You are on page 1of 77

Macroeconomic Theory I

Cesar E. Tamayo
Department of Economics, Rutgers University
ctamayo@econ.rutgers.edu

Class Notes: Fall 2010


Contents

I Deterministic models 4
1 The Solow Model 5
1.1 Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2 Balanced growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 The golden rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 Quantitative implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.5 Solow growth accounting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2 Optimal growth 9
2.1 Optimal growth in discrete time . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 The sequential method: Lagrange . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4 The recursive method: dynamic programming . . . . . . . . . . . . . . . . . . . . 10
2.4.1 The Envelope Theorem approach . . . . . . . . . . . . . . . . . . . . . . . 13
2.5 Balanced growth and steady state . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.6 Linearization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.7 Equilibrium growth and welfare theorems . . . . . . . . . . . . . . . . . . . . . . 16
2.8 Extensions to the optimal growth model . . . . . . . . . . . . . . . . . . . . . . . 18
2.8.1 Assets in the OGM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.8.2 The role of government in OGM . . . . . . . . . . . . . . . . . . . . . . . 19
2.9 Optimal Growth in continuous time . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.9.1 Steady State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.9.2 Tobins q . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3 Overlapping generations (OLG) 25


3.1 OLG in economies with production (Diamonds) . . . . . . . . . . . . . . . . . . 25
3.1.1 Log-utility and Cobb-Douglas technology . . . . . . . . . . . . . . . . . . 27
3.1.2 Steady state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.1.3 Golden rule and dynamic ine ciency . . . . . . . . . . . . . . . . . . . . . 29
3.1.4 The role of Government. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.1.5 Social security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.1.6 Restoring Ricardian equivalence . . . . . . . . . . . . . . . . . . . . . . . 31
3.2 OLG in pure exchange economies (Samuelsons) . . . . . . . . . . . . . . . . . . . 32
3.2.1 Homogeneity within generation . . . . . . . . . . . . . . . . . . . . . . . . 32
3.2.2 The role of money . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.2.3 Fiscal policy and the Laer curve . . . . . . . . . . . . . . . . . . . . . . . 36
3.2.4 Monetary equilibria with money growth . . . . . . . . . . . . . . . . . . . 37
3.2.5 Within generation heterogeneity . . . . . . . . . . . . . . . . . . . . . . . 37
3.2.6 The real bills doctrine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

1
II Stochastic models 41
4 Stochastic Optimal growth 42
4.1 Uncertainty in the neoclassical OGM . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.1.1 Non-stochastic steady state . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.1.2 Stationary distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.1.3 Log-linear approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.2 Solution method 1: Blanchard-Khan . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.3 Impulse response functions (IRF) . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

5 RBC models 46
5.1 The baseline RBC model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.1.1 The general case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.1.2 CRRA utility and Cobb-Douglas production . . . . . . . . . . . . . . . . 47
5.1.3 The log-linear system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.2 Labor productivity (King, Plosser & Rebelo, 1988) . . . . . . . . . . . . . . . . . 48
5.3 Solution method 2: Sims GENSYS . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.4 Varieties of RBC models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.4.1 Asset pricing models (Lucas, Shiller) . . . . . . . . . . . . . . . . . . . . . 51
5.5 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.6 Estimation methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.6.1 Generalized method of moments (GMM) . . . . . . . . . . . . . . . . . . . 52

III Appendixes 56
A The dynamic programming method 57
A.1 Guess and verify . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
A.2 Value function iteration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
A.3 Solving for the policy functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
A.4 Properties of the BFE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
A.5 The Envelope Theorem: an application . . . . . . . . . . . . . . . . . . . . . . . . 62

B The Maximum Principle 66


B.1 Discrete time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
B.2 Continuous time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
B.2.1 Current value vs. present value Hamiltonian . . . . . . . . . . . . . . . . 68

C First-Order Dierence Equations and AR(1) 70


C.1 The AR(1) process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
C.1.1 Representation and properties . . . . . . . . . . . . . . . . . . . . . . . . . 70
C.1.2 Conditional Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
C.1.3 Unconditional Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . 71
C.2 Linear First-Order Dierence Equations (FODE) . . . . . . . . . . . . . . . . . . 72
C.2.1 Induction & Geometric Series . . . . . . . . . . . . . . . . . . . . . . . . . 72
C.2.2 Homogeneous part and General solution . . . . . . . . . . . . . . . . . . . 73
C.2.3 Asymptotic Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
C.3 Systems of linear FODE (or VDE) . . . . . . . . . . . . . . . . . . . . . . . . . . 75
C.3.1 Asymptotic stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

2
Summary
These notes summarize the material of a rst semester graduate course in Macroeconomic
theory. The rst sections focus on deterministic growth models and models of overlapping
generations (OLG). Later sections are dedicated to stochastic models, including neoclassical
growth and real business cycles models. The appendixes cover some mathematical material
required for solving simple macro models. The notes are freely based on: Acemoglu (2008),
Romer (2001), Stokey and Lucas (1989), Ljungqvist and Sargent (2004), Dave and De Jong
(2011), Dixit (1990), Levy (1992) and lecture notes from Professor Roberto Chang from Rutgers
University. Naturally, all errors and omissions are my own.
A word on notation: Throughout these notes, x will be used to denote a specic solution
(optima), x will be used for steady states in dierence (dierential) equations and x ^ will be
used for ln (x=x). When applicable, upper cases will stand for economy-wide values of variables,
while lower cases will stand for per-capita (per-eective labor) variables. Finally, in stochastic
models E0 or simply E will denote unconditional expectations while Et will stand for expectations
conditional on information available at t:

3
Part I

Deterministic models

4
Chapter 1

The Solow Model

In the original Solow model, time is continuous and the horizon is innite. Without loss of
generality (WLOG) assume that time is indexed t 2 (0; 1). At each point in time, there is only
one fnal good Y (t).

Assumption 1 The nal good is produced with Harrod-neutral or labour augmenting technology:

Y (t) = F (K(t); A(t)L(t))

Assumption 2 F ( ) is twice di erentiable and F ( K; AL) = F (K; AL). This implies con-
stant returns to scale or no gains from specialization and that one can write the production
function in intensive form:
K
y = f (k) = F ;1
AL

Assumption 3 f (0) = 0; f 0 (k) > 0; f 00 (k) < 0, lim f 0 (k) = 0 and lim f 0 (k) = 1
k!1 k!0

Assumption 4 Savings are a constant fraction of income.

Assumption 5 Existence of a representative household

To ilustrate the assumptions about the production technology:

Example 1 The Cobb-Douglas production function satises assumptions 1-3. To see this con-
sider:

F (K(t); A(t)L(t)) = K (AL)1


K K (AL)1
F ;1 =
AL AL
K
=
AL
f (k) = k

and note that f ( k) = k = f (k). Also note that f 0 (k) = k 1 > 0 since > 0 and k 1 >
0: Likewise, f 00 (k) = ( 1) k 2 < 0 Furthermore lim k 1 = 1 and lim k 1 = 0 since
k!0 k!1
1 < 0:
1 K 1
Remark 1 Notice that: FK (K; AL) = K (AL)1 = AL = k 1
= fk (k)

5
1.1 Dynamics
Suppose that inputs grow as follows:
_
Labor: L(t) = nL(t) so that eln L(t) = L(t) = L(0)ent
_
Technology: A(t) = gA(t) so that eln A(t) = A(t) = A(0)egt
_
Capital law of motion: K(t) = sY (t) K(t)

To derive the last expression in intensive form apply the quotient rule and the product rule
to the expression:

d (K=AL) _
K(t) K(t) h i
= _
A(t)L(t) _
+ A(t)L(t)
dt A(t)L(t) 2
[A(t)L(t)]
sY (t) K(t)
= k(t)n k(t)g
A(t)L(t)
_
k(t) = sf (k(t)) k(t) [n + g + ]

the key equation of the Solow model.

1.2 Balanced growth


Suppose that the economy nds itself in a path in which K(t) and A(t)L(t) are growing at the
same rate. This is a special case of balanced growth which itself induces a so-called steady state1
_
for k since k(t) = 0: Thus
sf (k(t)) = k(t) [n + g + ] (1.1)
and one can see that starting from any level of capital per eective worker, k ! k: Furthermore,
at the level k one can see that:
_
K(t)
_
k(t) =0) =n+g
K(t)

and given the assumption of homogeneity (CRTS):

_ Y_ (t)
k(t) = 0 ) y(t)
_ =0) =n+g
Y (t)
_ _
nally, note that K(t) Y (t)
L(t) = g = L(t) . That is, the economy reaches a Balanced Growth Path (BGP),
where each variable fY; K; A; Lg is growing at a constant rate.

1.3 The golden rule


Suppose starting from the BGP, theres a shift in s. Then k_ jumps since sf (k(t)) > k(t) [n + g + ]
and then falls gradually until k ! knew : In turn YL(t)
(t)
grows by g and k_ > 0 so that YL(t)
(t)
jumps
C
but falls gradually too. Consumption AL , falls by denition since s jumps:

c (t) = (1 s) f (k(t))
1 The generic notion of balanced growth path is a situation in which all variables growth at a constant rate
over time (though this rate need not be the same across variables). A special case of a balanced growth path is
a steady state, in which the growth rate of all variables is equal to zero.

