binds (λt+i > 0), then we will have:

βuc (c2t+1+i ) > γuc (c1t+1+i )

which states that the household would be better off by reducing bequests and therefore
increasing his own consumption when old and decreasing the consumption of his offspring.
– But this transfer cannot be done due to the non-negativity constraint on bequests.
(λt+i > 0 ⇔ bt+1+i = 0)

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 33/62

 Introduction The basic model Optimality Altruism SS

The steady state equilibrium

⊳ The firm problem and the equilibrium conditions are identical to the basic OG model.

⊳ The steady state requires: kt+1 = kt = k ∗ and cit = cit+1 = c∗i

⊳ Then, the optimality conditions become,

uc (c∗1 ) = β (1 + r∗ ) uc (c∗2 )
β ∗
βuc (c∗2 ) = γuc (c∗1 ) + λ
1+n

⊳ And combining them both,

β ∗
uc (c∗1 ) = (1 + r∗ ) [γuc (c∗1 ) + λ ]
1+n

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 34/62

 Introduction The basic model Optimality Altruism SS

The steady state equilibrium

Situation 1: positive bequests
● If b∗ > 0 and λ∗ = 0,
1 1−γ
uc (c∗1 ) = γ (1 + r∗ ) uc (c∗1 ) ⇒ (1 + r∗ ) = ⇒ fk (k ∗ ) = +δ
γ γ
The economy satisfies the modified golden rule,
(where the discount factor is not given by the time discount β but by the weight of future generations γ .
Let’s call it the intergenerational
intergenerational modified
intergenerational modified golden
modified golden rule
rule).
golden rule
rule
rule

● Since γ (1 + n) < 1 ⇒ 1
γ
> (1 + n) we have
(1 + r∗ ) > (1 + n) ⇒ fk (k ∗ ) > δ + n = fk (kg∗ ) ⇒ k ∗ < kg∗
Therefore, there is no over-accumulation: the decentralized solution is Pareto-efficient
as in the Ramsey model

● Finally, note that c∗2 /c∗1 is determined by β/γ

β
uc (c∗1 ) = uc (c∗2 )
γ
J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 35/62
 Introduction The basic model Optimality Altruism SS

The steady state equilibrium

Situation 2: binding constraint for non-negative bequests

● If b∗ = 0 and λ∗ > 0,

1 1−γ
uc (c∗1 ) > γ (1 + r∗ ) uc (c∗1 ) ⇒ (1 + r∗ ) < ⇒ fk (k ∗ ) < +δ
γ γ

hence k ∗ would be above the intergenerational modified golden rule

● And note that it could even be that k ∗ > kg∗

→ The economy with a bequest motive is dynamically inefficient if and only if the associated
Diamond economy is also dynamically inefficient
Weil (JME 1987)

⇒ Altruism cannot fix the dynamic inefficiency of the standard OG model

(fixing dynamic inefficiency requires transfers from young to old, but altruism does the opposite, it generates transfers
from old to young)

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 36/62

 Introduction The basic model Optimality Altruism SS

The steady state equilibrium

Summary

● We can have three different types of steady states.

1) One where in equilibrium bequests are positive and the economy satisfies
satisfies the
intergenerational modified golden rule,
(where the discount factor is not given by the time discount β but by the weight of future generations γ )

2) Another where the transfers are zero and the steady state capital
capital is
is above
above the
intergenerational modified golden rule

3) Finally, also with zero transfers, we can have capital above, not only the modified golden
rule, but also above the golden rule itself yielding therefore dynamic inefficiency.

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 37/62

 Introduction The basic model Optimality Altruism SS

Infinitely-lived economy
Now, what if people care about their offspring as much as about themselves?
● Mathematically, we would have:
γ=β

● In this case, the steady state condition gives us:

1 1−β
⊳ If λ∗ = 0 and b∗ > 0 ⇒ (1 + r∗ ) = ⇔ fk (k ∗ ) = +δ
β β
1 1−β
⊳ If λ∗ > 0 and b∗ = 0 ⇒ (1 + r∗ ) < ⇔ fk (k ∗ ) < +δ
β β

● The case λ∗ = 0 is the discrete-time counterpart of the modified golden rule in the
infinitely-lived Ramsey problem seen earlier in the course.