6
To see what happens when the economy reaches the new BGP:

c = (1 s) f (k(t))
= f (k(t)) sf (k(t))
= f (k(t)) k(t) [n + g + ] by denition of BGP

and dierentiate w.r.t. s


@c @ k(s; n; g; )
(s; n; g; ) = f 0 (k(s; n; g; ) (n + g + ) (1.2)
@s @s
@c
since the last term is unambiguously positive, the sign of @s depends on whether f 0 (k(s; n; g; ) ?
(n + g + ). In fact, the BGP level of capital (per AL) that brings:

f 0 (k(s; n; g; )) = (n + g + )
@c
so that @s = 0 (BGP-consumption is at its maximum) is called the golden rule level of capital,
kGold : Therefore, if k > kGold one has that f 0 (k) < f 0 (kGold ) = (n + g + ) and therefore the
economy can increase c by dis-saving.

Exercise 2 Suppose that f (k(t)) = k(t) . Show that for some , the Solow model can predict
overaccumulation of capital in the sense that k > kGold :

Solution. Simply note that (1.2) is now:

@c h i @ k(s; n; g; )
1
(s; n; g; ) = k (n + g + ) (1.3)
@s @s
1
so one needs to show that k (n + g + ) < 0 (and therefore @c=@s < 0). Using (1.1)
one has 1
s 1
k=
n+g+
and replacing k in the golden rule condition, it can be seen that < s ) k > kGold and the
Solow model predicts overaccumulation of capital.

1.4 Quantitative implications


In order to quantify the eect of savings on long-run growth (i.e., BGP y):

@y @ k(s; n; g; )
= f 0 (k)
@s @s
@ k(s;n;g; )
to "quantify" @s it su ces to dierentiate implicitly the (BGP) equation of k_ = 0:

ds df (k) d (n + g + ) dk(s; n; g; )
f (k) + s = k(s; n; g; ) + (n + g + )
ds ds ds ds
@k dk(s; n; g; )
s f 0 (k) + f (k) = (n + g + )
@s ds
@k f (k)
=
@s (n + g + ) sf 0 (k)
substituting:
@y f 0 (k)f (k)
=
@s (n + g + ) sf 0 (k)

7
k[n+g+ ]
simplify multiplying by s
y and using y = f (k) and s = f (k)
from the equation k_ = 0 to
obtain:
s @y kf 0 (k)=f (k)
=
y @s 1 kf 0 (k)=f (k)
s @y k
=
y @s 1 k

1.5 Solow growth accounting


To obtain equation (1) in growth form dierentiate w.r.t. time (recall dY
dt = Y_ ), using the chain
rule and omitting the (t):

Y_ = Fk K_ + Fk L_ + FA A_
Y_ Fk K_ Fk L_ FA A_
= + +
Y Y Y Y
Y_ KFk K _ LFk L AFA A_
_
= + +
Y Y K Y L Y A
Y_ K_ L_ A_
= "k + "L + "A
Y K L | {zA}
Y_ L_ K_ L_
= "k + ( "L 1) + X
Y L K L
_
Note that this equation cannot be estimated with data for K K and obtaining the residual as X
since the residual is correlated (by construction) with capital per worker. Instead, rearrange:
!
y_ K_ L_
X= "k
y K L

and now one has everything measurable in the RHS. This is the key equation of growth account-
ing used to measure Solows residual. Usually "k is the share of capital on the economy and in
a Cobb-Douglas like the example above, "k = :

8
Chapter 2

Optimal growth

2.1 Optimal growth in discrete time


Suppose that, savings are not a xed share of income but rather that households decide how
much to consume and how much to save on each period. For now, assume that there is neither
technical change nor poppulation growth (g = n = 0) so that aggregate production takes the
form F (Kt ; Lt ): As before, suppose that F ( ) is homogeneous of degree one so that production
K
can be written in per labor units: y = f (k) = F AL ;1 :
Next, suppose there exists a representative household (RH) or, equivalently, that preferences
can be aggregated economy-side. Then the RH solves the (discrete-time) problem:
1
X
t
max u(ct )
ct ;kt+1
t=0
s:t: (2.1)
ct + kt+1 f (kt ) + (1 )kt
k0 given
1
so that the RH chooses a consumption-saving plan fct ; kt+1 gt=0 under the condition that the
resource constraint holds at every t = 0; 1; :::

2.2 Assumptions
Assumption 6 Assumptions 1-3 about f ( ) from section 1.1 hold.

Assumption 7 The objective function u( ) is continuous, twice di erentiable and satises the
Inada conditions: u(0) = 0; u0 (k) > 0; u00 (k) < 0; lim u0 (k) = 0 and lim u0 (k) = 1
k!1 k!0

Assumption 8 The constraint set fkt+1 j kt+1 f (kt ) + (1 )kt ct g is compact, convex.

Assumption 9 Preferences are additive. This ensures dynamic consistency.

Under these assumptions any organization of markets and production will yield the same
competitive equilibrium allocation. Hence the competitive equilibrium is unique and the rst
and second welfare theorems hold. That is, the competitive equilibrium will be Pareto e cient
and the planners problem can be descentralized as the outcome of a competitive equilibrium as
will be shown below.

9
2.3 The sequential method: Lagrange
The problem above can be approached by the innite-horizon Lagrange method:
1
X
t
L= fu(ct ) + t [f (kt ) + (1 )kt ct + kt+1 ]g
t=0

with F.O.C.:
@L t 0
= 0 =) u (ct ) = t (2.2a)
@ct
@L
= 0 =) t = t+1 (f 0 (kt ) + 1 ) (2.2b)
@kt+1
@L
= 0 =) ct + kt+1 = f (kt ) + (1 )kt (2.2c)
@ t+1
t+1 0
replacing t and using the fact that t+1 = u (ct+1 ) one obtains the Euler Equation:

u0 (ct ) = u0 (ct+1 ) (f 0 (kt ) + 1 ) (2.3)

Along with the resource constraint (F.O.C. (2.2c)) these two dierence equations fully charac-
terize the solution to the optimal plan. The boundary condition required for the solution to
(2.2c)-(2.3) to exist is the transversality condition:
t 0
lim t kt+1 = lim u (ct )kt+1 = 0
t!1 t!1

2.4 The recursive method: dynamic programming


This problem can also be solved as a discrete time, deterministic, stationary, dynamic program-
ming one. In fact, when the problem is stationary (i.e. the problem faced at every period is
identical), the sequential and recursive methods are equivalent. If one separates the problem into
periods, each problem depending on the state kt and consisting on choosing controls ct ; kt+1 ; the
Bellman functional equation (BFE) for the iterated (recursive) one-period optimization problem
is:

V (kt ) = max [u(ct ) + V (kt+1 )]


ct
s:t: (2.4)
yt = ct + it = f (kt )
kt+1 = (1 )kt + it
k0 given

In fact, this problem can be seen to have only one control variable since choosing ct is
equivalent to choosing kt+1 given the combination of the resource constraint and the capital
accumulation equation. To be more precise notice that the problem is (2.4) is equivalent to:

V (kt ) = max [u(f (kt ) kt+1 (1 )kt ) + V (kt+1 )] (2.5)


kt+1

V (kt ) = max [u(ct ) + V ((1 )kt + f (kt ) ct )] (2.6)


ct

and associated transversality condition:


t+1
lim V 0 (kt+1 )kt+1 (2.7)
t!1

10
Under familiar assumptions (see Appendix A), the solution to the Bellman equation, will yield
time-invariant, sate-dependent rules for consumtion and capital accumulation, i.e., policy func-
tions:

ct = h(kt ) (2.8)
kt+1 = g(kt ; ct ) = g(kt ; h(kt )) (2.9)

and notice that g( ) is precisely the state-transition function that results after consolidating the
two restrictions in (2.4); that is:

kt+t = (1 )kt + f (kt ) ct = g(kt ; ct )

However, since (2.4) is a functional equation, one needs to solve for V ( ) in order to obtain
(2.8)-(2.9). There are mainly three approaches to solve for V ( ):

Guess and verify the value function


Value function iteration or method of successive approximations, and,
Policy functions iteration or Howards improvement algorthm.

These three approaches are developed in Appendixes A.1-A.3, and some examples are pro-
vided. The main conditions for the existence of a unique solution, V ( ); to the BFE (2.4) are:

Condition 1 Assumptions 1-3 at the begining of this chapter hold.

Condition 2 The control space and the state space are convex, compact sets,

Condition 3 The operator V 7! T (V ) in (2.4) maps the set of continuous, bounded, real-valued
functions M into itself.1

Condition 4 The operator V 7! T (V ) (2.4) is a contraction mapping dened on a complete


metric space of continuous, bounded, real-valued functions M.

Condition 3 requires in turn that u ( ) ; f ( ) are continuous and bounded and that the corre-
spondence (kt ) = fct jct (1 )kt + f (kt ) kt+1 g is UHC, LHC, non-? and compact-valued.
Appendix A.4 provides the relevant proofs and also proves some additional properties of the oper-
ator T ( ) : Condition (iv) requires a contraction mapping; recall that a contraction is a mapping,
'; on a metric space (W; d) that satises d (' (f ) ; ' (g)) d (f; g) for all f; g 2 M and some
< 1: Since M above is a complete metric space, conditions (i)-(iv) would su ce for a unique
solution to the BFE since the Banach xed point theorem asserts that every contraction mapping
on a complete m.s. has a unique xed point, i.e., 9 V such that V = T (V ). Moreover, the
sequence dened by Vi+1 = T (Vi ) converges to the unique xed point V : Now, the Blackwell
conditions for an operator to be a contraction mapping are:

Monotonicity: ' > (w.r.t. box metric)) T (') > T ( ) whenever '; 2 M.
Discounting: for any constant function A = h( ), and ' 2 M one has T ('+A) T (')+ A
for some < 1:
1 In this case the assumption that V is bounded is redundand since the compactness of the control space

ensures this property.

11
Both conditions are easily seen to hold in this case: Assuming that one obtains V ( ) by any
of the abovementioned methods, it is easy to obtain the F.O.C. for the problem (2.6):
V (kt ) = max [u(ct ) + V ((1 )kt + f (kt ) ct )]
ct

the associated F.O.C. is:


u0 (ct ) V 0 (kt+1 ) = 0 (2.10)
and the resource constraint:
ct + kt+1 = f (kt ) + (1 )kt (2.11)
0
At this point one can use the knowledge of V ( ) in order to obtain V ( ) and solve the system of
dierence equations (2.10)-(2.11).
Example 3 (Guess and verify) Cobb-Douglas technology with full depreciation and logarith-
mic utility. The problem is:
V (kt ) = max [ln ct + V (kt+1 )]
ct ;kt+1

s:t: (2.12)
ct + kt+1 = kt
k0 given
the F.O.C is therefore:
1
= V 0 (kt+1 )
ct
1
= V 0 (kt+1 )
kt kt+1
If one guesses the form of V ( ) as:
V (kt ) = F + G log kt
so that one replaces for V 0 (kt+1 ) in the F.O.C.:
1 G
=
kt kt+1 kt+1
and using the resource constraint arrives to the policy functions (with undetermined coe cient
G):
G
kt+1 = k
1+G t
G
ct = 1 kt
1+G
now using this in the BFE:
G G
V (kt ) = F + G log kt = log 1 kt + F + G log k
1+G 1+G t
which solving for the undetermined coe cients F,G yields:

G =
1
log (1 )+ (1 ) log( )
F =
1

12
1
so that nally one can generate optimal plans fct ; kt+1 gt=1 from the fully specied policy func-
tions:

ct = (1 ) kt
kt+1 = ( ) kt

Finally, note that the transversality condition is satised:

t+1 t+1
lim V 0 (kt+1 )kt+1 = lim =0
t!1 t!1 1

2.4.1 The Envelope Theorem approach


A dierent avenue that avoids dealing with the value function explicitely is as follows. If con-
ditions 1-4 above regarding the objects in the problem are satised (i.e., concavity, convexity,
continuity, compactness, monotonicity, discounting), then the unique solution to the BFE V ( ),
would be continuous, concave, increasing and importantly, di erentiable. Hence, the Envelope
theorem applies and one can use the envelope condition for the value function (see the Appendix
A for a more elaborate example). To see how the Envelope theorem works in this particular
case, recall the policy functions:

ct = h (kt )
kt+1 = g(kt ; ct )
= (1 )kt + f (kt ) h (kt )

then the value function becomes:

V (kt ) = u(h (kt )) + V ((1 )kt + f (kt ) h (kt ))

dierentiating w.r.t. kt :

@u(h (kt )) @h (kt ) @f (kt ) @h (kt )


V 0 (kt ) = + V 0 (kt+1 ) (1 +
@h (kt ) @kt @kt @kt
@u(h (kt )) @h (kt ) @f (kt ) @h (kt )
= + V 0 (kt+1 ) (1 + V 0 (kt+1 )
@h (kt ) @kt @kt @kt
@f (kt ) @h (kt ) @u(h (k t ))
= V 0 (kt+1 ) (1 + + V 0 (kt+1 )
@kt @kt @h (kt )

but the F.O.C. (2.10) implies:

@u(h (kt ))
= u0 (ct ) = V 0 (kt+1 )
@h (kt )

and therefore:
@f (kt )
V 0 (kt ) = V 0 (kt+1 ) (1 +
@kt
@f (kt )
= u0 (ct ) (1 +
@kt

and since this holds for every period:

V 0 (kt+1 ) = u0 (ct+1 ) [f 0 (kt+1 ) + (1 )]

13
replacing in the F.O.C. one obtains once more the system of dierence equations that (under
the assumptions above) will characterize the solution to the RH problem:

u0 (ct ) = u0 (ct+1 ) [f 0 (kt+1 ) + (1 )] (2.13)


ct + kt+1 = f (kt ) + (1 )kt (2.14)

Alternatively, one can use the policy functions (2.8)-(2.9) to express the Euler equation in terms
of kt :
u0 (g(kt )) = u0 (g(h(kt ))) [f 0 (h(kt )) + (1 )]
Finally, note that V 0 (kt+1 ) is something similar to a shadow price associated with the resource
constraint.

Remark 2 Note that in this problem it is the case that 6= 1 and therefore, f 0 (kt ) + 1 = Rt .
If one was to consider the case where = 1, as will be the case in some sections below then one
would have: f 0 (kt ) = rt = Rt :

2.5 Balanced growth and steady state


As mentioned earlier, the generic notion of balanced growth path is a situation in which all
variables grow at a constant rate over time (though this rate need not be the same across
variables). Perhaps the simplest example of balanced growth that solves the optimal growth
problem is:

Example 4 Consider the so-called AK model.under CRRA utility and full depreciation. The
social planners problem is:
1
X
t c1t
max
t=0
1
s:t: yt ct + it
yt = f (kt ) = Akt
it = kt+1

the BFE for this problem is:

V (kt ) = max fu (Akt kt+1 ) + V (kt+1 )g


kt+1

with F.O.C.:
u0 (ct ) + V 0 (kt+1 ) = 0
and envelope condition:
V 0 (kt+1 ) = u0 (ct+1 ) A
so the Euler equation becomes:
u0 (ct ) = u0 (ct+1 ) A
or, using the functional form for u (ct ) :

ct+1
= A
ct

Since ct = Akt kt+1 :


(Akt+1 kt+2 ) = A (Akt kt+1 )

14
A second order di erence equation that, without a (on k) that, without a terminal condition has
multiple solutions for any given k0 : The interest is in that which satises the TVC (2.7) which
in this case becomes (using the EC):
t+1 t+1 0
lim V 0 (kt+t ) kt+1 = lim u (ct+1 ) Akt+1
t!1 t!1

Now to nd conditions that ensure holding of the TVC note that in a BGP capital and consump-
tion grow at the same constant rate so kt+1 = kt and ct+1 = ct for some : Thus:
( ct ) 1=
= A) = ( A)
ct

which brings positive growth iif > 1=A. Next, rewrite u0 (ct+1 ) = ct = A and kt+1 = t+1
k0
in the TVC:

t+1 0 t+1 ct t+1


lim u (ct+1 ) Akt+1 = lim A k0
t!1 t!1 A
t+1 ct t+1
= lim k0
t!1
t
= lim ( ) ct k0
t!1

1= 1
then the TVC holds iif < 1 )if ( A) < 1: Hence by assuming that <A 1+ one can
ensure that there is balanced growth and the TVC holds.

A special case of a balanced growth path is a steady state, in which the growth rate of all
variables is equal to zero. From (2.13)-(2.14) one can compute the steady state by noting that
in the SS, xt = xt+1 = x for x = fc; k; yg. Hence:

u0 (c) = u0 (c) f 0 (k) + (1 )


1
= f 0 (k) + (1 )

dening a modied golden rule for capital accumulation, and

c k = f (k) = y

2.6 Linearization
Next it is possible to study the behavior of the model around the steady state (SS). Recall that
in the SS, xt = xt+1 = x for x = [c; k]0 and x^t = (xt x)=x . So, linearize the system around
the SS. Recall that to linearize G(x; y) = 0 around the SS (x; y):
@G @G
(x; y)x x
^t + (x; y)y y^t = 0
@xt @yt
Linearizing (2.13):
c^t = c^t+1 + k^t+1 (2.15)
where division by u0 (c) on both sides has been used and the fact that 1= = f 0 (k) + (1 ) from
the SS and ; are the elasticities of u0 ( ) and f 0 ( ), respectively. Finally, the linear constraint
is linearized as:

ct + k k^t+1
c^ = f (k) + (1 ) k k^t
ct + k^t+1
!^ = [ +!+ 1] k^t (2.16)

15
where ! = yc , = ky and is the elasticity of f ( ):Now one can suppose that the linearized policy
functions are of the form:

c^t = ^
1 kt
k^t+1 = ^
2 kt

so that replacing in (2.15)-(2.16) one obtains:

1 = +1 2 2
! 1 + 2 = [ +!+ 1]

and one can solve for the undetermined coe cients 1 ; 2 as functions of parameters of
the model ( ; !; ). Note that the rst of these equations will be a cuadratic one so one has
two solutions, from which one selects 2 < 1 since this is the only solution that satises the
transversality condition. With this log-linear policy functions and the "true" coe cients as
ct ; k^t+1 g1
functions of the parameters of the model, one can generate optimal sequences f^ t=0 , that
is sequences of the variables in deviation from SS form.