● The case λ∗ > 0 and b∗ = 0 will never arise in the steady state of the Ramsey model.
The TVC prevents this from happening by restricting β(1 + n) < 1

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 38/62

Outline

1 Introduction

2 The basic OG model

3 Optimality

4 Altruism

5 Social security
 Introduction The basic model Optimality Altruism SS FF PAYG Transition

Social security
Introduction

● OG models are a very natural framework to study the effects on capital accumulation
of a welfare program such as the social security.

● In general, the provision of resources for retirement is likely to have important

macroeconomic (and social) consequences

● In modern and rich societies, people live 20+ years after normal retirement age.

Life Expectancy at 60 (year 2021, WHO data) f m

Spain 27.3 23.3

France 27.2 23.3
Greece 25.5 22.1
US 24.4 21.8
Uruguay 24.0 19.0
China 23.1 19.2
Argentina 23.1 18.8

⊳ 20 years of retirement is a long time compared to 40/45 years of working life.

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 39/62
 Introduction The basic model Optimality Altruism SS FF PAYG Transition

Life Expectancy at 65
(OECD Data)

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 40/62

 Introduction The basic model Optimality Altruism SS FF PAYG Transition

Social security
Introduction

● There are different ways to provide resources for the elderly:

1. Informal extended-family insurance. Your offspring takes care of you after retirement.
2. Voluntary saving. People decide how much to save and when they retire they just eat up
those savings.

3. Fully funded social security. The state collects taxes from workers and saves the money to
be given back to people at retirement age.

4. Pay-as-you-go social security. The state collects taxes from current workers and uses these
resources to pay a pension to the current retirees.

● Further considerations:
– Lump sum payments or life annuities?
(What happens if you live for too long?)

– Optimal retirement age

– Optimal fertility rate
J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 41/62
 Introduction The basic model Optimality Altruism SS FF PAYG Transition

A fully funded system

The household problem

● The government does the following:

– At period t, it collects dt from workers (lump sum);
– then, it invests this amount in form of capital;
– at period t + 1 it pays dt (1 + rt+1 ) back to retirees.

(no administration costs and the same saving technology for the government)

● The household budget constraints become:

c1t + st = wt − dt
c2t+1 = (1 + rt+1 ) st + (1 + rt+1 ) dt

● Therefore, the household problem will be:

max {u (wt − dt − st ) + βu ((1 + rt+1 ) (st + dt ))}

st

(where the household takes dt as given)

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 42/62
 Introduction The basic model Optimality Altruism SS FF PAYG Transition

A fully funded system

● The key equations are:

1 Household optimality condition:

uc (wt − dt − st ) = β (1 + rt+1 ) uc ( (1 + rt+1 ) (st + dt ) ) ⇒ st = s (wt , rt+1 , dt )

2 Firm optimality conditions:

rt+1 = fk (kt+1 ) − δ and wt = f (kt ) − kt fk (kt )

3 The market clearing condition for capital:

Kt+1 = Nt st + Nt dt ⇒ kt+1 (1 + n) = st + dt

● They can be combined to get a law of motion of kt+1 as a function of kt

uc (wt − kt+1 (1 + n) ) = β (1 + rt+1 ) uc ( (1 + rt+1 ) kt+1 (1 + n) )

→ Note that dt does not appear

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 43/62
 Introduction The basic model Optimality Altruism SS FF PAYG Transition

A fully funded system

● In the basic case without social security we had:

1 Household optimality condition:

uc (wt − st ) = β (1 + rt+1 ) uc ( (1 + rt+1 ) st ) ⇒ st = s (wt , rt+1 )

2 Firm optimality conditions:

rt+1 = fk (kt+1 ) − δ and wt = f (kt ) − kt fk (kt )

3 The market clearing condition for capital:

Kt+1 = Nt st ⇒ kt+1 (1 + n) = st

● They can be combined to get the same law of motion of kt+1 as a function of kt

uc (wt − kt+1 (1 + n) ) = β (1 + rt+1 ) uc ( (1 + rt+1 ) kt+1 (1 + n) )

Note that dt does not appear

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 44/62
 Introduction The basic model Optimality Altruism SS FF PAYG Transition

A fully funded system

Neutrality

Therefore, the kt+1 that solves the system without social security also solves the system
with a fully funded social security system.