2.7 Equilibrium growth and welfare theorems


The solution to the optimal growth model can in fact be deduced as the outcome of a competitive
equilbrium. To see this, state the problem of the RH and the rm separately.

Households
The representative household maximizes lifetime discounted utility subject to its resource con-
straint. Households own the factors of production k; l and own the rms. For simplicity, suppose
there is full depreciation ( = 1). At each period, the RH receives income from renting all of its
available capital, working all its endowed labor and earning prots from the rms.2 With this
income, the RH and decides how much to consume and how much to invest (save):
1
X
t
max u(ct )
ct
t=0
s:t:
h
ct + kt+1 rt kth + wt lth + t = yth

Firms
Firms produce a single good by renting production factors from the RH and maximize prots
subject to their production technology:
1
X 1
X
max t = max pt ytf wt ltf rt ktf
t=0 t=0
s:t:
ytf F (ktf ; ltf )

where F ( ) is continuous, dierentiable, strictly increasing and homogeneous of degree one. Since
there is only one good in the economy, there are no relative prices and one can set pt = 1. Also,
2 The assumption that households work their entire endowment of labor reects the fact that there is no
disutility of labor in this model. The RBC models surveyed in the sections below relax this assumption so that
labor becomes in fact a choice variable.

16
since there is no discounting, lifetime prots are maximized , prots are maximized at every
period t:3

Equilibrium
For simplicity suppose that lth = 1: Then a competitive equilibrium consists of a set of prices
1 1
fpt = 1; wt ; rt gt=0 and allocations fkt ; lt = 1; yt ; ct gt=0 such that 8 t :
1. The rm maximizes prots. To do so, note that since F ( ) is strictly increasing, the
technology constraint will hold with equality ytf = F (ktf ; ltf ) . Thus, the F.O.C.s of the
rm are:
@ (ktf ; ltf )
= 0 =) wt = Fl (ktf ; ltf )
@ltf
@ (ktf ; ltf )
= 0 =) rt = Fk (ktf ; ltf )
@ktf

2. The RH maximizes utility. The F.O.C.s for the RH are, as before:


u0 (ct ) = u0 (ct+1 )rt+1
ct + kt+1 = rt kth + wt lth +
h
t

3. Markets clear in all periods (t = 1; 2:::):

yth = ytf = F (ktf ; ltf )


lth = ltf = 1
kth = ktf

Next, replace the F.O.C.s for the rm in the prot function at t and recalling that lt = 1:

t = F (kt ; lt ) Fl (kt ; lt ) Fk (kt ; lt )kt


and because F ( ) is homogeneous of degree one, Eulers theorem (x rf (x) = f (x)) implies
that t = 0 so that:
wt = F (kt ; lt ) Fk (kt ; lt )kt
P1
and t=0 t = 0 . Replacing in the F.O.C.s for the RH yields:
u0 (ct ) = u0 (ct+1 )Fk (kt ; lt )
and
ct + kt+1 = Fk (kt ; lt )kt + F (kt ; lt ) Fk (kt ; lt )kt
= F (kt ; lt )
which are of course, the same optimality conditions derived under the centralized approach in
the section above. Hence one has found a vector of prices that delivers the (planned) Pareto
optimal allocation which results as a solution to (2.13)-(2.14). That is, the optimal allocation
has been descentralizedas a competitive equilibrium of the economy This is an ilustration of
the second fundamental theorem of welfare economics. 4
.
3 It is straightforward to extend this model to the case where rms discount future prots. A natural candidate
1
for discounting would be R where R is the gross interest rate (in this economy all assets would earn R).
4 Recall that the rst welfare theorem states that whenever households are non-satiated, a competitive equi-

librium allocation is Pareto optimal.

17
2.8 Extensions to the optimal growth model
2.8.1 Assets in the OGM
Recall the budget constraint for the RH is, in general:

kt+1 = (1 + rt )kt + wt lt + t ct

and if one assumes lt = 1 8 t:

kt+1 = (1 + rt ) kt + wt ct + t (2.17)
| {z }
Rt = gross return

Now, allow for assest at that HH carry from previous periods. Note that in aggregate at = kt +bt
where we may have bt < 0 at some t. So we can re-write the ow budget constraint as:

ct + at+1 = Rt at + wt + t
ct at+1
+ = at + wt + t
Rt Rt
At this point one carries out forward substitution of at+1 from the equation for at+2 , then sub-
stitute the latter from the equation for at+3 and so on. For some nite T 1; the intertemporal
budget constraint (IBC) is given by:

XT XT
1 aT +1 1
ct + = (1 + r0 )a0 + (wt + t) (2.18)
t=0
R t RT t=0
R t

where:
t
Y t
Y
Rt = Rj = (1 + rj )
j=1 j=1

But since were assuming an innite horizon, we must prevent aRT T+1 ! 1 as T ! 1 (build
P1
debt forever) which would result in t=0 R1t ct ! 1: So we impose the additional no Ponzi
games condition:
aT +1
lim =0
T !1 RT

so that when the horizon is innite, taking limits on both sides of (2.18) and using the no-Ponzi
condition, the IBC can be expressed as:
X1 X1
1 1
ct = (1 + r0 )a0 + (wt + t) (2.19)
t=0
R t | {z } t=0
R t
| {z } | {z }
P V consumption = initial asset income + PV non-asset income

Note that a stream of consumption that satises the (IBC) also satises the ow constraint at
1
each t. To see this, consider a consumption plan f^ct gt=0 that satises the IBC. At some , by
(2.18) the household will have accumulated assets:

a +1 X 1
= (1 + r0 )a0 + (wt + t c^t )
R t=0
Rt
X1 1 1
= (1 + r0 )a0 + (wt + t c^t ) + (w + c^ ) (2.20)
t=0
Rt R

18
obviously since:

a X1 1
= (1 + r0 )a0 + (wt + t c^t )
R 1 t=0
Rt
X1 1 a
) (wt + t c^t ) = (1 + r0 )a0 +
t=0
Rt R 1

we can replace this in (2.20) to get:


a +1 1 a
= (w + c^ ) +
R R R 1

simply multiply this last expression by R on both sides to obtain:


a +1 =w + c^ + R a
which is of course the sequential BC for period .
Using the IBC just derived, the optimal growth problem can be stated more compactly as:
1
" 1 1
#
X X 1 X 1
t
L= u(ct ) + (1 + r0 )a0 + (wt + t ) ct
t=0 t=0
Rt t=0
Rt

with associated F.O.C. (noting that in equilibrium t = 0):


@u(ct ) t 0
) u (ct ) =
@ct Rt
@u(ct+1 ) t+1 0
) u (ct+1 ) =
@ct+1 Rt+1
so equating them yields the same Euler equation derived earlier:
u0 (ct ) = Rt+1 u0 (ct+1 )
= u0 (ct+1 )(1 + r0 )
which naturally relies on the fact that in equilibrium all rates of return are equilized (so all
assets receive and are discounted to the same rate).

2.8.2 The role of government in OGM


To introduce the government, derive restrictions similar to (2.17) and (2.19):
bt+1 = (gt t) + R t bt
where bt+1 = debt outstanding at t + 1, (gt t ) = net income or primary decit at t and
Rt bt =debt service in t: [ Note that the same rate of return is used for the government, all
nancial assets and capital which is the case if there is no uncertainty (no default risk) and
perfect complete nancial markets!]. Next, set up the IBC for the government:
1
X
1 t gt
(1 + r0 )b0 lim bt+1
t!1 Rt Rt Rt
t=0

Naturally, if we allow limt!1 R1t bt+1


! 1, the LHS of this inequality approaches 1 and
therefore the government could run primary decits forever. Hence, the following no-Ponzi
condition for the Gvt.is imposed:
1
lim bt+1 = 0
t!1 Rt

19
so that the IBC for the government is:
1
X t gt
(1 + r0 )b0
t=0
Rt Rt

Note that now HH would have to pay taxes so:


X1 1
X
1 wt + t t
ct = (1 + r0 )a0 +
t=0
Rt t=0
Rt Rt

and replacing PV taxes from the Gvt. IBC:


X1 1
X X1
1 wt + t gt
ct = (1 + r0 )a0 + (1 + r0 )b0
t=0
R t t=0
Rt t=0
R t

X1 X1
wt + t gt
= (1 + r0 )k0 + (2.21)
t=0
Rt t=0
R t

where the denition of assets a0 = b0 + k0 has been used. This is a crucial result, for, it shows
that only the (PV) level of government spending matters and not the means through which it
is nanced (debt or taxes). This is naturally a Ricardian Equivalence statement.
The new IBC for the HH (2.21) can be used again in the Lagrangian as above to derive the
system of equilibrium conditions for the optimal growth problem. Furthermore, one can see that
for a given initial stock a0 , at any t :

kt + bt = at
= Rt a0

and since a0 is constant, " bt )# kt and/or " Rt which are both ways of crowding out.

Example 5 (OGM with assets and government) The problem can be stated as:
1
X
t
max u(ct )
ct
t=0
s:t:
at+1 = (1 + r)at + wt ct

so that the state variable is at and the control variable again ct : Note that this problem is not
well dened even with the transversality condition:
t+1
lim V 0 (at+1 )at+1 = 0
t!1

since it may happen that at+1 ! 1. Therefore, one needs the no-Ponzi condition introduced
above:
aT+1
lim =0
RT
T !1

so that the TVC holds. With the problem well specied, the solution can be found by solving:

V (at ) = max fu((1 + r)at + wt at+1 ) + V (at+1 )g

with F.O.C.:

u0 ((1 + r)at + wt at+1 ) = V 0 (at+1 )


u0 (ct ) = V 0 (at+1 )

20
and envelope condition:

V 0 (at+1 ) = u0 ((1 + r)at+1 + wt at+2 )(1 + r)


= u0 (ct+1 )(1 + r)

and the usual Euler equation:


u0 (ct ) = u0 (ct+1 )(1 + r)

2.9 Optimal Growth in continuous time


State the problem (2.1) in continuous time:
Z1
t
max u(c(t))e dt
c(t);I(t)
0
s:t: (2.22)
c(t) + i (t) f (k(t))
_
k(t) = i (t) k(t) (2.23)
k(0) given

The continuous time problem is actually easier to solve using the tools of optimal control; a
review of The Maximum Principle and the Hamiltonian approach is presented in Appendix B.
Set up the (present-value) Hamiltonian:
t
H = u(c(t))e + (t) [f (k(t)) k(t) c(t)] (2.24)

where (t) is the co-state variable. The condition for c(t) to maximize the Hamiltonian is:

@H
i) = 0 =) u0 (c(t))e t
= (t) (2.25)
@c(t)

solving this yelds c (t) as a function of the co-state and parameters. Next, we can replace c (t)
in (2.24) to obtain the maximum value function of the Hamiltonian, H , dierentiate w.r.t. state
and co-state variables. Alternatively, simply state the Pontryagin conditions similar to those
derived in the Appendix B (B.8)-(B.9):

@H
ii) _ (t) = = (t) [f 0 (k(t)) ] (2.26)
@k(t)
_ @H
iii) k(t) = = f (k(t)) k(t) c(t) (2.27)
@ (t)

Conditions (2.25)-(2.26)-(2.27) along with the TCV:

lim (t)k(t) = 0
t!1

are the three dierential equations (and terminal condition) that characterize the solution to
the HH problem.

Example 6 Suppose that u (c (t)) = ln (c (t)). Then from (2.25):


t
e
c(t) =
(t)

21
so that replacing in (2.27) we get the two di erential equations on k and that characterize the
solution:
t
_ e
k(t) = f (k(t)) k(t)
(t)
0
_ (t) = (t) [f (k(t)) ]

Alternatively, one can solve the system in terms of dierential equations in c and k, which is
more consistent with the idea of policy rules described in previous sections. To do so, dierentiate
(2.25) w.r.t. time:
u00 (c(t))e t c(t)
_ e t u0 (c(t)) = _ (t) (2.28)
and replace in (2.26) to obtain:

u00 (c(t))e t
c(t)
_ e t 0
u (c(t)) = (t) [f 0 (k(t)) ]
u00 (c(t))e t
c(t)
_ e t 0
u (c(t)) = t 0 0
e u (c(t)) [f (k(t)) ]

which upon rearranging:


c(t)
_ f 0 (k(t))
= (2.29)
c(t) [c(t)u00 (c(t))=u0 (c(t))]
where the denominator is naturally, the Arrow-Pratt CRRA coe cient: This condition tells us
that whenever f 0 (k(t)) is "large", which happens when k(t) is "low", we have c=c
_ > 0 so that
the HH consumes less today and more in the future. Likewise, c=c
_ < 0 if is large enough which
suggests that the HH is more "patient".
Finally, the TVC is:

lim (t)k(t) = 0
t!1
t 0
lim e u (c(t))k(t) = 0
t!1

2.9.1 Steady State


Recall, in SS c(t) _
_ = 0 and k(t) = 0. Therefore:
c(t)
_
= 0 =) f 0 (k) = (2.30)
c(t)
which is, again, a modied Golden Rule for capital accumulation and, once more, independent
from the shape of u( ): Next:
_
k(t) = 0 =) f (k) c = k
_
so that SS-investment just breaks-even (making k(t) = 0).

Remark 3 Note that (t) is the shadow price of capital, i.e., it measures the impact of a small
increase in kt on the optimal value of the program. By comparison, V (kt ) is the value of of the
optimal program from t given the level of capital kt : Therefore, we have that:

(t) = V 0 (kt )

and note that _ (t) represents the appreciation of capital since its the change in the value of a
unit of the state variable.

Exercise 7 In order to make meaningful comparisons with the Solow model, reintroduce techni-
cal change and poppulation growth. (i) Write the equilibrium conditions of the OGM considerng
these features and (ii) Show that along the balanced growth path, k < kGold where kGold is the
Golden Rule level of (per-e ective-labor) capital in the Solow model.

22
Solution. (i) First, derive the analogue to (2.30) after reintroducing poppulation growth and
technological progress. To do so, write the constraints in absolute levels:

C (t) + I (t) = F (K (t) ; A (t) L (t))


K_ (t) = I (t) K(t)

derive the capital accumulation equation:


d d
_ d K (t) [A (t) L (t)] dt K (t) K (t) dt [A (t) L (t)]
k(t) = = 2
dt A (t) L (t) [A (t) L (t)]

and thus:

_ K_ (t)
k(t) = k (t) (g + n)
A (t) L (t)
I (t) K(t)
= K (t) (g + n)
A (t) L (t)
= i (t) ( + g + n) k (t)
_
replace with i (t) = f (k(t)) c (t) so that k(t) = f (k(t)) c (t) ( + g + n) k (t) : Therefore
conditions (2.26) and (2.29) become:

@H
_ (t) = = (t) [f 0 (k(t)) g n]
@k(t)
c(t)
_ f 0 (k(t)) g n
=
c(t)

(ii) Now, along the BGP c(t)


_ = 0 so

f 0 (k) = + +g+n
> +g+n
= kGold

which in turn means that k < kGold :

2.9.2 Tobins q
Consider the following model of investment under adjustment costs. Assumptions 2 and 3 in
section 1.1 are satised. Theres a single nal good and therefore one can normalize its price
to 1. The problem is that of a RH which must decide how much to consume and how much
to invest in the representative rm at each given t. However, installing capital is costly. When
there are quadratic investment adjustment costs the problem can be stated as (for simplicity
ignore depreciation):
Z1
t
max e u(c(t))dt (2.31)
fc(t);I(t)g1
t=0
0
_
s:t. k(t) = I(t)
I 2 (t)
c(t) + I(t) = f (k(t))
2 k(t)

where I(t) is investment and > 0 is a constant:

23
Note that in this problem, the control variables are c(t) and I(t) (or kt+1 in the OGM), while
the state variable is k(t). On what follows, the present-value optimization problem is solved
and then it is expressed in current value terms since the latter form lends itself to intuitive
interpretation. First, we get rid of c(t) by using the second constraint. Notice that we can do
this only because the constraint is specied with equality. Notice also that after eliminating this
constraint, the Lagrangian and the Hamiltonian are obviously the same so that in the language
of Appendix B, G ( ) dissapears and LI = HI : Thus, set up the present-value Hamiltonian:
I 2 (t)
H pv = e t
u f (k(t)) I(t) + (t)I(t)
2 k(t)
The F.O.C. w.r.t. the control is simply obtained:

t 0 I(t)
(t) = e u (c (t)) 1 + (2.32)
k(t)
Next, using the Pontryagin conditions corresponding to the present-value problem (B.8)-
(B.9):
" #
2
t 0 0 I(t)
_ (t) = e u (c (t)) f (k(t)) + (2.33)
2 k(t)
_
k(t) = I(t) (2.34)
Or, using the F.O.C. to solve for I (t) we can rewrite (2.34) as:

_ k (t) e t (t)
k(t) = 1
u0 (c (t))
And just as in the discrete time problem, the TVC is given by: limt!1 (t)k(t): Naturally,
with an explicit functional form for u, we could solve for c (t) ; I (t) using the F.O.C. (2.32)
and the constraint, replace this in (2.33)-(2.34) to obtain a pair of dierential equations on (t)
and k(t) that would characterize the solution to the problem.
An interesting avenue to take in this problem is to express the equilibrium conditions in
current-value terms. To do so, multiply (2.33) by e t on both sides (the other conditions do
not involve (t)) and dene q(t) = e t (t). Then, since q(t) _ = _ (t)e t + (t)e t , one has that
_ (t)e t = q(t)
_ q(t) and therefore the Pontryagin conditions can be written:
" #
2
I(t)
q(t)
_ q(t) = u0 (c (t)) f 0 (k(t)) + (2.35)
2 k(t)

_ k (t) q(t)
k(t) = 1 (2.36)
u0 (c (t))
Finally, using (2.35) we can arrive at:
" #
2
0 0 I(t)
q(t)
_ q(t) = u (c (t)) f (k(t)) +
2 k(t)
2 3
Z 1 2
I(t)
q(t) = e u (c(s)) 4f 0 (k(t)) +
(s t) 0 5 ds
t | {z } 2 k(t)
| {z }
disc. marg. util of output marg. prod. of k - marg. adj cost (2.37)
that is, Tobins q summarizes the informarion of the discounted social benet of installing an
additional unit of capital.