● Both ways of saving (private and social security induced) yield the same return.
Therefore, the household is indifferent between them.

● The households undo through their private savings what the government does
through compulsory saving.
In particular,
dst
= −1
ddt

● Therefore, aggregate savings and consumption are unchanged.

● Notice we have abstracted from distortionary taxation and uncertain lifetimes, which
might break this equivalence.

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 45/62

 Introduction The basic model Optimality Altruism SS FF PAYG Transition

A fully funded system

Excessive government savings (an extreme case)

There is a case, however, in which the allocations with and without social security may
not be identical.

● We have seen that dst

ddt
= −1.
Households substitute private savings for public savings on a one-to-one basis, which leaves
aggregate savings in the economy unchanged

● If we impose a no-borrowing constraint (st ≥ 0) and the government sets dt too big
(larger than private savings without social security), then this substitution cannot be
performed.

⊳ The funded social security system may increase aggregate savings of the economy if
(a) households cannot borrow against future pensions and
(b) the government sets contributions larger than what households would have chosen in
absence of a public system.

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 46/62

 Introduction The basic model Optimality Altruism SS FF PAYG Transition

A pay-as-you-go system
The transfer scheme

● The government does the following:

– At period t, it collects d1t Nt from the young workers;
– it distributes this amount among the Nt−1 retirees giving each d2t :

d1t Nt
d2t = = d1t (1 + n)
Nt−1

– At period t + 1 the retirees (young workers at t) receive each

d1t+1 Nt+1
d2t+1 = = d1t+1 (1 + n)
Nt

● Therefore, the return of the PAYG social security system will be

d2t+1 d1t+1
= (1 + n)
d1t d1t

where (1 + n) = Nt+1
Nt
is the inverse of the dependency ratio at t.
J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 47/62
 Introduction The basic model Optimality Altruism SS FF PAYG Transition

A pay-as-you-go system
The transfer scheme

● Let’s impose a steady state condition for the social security policy:

d1t+1 = d1t

● Then, the rate of return of social security in steady state is (1 + n).

● That is to say, in a two-period model, the fact that population grows means that at any
point in time there are more workers than retirees and therefore the social security
system can pay a positive return.

● In a model with non-stationary demographics, we want to look at the dependency

ratio: number of retirees divided by working age population

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 48/62

 Introduction The basic model Optimality Altruism SS FF PAYG Transition

A pay-as-you-go system
The (inverse of the) dependency ratio in Spain
Working age population / 65+
7

Data (INE)
6
Projection (INE)

0
1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060 2065 2070

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 49/62

 Introduction The basic model Optimality Altruism SS FF PAYG Transition

A pay-as-you-go system
The household problem

● The budget constraints become:

c1t + st = wt − dt
c2t+1 = (1 + rt+1 ) st + (1 + n) dt

(where we use dt instead of d1t for notational simplicity)

● Therefore, the household problem will be:

max {u (wt − dt − st ) + βu ((1 + rt+1 ) st + (1 + n) dt )}

st

(where the household takes dt as given)

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 50/62

 Introduction The basic model Optimality Altruism SS FF PAYG Transition

A pay-as-you-go system
● The key equations are:

1 Household optimality condition:

uc (wt − dt − st ) = β (1 + rt+1 ) uc ( (1 + rt+1 ) st + (1 + n) dt )

2 Firm optimality conditions:

rt+1 = fk (kt+1 ) − δ and wt = f (kt ) − kt fk (kt )

3 The market clearing condition for capital:

Kt+1 = Nt st ⇒ kt+1 (1 + n) = st

● This leads to a different law of motion for capital than the case w/o Social Security:
a) The return of the PAYG is different from the return of private savings
b) The capital market clearing condition is also different
J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 51/62
 Introduction The basic model Optimality Altruism SS FF PAYG Transition

A pay-as-you-go system
How does dt affect private savings?

● The Euler equation for the household defines an implicit function

st = s (wt , rt+1 , dt )

● We can obtain the effect of dt on private savings:

dst ucc (c1t ) + β (1 + rt+1 ) (1 + n ) ucc (c2t+1 )

= sd = − <0
ddt ucc (c1t ) + β (1 + rt+1 ) (1 + rt+1 ) ucc (c2t+1 )

⊳ As in the funded system, the pay-as-you-go pension system crowds out private savings

⊳ However, the size of the crowding out is not necessarily equal to one.

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 52/62

 Introduction The basic model Optimality Altruism SS FF PAYG Transition

A pay-as-you-go system
The size of the crowding out

● If rt+1 = n [golden rule case] → ∣ dddst ∣ = 1

t

– SS pension substitutes private savings in a one-to-one basis

(the partial equilibrium effect of pensions is the same as in the funded system)

● If rt+1 < n [dynamic inefficient region] → ∣ dddst ∣ > 1

t

– SS pension crowds out more than one unit of private savings

– The return of the social security is larger than the return to saving. A one-to-one reduction
generates the same consumption as young and more consumption as old ⇒ a further
reduction of savings spreads the gains between both periods.

● If rt+1 > n [below golden rule] → ∣ dddst ∣ < 1

t

– SS pension crowds out less than one unit of private savings

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 53/62

 Introduction The basic model Optimality Altruism SS FF PAYG Transition

A pay-as-you-go system
Equilibrium capital

● We have seen the effect of dt on private savings st

● However, this is a partial equilibrium analysis

(we have held wt and rt+1 constant)

● Since public savings are independent from dt in the PAYG system, aggregate capital
will fall along with private savings:

Kt+1 = Nt st

and this will affect prices (wt and rt+1 ) and prices affect back to individual allocations.

⊳ This is the fundamental difference with respect to a funded system:

in the funded system there is no effect on aggregate capital because public and
private savings compensate each other.

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 54/62

 Introduction The basic model Optimality Altruism SS FF PAYG Transition

A pay-as-you-go system
Steady state capital
● Following manipulations already familiar to us we have:

Kt+1 = Nt st ⇒ (1 + n) k ∗ = s [f (k ∗ ) − k ∗ fk (k ∗ ) , fk (k ∗ ) − δ, d∗ ]

● This lets us obtain:

dk ∗ sd
=
dd∗ (1 + n) − sr fkk (k ∗ ) + sw k ∗ fkk (k ∗ )

< 0 if the denominator is positive.

∗
dk
● Therefore, we will have dd∗
(We have just seen that sd < 0)

– The condition for the denominator to be positive is the same as the one for having a locally
stable and non-oscillatory steady state.

⊳ An economy with a bigger pay-as-you-go SS has less capital in steady state.

⇒ PAYG may alleviate the dynamic inefficiency
J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 55/62
 Introduction The basic model Optimality Altruism SS FF PAYG Transition

Policy reform and transitional dynamics

Moving from a pay-as-you-go system to a fully funded system

⊳ Does the previous result mean that a movement from a pay-as-you-go Social Security
system to a fully funded system will increase aggregate capital in the economy?

● Imagine that at time t the government unexpectedly moves from a pay-as-you-go to a

fully funded Social Security system, but with two constraints:

1 The government has to honor the social security payments to the transitional generations
(which have been caught by surprise)

2 The government has to spread the burden of this transitional generation evenly across all
current and future cohorts.

● This suggests a policy line: issue public debt Bt to take care of the transitional
generations’ pensions and refinance it perpetually.

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 56/62

 Introduction The basic model Optimality Altruism SS FF PAYG Transition

The equilibrium allocations at t

● What is owed to the current old? Let Bt be the stock of debt issued at t to pay for the
pensions of the generation born at time t − 1:

Bt = Nt−1 dt−1 (1 + n)

● Then, the equilibrium of the economy at t will be described by

1 Household optimality condition for the young at time t:

uc (wt − dt − st ) = β (1 + rt+1 ) uc ( (1 + rt+1 ) (st + dt ))

2 Firm optimality conditions:

rt+1 = fk (kt+1 ) − δ and wt = f (kt ) − kt fk (kt )