24
Chapter 3

Overlapping generations (OLG)

3.1 OLG in economies with production (Diamonds)


Assume that a generation is born in every period of time. Time is discrete indexed t = 1; 2:::
Each generation lives two periods. Therefore, at each t, there are two generations alive; "young
households" and "old households", call them HH types 1 and 2. For completeness, suppose that
at period t = 1 a generation is already alive (agent type 0). Therefore generation t is that born
in period t: Each new generation is larger than the previous one by a factor of (1 + n), n 2 (0; 1).
Therefore:
Lt = (1 + n)Lt 1
In its rst year of life, each HH works, saves and consumes. In its second year of life each HH
only consumes. Therefore:

c1;t ! period t consumption of a typical HH from generation t ("youngs")


c2;t+1 ! period t + 1 consumption of a typical HH from generation t ("olds")

Production is carried out by the use of capital, technology and "young" agents as the labor force
with Harrod-neutral production function:

F (Kt ; At Lt )

Moreover, production technology satises assumptions 1)-3) in section 1.1 and utility satises
assumptions 2)-4) in section 2.2. Technology follows an exogenous growth process:

At = (1 + g)At 1

Hence, there are three markets; nal goods, capital goods and labor. Markets are competitive
and for simplicity, there is full depreciation ( = 1). Acordingly, t = 0, FK = 1 + rt = Rt and
FL = wt : In period t, each HH from generation t works, receives (technology-enhanced) labor
income, consumes and saves1 :
At wt = c1;t + st
Generation ts savings are rented in the form of capital for production at t + 1 so that capital
available is:
Kt+1 = Lt st = St (3.1)
1 Since only "youngs" save on each period, there is no s
2;t ("olds" dont save), to simplify notation st =
s1;t =total savings of the economy at t:

25
and are returned with the corresponding rental income. Generation t (i.e. agent type 2 in period
t + 1) then consumes the proceedings:

c2;t+1 = Rt+1 st
= (1 + rt+1 )st
= (1 + rt+1 ) (At wt c1;t )

Generation t then maximizes its (discounted) lifetime utility from consumption:

max U (c) = u(c1;t ) + u(c2;t+1 ) (3.2)


c1;t ;c2;t+1
s:t:
c2;t+1
At wt = c1;t + (3.3)
Rt+1
alternatively, one the RH solves the unconstrained optimization problem:

max U (c) = u(c1;t ) + u(Rt+1 (At wt c1;t ))


c1;t

both yielding the same F.O.C. and Euler equation:

u0 (c1;t ) = u0 (c2;t+1 )Rt+1 (3.4)

which is analogous to the one found in the OGM. Equations (3.3)-(3.4) are the pair of dierence
equations describing the solution to the typical generation t HH problem. Next, note that the
Euler equation can be expressed as:

u0 (c1;t ) = ( ; u0 (c2;t+1 ); Rt+1 )

and that u00 < 0 ) u0 1


(c1;t ) = c1;t . Therefore:

u0 1
(c1;t ) = u0 1
(c2;t+1 ) 1
( ; u0 (c2;t+1 ); Rt+1 )
c1;t = u0 1
(c2;t+1 ) 1
( ; u0 (c2;t+1 ); Rt+1 )

hence, if one replaces in:

At wt = c1;t + st
st = At wt u0 1 (c2;t+1 ) 1
( ; u0 (c2;t+1 ); Rt+1 )
st = (At wt ; Rt+1 )
(+) (?)

where is called the savings function. Consider a rise in labor income At wt , then, everything
else constant, the properties of u ( ) imply that consumption in both periods would rise, which
implies that st increases; thus the sign of @st =@At wt is unambiguous. However, consider a rise
in Rt+1 . There are two eects to consider. First, since the opportunity cost of consumption
in t rises, the HH may want to substitute current for future consumption (substitution eect).
Second, a rise in Rt+1 has an income eect since each unit saved yields higher return so the
HH will want to consume more on both periods. The total eect is ambiguous as is the sign of
@st =@Rt+1 :
Next, to derive the law of motion of capital come back to (3.1):

Kt+1 = Lt st = St
Kt+1 = Lt (At wt ; Rt+1 )
= Lt (At wt ; FK )

26
or in per-eective worker terms (i.e., dividing by At+1 Lt+1 since were deriving t + 1 capital):

Kt+1 Lt (At wt ; FK(t+1) )


=
At+1 Lt+1 At+1 Lt+1
((1 + g) 1 wt ; fk(t+1) )
kt+1 =
(1 + n)
((1 + g) 1 (f fk kt ); fk(t+1) )
kt+1 =
(1 + n)
where, in the above derivation one uses the following: 1) Rt+1 = @F=@Kt+1 = FK(t+1) =
@f =@kt+1 = fk(t+1) , 2) L: =Lt+1 = 1=(1 + n), 3) At =At+1 = 1=(1 + g) and 4) wt = fl = F FK =
f fk by homogeneity of degree one (recall Taylors theorem):Therefore, the key equation for
the law of motion of capital in Diamonds model is:
1
(1 + n)kt+1 = ((1 + g) (f fk kt ); fk(t+1) ) (3.5)

at this point one needs to specify some functional form for utility and production since kt+1
shows up in both sides of the above equation and cannot be solved for explicitely.

3.1.1 Log-utility and Cobb-Douglas technology


Under u(cit ) = log cit and F (Kt ; At Lt ) = Kt (At Lt )1 ) f (k) = k one can re-write the
(generation t HH) problem as:

max U (c) = log c1;t + log c2;t+1


c1;t ;c2;t+1
s:t:
c2;t+1
At wt = c1;t + (3.6)
Rt+1
so the Euler eq. can be re-written as:
1 1
= Rt+1
c1;t c2;t+1
c2;t+1 = c1;t Rt+1

now using the resource constraint:


c1;t Rt+1
At wt = c1;t +
Rt+1
At wt
c1;t =
1+
1 1
= k k
1+
a policy function for consumption; replacing in the savings function:

At wt = c1;t + st
At wt
st = At wt
1+

st = At wt
1+
1 1
= k k
1+

27
that is, as in the growth models of the sections above, savings are a constant fraction of HH
income. Next, using the equation for capital:
st
(1 + n)kt+1 =
At+1
1
= At wt
At+1 1+

= wt
(1 + ) (1 + g)

= (kt kt )
(1 + ) (1 + g)
(1 )
kt+1 = k (3.7)
(1 + ) (1 + g)(1 + n) t

a policy function for capital accumulation. Equation (3.7) is the key equation of Diamonds
model unde log-utility and Cobb-Douglas production.

3.1.2 Steady state


The steady state value for capital can be computed as:

(1 )
k = k
(1 + ) (1 + g)(1 + n)
1
(1 ) 1
k = (3.8)
(1 + ) (1 + g)(1 + n)

steady state output, is:

y = k
(1 ) 1
=
(1 + ) (1 + g)(1 + n)

To obtain SS consumption rst note that cit is the amount consumed by one typical household
of generation i in its rst year of life. Since now the poppulation size is not normalized to 1 as
in the growth models of previous sections, in order to nd economy-wide consumption at t one
must compute:
Ct = Lt c1t + Lt 1 c2t
so that total consumption per (eective) worker in period t:
c2t
ct = c1t +
(1 + n)

now, using:

Ct = Yt St
ct = yt st
c = k (1 + n)k

so that: 1
(1 ) 1
(1 ) 1
c= (1 + n)
(1 + ) (1 + g)(1 + n) (1 + ) (1 + g)(1 + n)

28
3.1.3 Golden rule and dynamic ine ciency
As before, the golden rule for capital accumulation will follow:

kGold = arg max k (1 + n)k


k | {z }
=c

with F.O.C. as:


1
k = 1+n
1
1+n 1
kGold = (3.9)

or, in the general case:

f 0 (kGold ) = 1 + n
R = 1+n
r = n

that is, only when the interest rate equals the rate of poppulation growth, capital is at its golden
rule level.2
The possibility of dynamic ine ciency arises by comparing (3.8) with (3.9):

?
k = kGold
1 1
(1 ) 1
? 1+n 1
=
(1 + ) (1 + g)(1 + n)

therefore, if k 6= kGold the current allocation of resources would be Pareto ine cient and the
planner could make every one else by rising/lowering investment (savings)

3.1.4 The role of Government.


Suppose that for some reason the economy nds itself under dynamic ine ciency, i.e., k > kGold .
Then the Government could reallocate resources as follows: spend Gt and fund it entirely with
(lump-sum) taxes in that same period. Then key equation (3.7) becomes:

(1 )
kt+1 = [kt Gt ]
(1 + ) (1 + g)(1 + n)

which unambiguously leads to a lower k: The intuition is as follows: since the taxes are only on t,
then generation t is taxed only on its rst year of life. The HH would have to reduce consumption
in its rst period but since it wants to smooth consumption (under CRRA) it would have to
reduce its savings too, so that consumption in period t falls less than the full amount of taxes.
Consequently, savings fall and the economy moves towards kGold (which maximizes the SS level
of consumption). The Government in turn can reallocate the taxes as Gt to increase consumption
of those hurt and therefore increase economy-wide consumption.

Claim 8 In the OLG presented above, the Ricardian equivalence result does not hold in general.
2 Note that if 6= 1 then this condition becomes (1 + r ) = 1 + n or r = n:

29
To se why, suppose that in period t; Government purchases Gt are introduced and funded by
issuing one-period bonds bt = Gt . In turn, in period t + 1 the government levies a tax in order to
repay for the bonds, so t+1 = (1 + rt+1 )bt : In the OGM the RH would then save whatever the
government spends Gt , so that it yields reutrn (1 + rt+1 )Gt = t+1 . Therefore, the consumption
plan operates as if the government was funding its purchases via taxes in period t: Hence, the
irrelevance of distinguishing between taxes and bonds. On the other hand, in the OLG model,
at period t only the young agents care about future taxes; therefore, introducing Gt has real
eects on the consumption plan of generation t 1:

3.1.5 Social security


Consider the problem of social security in the OLG model. "Old" HH receive benets bt that
are funded by some sort of contriubution d: There are two systems; fully funded social security
and pay-as-you-go (PAYGO).
Fully funded social security. A HH of generation t contributes d in its rst period of
life, this contribution is invested at market rates and then in its second period of life the HH
receives (1 + Rt )d: The HH therefore faces the problem3 :

max u(c1t ) + u(c2t+1 ) (3.10)


c1t ;c2t+1

s:t:
c1;t + st + d At wt (3.11)
c2;t+1 Rt+1 (d + st ) (3.12)

Claim 9 Under the above assumptions and denitions, the HH problem with fully funded social
security is equivalent to the simple HH problem (3.2)-(3.3) of section 3.1

Proof. To see why, rst solve for st in the constraint (3.12):

c2;t+1 = Rt+1 (dt + st )


c2;t+1
st = d
Rt+1

and next replace in constraint (3.11):

c1;t + st + d = At wt
c2;t+1
c1;t + d +d = At wt
Rt+1
c2;t+1
c1;t + = At wt
Rt+1

which is exactly the resource constraint (3.3). Hence, the problem is identical to that presented
in section 3.1
The intuition for this result is simple: if contributions are invested at market rates, then
they are simply a dierent way of saving. Therefore, the HH only needs to choose how much
it wants to save at t, then pay d as contributions and invest the remaining as st : Note that,
depending on the size of d; it may be optimal for the HH to set st = 0: Also notice that now the
amount invested (which matches exactly the amounf of capital available at t + 1 since = 1) is
st + d = (1 + n)kt+1
3 In this set up it is assumed that each HH of all generations make the same contribution, i.e., the case where

dt 6= dt+1 is ruled out. This possibility is explored in Acemoglu (2008) and in fact the case where dt is a choice
variable is discussed.

30
Pay-as-you-go social security. In this case, a HH from generation t contributes d in its
rst period of life. This amount is transfered in the same period to generation t 1 (the olds
at t) in the form of benets bt . In its second period of life, each HH of generation t receives the
contributions of generation t + 1 HH: bt+1 = (1 + n)d: Therefore, the (generation t) HH problem
is:

max u(c1t ) + u(c2t+1 ) (3.13)


c1t ;c2t+1

s:t:
c1;t + st + d At wt (3.14)
c2;t+1 Rt+1 st + (1 + n)d (3.15)

so that the consolidated (ow) resource constraint becomes:

c2;t+1 (1 + n)
c1;t + d +d At wt
Rt+1 Rt+1

therefore, since the return on contributions is now (1 + n) rather than Rt+1 the problem is
equivalent to the simple model of 3.1 only if Rt+1 = (1 + n). Also notice that only st goes
into capital accumulation, rather than st + d as before. However, if the economy faces dynamic
ine ciency and overaccumulation of capital (i.e., Rt+1 < 1 + n) then PAYGO social security
results in capital accumulation that is lower by d, hence steering the economy again towards
the golden rule level of capital. Likewise, if d was not xed but a choice of the HH (or at least
a fraction of it), HH would prefer d to st as long as Rt+1 < 1 + n. This in turn would reduce
capital until rates of return again equalize.

3.1.6 Restoring Ricardian equivalence


Suppose the following set up. Instead of new families arriving, assume that each HH beguets
(1 + n) osprings which count as HH in their second period of life. Therefore, the demographic
structure of the models above remain unchanged. However, suppose now that each HH only
works in its second period of life. Moreover, HH consumption on its rst period of life is incor-
porated in its "parent" HH. Finally, the "parent" HH cares about its osprings consumption
ct+1 ; and therefore considers a bequest bt+1 . The problem of generation t 1 at t is therefore:

max u(ct ) + u(ct+1 )


ct ;bt+1

s:t: (3.16)
ct + bt+1 yt At wt + Rt bt

where, as before:
wt = f (kt ) f 0 (kt )kt
and:
Rt = f 0 (kt )
moreover, assuming = 1, capital available for production in period t + 1 equals generation ts
wealth in that period, which was bequested to him by its parent HH bt :

Kt+1 = Lt+1 bt+1


1
kt+1 = bt+1
At+1
At+1 kt+1 = bt+1

31
where, Kt+1 =Lt+1 is the capital per-worker. Note that the constraint in (3.16) could be re-
written as:
ct + bt+1 At (wt + Rt kt )
Therefore, a HH from generation t 2 loosens its osprings resource constraint by bequesting
bt . Under this set-up, in each time period, only one type of HH decides how much to consume
and how much to save as bequest for its ospring. In turn, what the parent HH leaves as bequest
is used for production in t + 1 which constraints consumption of its ospring in that period.
Therefore, the model is akin that of the OGM of chapter 2, more specically, to that presented
in example 3. The generation t HHs problem can be summarized by the following BFE:

V (bt ) = max fu(ct ) + V (bt+1 )g


ct ;bt+1

or:
V (bt ) = max fu(At (wt + Rt kt ) bt+1 ) + V (bt+1 )g
bt+1

with F.O.C.:

u0 (ct ) = V 0 (bt+1 )
u0 (ct ) = u0 (ct+1 )Rt+1

Therefore, all the results form chapter 2 apply, and, in particular, since the parent HH cares
about the osprings consumption, Ricardian equivalence is restored.

3.2 OLG in pure exchange economies (Samuelsons)


3.2.1 Homogeneity within generation
Assume again that there are agents from two generations alive at each period t. Mass for each
generation is normalized to 1, so poppulation size at each t is:

Lt + Lt 1 =1+1=2

Furthermormore, assume for now that there is no poppulation growth nor there is any produc-
tion. Instead, each agent is endowed with ! 1 when young and ! 2 when old. At the begining of
time (t = 0) generation 0 is already alive; it only lives one period and only consumes its endow-
ment. Assumptions 1)-4) in section 2.2 are satised. In an economy with borrowing/lending or
where endowments are either storable or denominated in money, agents could consume and save
at each period, thus facing the problem:

max u(c1;t ) + u(c2;t+1 )


c1t ;c2t+1 ;st

s:t:
c1;t + st ! 1
c2;t+1 ! 2 + Rt+1 st

or consolidating a (PV) lifetime budget constraint as in the previous section:

max u(c1;t ) + u(c2;t+1 )


c1t ;c2t+1

s:t:
c2;t+1 !2
c1;t + !1 +
Rt+1 Rt+1

32
the Lagrangean for this problem would be:

c2;t+1 !2
L = u(c1;t ) + u(c2;t+1 ) c1;t + !1
Rt+1 Rt+1

with associated familiar Euler equation:

u0 (c1;t )
= Rt+1
u0 (c2;t+1 )

Now, suppose that endowments are not storable and there exists no storable asset such as money
and just for simplicity let = 1. Under this set-up there would be no trade, borrowing/lending
or savings. To see why, rst notice that there would be no trade between generations. At any
given t, "olds" will never want to consume less but more so they would only be interested in
borrowing; on the other hand they would not be around in the next period to honor their debts
so "youngs" will not engage in any lending or temporary exchange in endowments with old
agents. Finally, there could not exist trade or borrowing/lending within generations since all
agents belonging to the same generation are identical. Therefore, an autarky equilibrium must
arise:

c1;t = !1
c2;t+1 = !2

that is, each agent must consume its entire endowment at every period. This is a trivial example
of a stationary equilibrium, i.e., that in which young agents from every generation have the
same level of consumption and old agents from every generation aslo have a constant level of
consumption. Naturally it is not necessarily the case that c1;t = c2;t+1 . Now, to support this
equilibrium, the autarky interest rate must satisfy:

u0 (c1;t ) u0 (! 1 )
RA = 0
= 0
u (c2;t+1 ) u (! 2 )

therefore by the concavity of u( ):

RA 1 ) u0 (! 1 ) > u0 (! 2 ) ) ! 2 > ! 1 (Classical case)


RA < 1 ) u0 (! 1 ) < u0 (! 2 ) ) ! 2 < ! 1 (Samuelsons case)

so that if ! 2 6= ! 1 marginal utility across consumption in dierent periods is not equalized.