3 The market clearing condition for capital:

Kt+1 + Bt = Nt st + Nt dt ⇒ kt+1 (1 + n) + dt−1 = st + dt

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 57/62

 Introduction The basic model Optimality Altruism SS FF PAYG Transition

The equilibrium allocations at t

● Combining the market clearing condition for capital and the Euler equation we get,

uc [wt − dt−1 − kt+1 (1 + n) ] =

β (1 + rt+1 ) uc [ (1 + rt+1 ) (1 + n) kt+1 + (1 + rt+1 ) dt−1 ]

● If nothing had changed (PAYG World) we would have had,

uc [wt − dt − kt+1 (1 + n)] =

β (1 + rt+1 ) uc [(1 + rt+1 ) (1 + n) kt+1 + (1 + n) dt ]

⊳ Therefore, if we keep the contributions to the system unchanged dt−1 = dt these two
equations are the same whenever rt+1 = n
→ The transition from PAYG to FF does not change kt+1

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 58/62

 Introduction The basic model Optimality Altruism SS FF PAYG Transition

The equilibrium allocations at t

Summary

● Why this result?

1) If we are at the golden rule , the household choices are unchanged after the reform
because the return to the funded system is the same as the return to the PAYG system.

2) The aggregate capital of the economy is also unchanged because the higher savings by the
public sector are exactly compensated by the debt of the government: the debt of the social
security to the transitional generation.

⊳ Therefore, prices won’t change either and the economy will be unchanged.

● If rt+1 ≠ n, individuals will modify their consumption and saving, and this will affect
aggregate capital and prices due to second order effects.

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 59/62

 Introduction The basic model Optimality Altruism SS FF PAYG Transition

The equilibrium allocations after t

What happens next, at t + 1?
● Let’s allow the government to issue more debt in order to pay for the interest services:

Bt+1 = Bt (1 + rt+1 ) = Nt−1 dt−1 (1 + n) (1 + rt+1 )

● The new asset market equilibrium condition will be:

Kt+2 + Bt+1 = Nt+1 st+1 + Nt+1 dt+1

● in per worker terms

1 + rt+1
kt+2 (1 + n) + dt−1 = st+1 + dt+1
1+n

⊳ Assuming again that dt−1 = dt = dt+1 , if we are at the golden rule, this equation is
exactly as in the previous period.
J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 60/62
 Introduction The basic model Optimality Altruism SS FF PAYG Transition

Moving from a pay-as-you-go system to a fully funded system

The equilibrium allocations after t

⊳ In the golden rule the government can perpetually refinance its public debt: its rate
of growth (r ) equals the growth of aggregate output (n).
⇒ The debt to income ratio is constant and thus sustainable.
⇒ The economy does not change and stays in its steady state.

⊳ What would happen with public debt outside the golden rule?
– rt+1 < n [dynamic inefficient region] → The growth of output is larger than the growth of debt
⇒ The debt to gdp ratio goes to zero asymptotically (unless the government uses the fiscal space)
– rt+1 > n [below golden rule] → The growth of output is smaller than the growth of debt
⇒ The debt to gdp ratio explodes (unless the government raises taxes to run a primary balance)

● Side note: some recent papers on public debt sustainability

Reis (JEP 2022), Angeletos, Lian, Wolf (ECTA 2024)

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 61/62

 Introduction The basic model Optimality Altruism SS FF PAYG Transition

Conclusions about Social Security

● In this very stylized model we learn that,

⊳ A funded pensions system generates the same steady state allocations as an economy w/o
public pensions because households are indifferent between public and private savings
(unless the contributions are set too large)

⊳ An economy with a pay-as-you-go system displays in steady state lower aggregate capital
and therefore lower output than an equivalent economy with a funded system. Aggregate
consumption, however, might be larger.

⊳ Moving from a pay-as-you-go system towards a funded system does not necessarily imply
that aggregate capital will increase. The pay-as-you-go system contains an implicit debt that
becomes explicit when ending the system.

● Uncertainty in labor earnings and life length, within-cohort inequality, multiple

periods, distortionary taxation, demographic transitions ...
→ ... all waiting for you in Quantitative Macro

J. Pijoan-Mas Macroeconomics I (CEMFI 2024-2025) 62/62