That is, if RA > 1 ) ! 2 > ! 1 , agents would want to consume more in their rst period and
less in their second period i.e., borrow in t and repay in t + 1 (give up some ! 2 in exchange
for ! 1 ). In turn, if RA < 1 ) ! 2 < ! 1 agents would want to save (lend) in their rst period
and consume more in ther last period of life. Notice that only in the latter case there is room
for Pareto-improving reallocation (no "old" agent is willing to lend to "youngs"). In particular,
whenever RA < Rt+1 (equilibrium return under autarky is less than the free-trade equilibrium
return), there would be room for intervention.
Notice also that, under perfect consumption smoothing in u( ) (e.g. log-utility, CRRA) and
autarky, one has that:
RGold = 1
so that:
s1t = s2t = S(Rt+1 ) = 0

33
3.2.2 The role of money
Suppose now that the economy nds itself in a place where RA < Rt+1 . An agent in generation
f0g has endowment (! 2 ; M ) where M is the total supply of money, which is constant. In turn,
agent from generation t 1 faces the (free-trade) problem (with = 1):

max u(c1;t ) + u(c2;t+1 )


c1t ;c2t+1 ;st

s:t:
c1;t + st ! 1
c2;t+1 ! 2 + Rt+1 st

Now, the unit price of a nal good is Pt and the price of money in terms of goods (i.e. the
relative price of money) is qt = 1=Pt : Notice that as Pt ! 1, the value of money qt ! 0: Real
money balances are denoted mt = qt M = M=Pt : As usual, the return on money is the inverse
of ination, and since interest rates equalize:
qt+1 Pt
Rt+1 = =
qt Pt+1
and therefore:
mt+1
Rt+1 =
mt
Under this set up, in period 1 agents from generation f0g can exchange their endowment of
money for goods and generation f1g can save part of its endowment in the form of money. In
period 2 agents form generation f1g again exchange their saved money for goods with agents
from generation f2g and so on. The resource constraints are now:
M
c1;t + !1
Pt
M M
c2;t+1 ! 2 + Rt+1 ) c2;t+1 !2 +
Pt Pt+1

Therefore, savings function at any period is given by (assuming perfect foresight):

mt+1 M
St (Rt+1 ) = St = = qt M
mt Pt

or, using the fact that Rt+1 = qt+1 =qt one has:

qt+1
St = qt M
qt
which is a FODE that can be expressed as:

qt+1 = (qt )

whose unique non-trivial steady-state (assuming ( ) is monotonic) would be given by:


qt+1
qt = (qt ) = qt+1 = q ) = 1 = Rt+1
qt
that is, when savings satisfy the golden rule. Local stability for the SS is guaranteed only if
0
(q ) < 1: In turn, if 0 (q ) > 1; then whenever qt < q then qt ! 0 so that the value of money
would collapse and the economy would experience hyperination. Notice that this possibility
may arise even as M is xed, so its a "bubble" of sorts.

34
Example 10 Consider the problem for generation t agent:

max log(c1;t ) + log(c2;t+1 )


c1t ;c2t+1 ;st

s:t:
M
c1;t + !1
Pt
M
c2;t+1 !2 +
Pt+1
Pt
= Rt+1
Pt+1

the Euler equation being:


1 Rt+1
=
c1;t c2;t+1
c2;t+1 = c1;t Rt+1

using the resource constraints:


M
c1;t = !1
Pt
M
c2;t+1 = !2 +
Pt+1

replace in the Euler eq.:


M M
!2 + = !1 Rt+1
Pt+1 Pt
and using Rt+1 = Pt =Pt+1 and M=Pt+1 = (M=Pt ) (Pt =Pt+1 ), solve for M=Pt which is the
savings function (recall agents save only on their rst period, in this case, t):

M 1 !2
St (Rt+1 ) = = !1
Pt 2 Rt+1

Next, solve the di erence equation:

M 1 ! 2 Pt+1
= !1
Pt 2 Pt
!2 2M
Pt = Pt+1 +
!1 !1
notice that for this FODE to have a unique bounded solution (i.e., non-explosive/hyperination)
it is required that !
! 1 < 1: To solve forward rst:
2

!2 2M
Pt+1 = Pt+2 +
!1 !1
replace in the equation for Pt :

!2 !2 2M 2M
Pt = Pt+2 + +
!1 !1 !1 !1
2
!2 ! 2 2M 2M
= Pt+1 + +
!1 !1 !1 !1

35
so the solution to the di erence equation that determines the price level is:
T 1
X k
!2 !2 2M
Pt = lim Pt+T +
T !1 !1 !1 !1
k=0
1
X k
!2 2M !2
Pt = (assuming < 1)
!1 !1 !1
k=0
2M 2M ! 2
Pt = =
!1 1 !2 ! 1 ! 2 ! 21
!1

3.2.3 Fiscal policy and the Laer curve


Suppose now that the Government introduces purchases and fuds them either entirely by printing
money. Thus, money stock is not constant anymore:
Mt Mt 1
G= = qt (Mt Mt 1)
Pt
or:
Pt 1
G = mt mt 1
Pt
= mt mt 1 Rt mt 1 + mt 1
= mt mt 1 + mt 1 (1 + Rt )
| {z } | {z }
segnioreage inf lation tax

solving for the (gross) rate of return:

mt G
Rt =
mt 1

Now the savings function becomes:

mt G
S (Rt+1 ) = S = mt
mt 1

so that G simply shifts the savings function. However, this may induce more than one non-trivial
SS (if, e.g., the savings function intersects the ordinate above 0). Now, in any steady state where
mt = mt 1 = m:
G m G
S (R) = S 1 = m and R =
m m
so that:

G = m (1 R)
G
G = S 1 (1 R)
m

This is a classical Laer curve equation. Since G is in both sides of this equation, and S 0 ( ) > 0,
G will have opposite eects, hence the bell shape of the Laer curve.

36
3.2.4 Monetary equilibria with money growth
Recall the equilibrium in an economy without at money was:

u1 (! 1 ; ! 2 )
St (Rt+1 ) = 0 and RA =
u2 (! 1 ; ! 2 )

introducing a xed stock of at money gives rise to:


M Pt qt+1 mt+1
St (Rt+1 ) = = mt and Rt+1 = = =
Pt Pt+1 qt mt

so that together these two conditions imply:

mt+1
St = mt ) mt+1 = (mt )
mt

with steady state:


m
m =S = S (R ) = S (1) ) m = (m )
m
so it is required that S (1) > 0 for agents to be willing to hold money when R = 1. Naturally,
if mt = 0 there is no savings and the economy is back to autarky so (0; 0) is the trivial steady
state. Note, in the classical case, where RA 1 the slope of 0 ( ) > 1 and the only steady
state is the trivial one. Intuitively, in this case agents want to consume more than their current
endowment and in the future consume less than their endowment. Therefore, under this case
there is no room for intervention, and, not surprisingly, there is no monetary equilibria. In
Samuelsons case, RA < 1 and there exists at least one non-trivial steady state (i.e., in which
money is valued).

Denition 1 (Monetary equilibrium) Suppose an OLG economy in which the government


nances its decit by printing money as above. A monetary equilibrium consists of sequences
fMt ; Pt g with Pt < 1 and Mt > 0 8t such that:

M M
Mt = arg max u ! 1 + u !2 + (HH optimize)
M 0 Pt Pt+1

and:
Mt Mt 1 = Pt Gt (Gvts constraint is satised)

An alternative representation of this economy is to replace money with government bonds.


In this case, bonds will yield interest rate Rt+1 in period t + 1 which the government should
pay and hence will enter the gvts constraint. In the same spirit, a straightforward extension is
the case when there is poppulation growth. If each (generation zero) agent is endowed with Mt
units of monet, the governments budget constraint (second condition of equilibrium) is replaced
by:
Lt Mt Lt 1 Mt 1 = Lt Pt Gt
and a similar aproach can be taken for the government bonds case.

3.2.5 Within generation heterogeneity


Suppose now that within each generation, there are Lj of type j agents, with j = 1; :::; N:
Suppose for simplicity that utility is logarithmic. Now type j of generation t will face the

37
problem:

max log(cj1;t ) + log(cj2;t+1 )


s:t:
cj1;t + sjt (Rt+1 ) ! j1 (3.17)
cj2;t+1 ! j2 + sjt (Rt+1 ) Rt+1
Pt
= Rt+1
Pt+1

where sjt (Rt+1 ) is the savings function for any type j agent from generation t. Thus, total
savings from type j agents would be Lj sjt (Rt+1 ) = Stj (Rt+1 ).

Denition 2 An equilibrium without valued currency n (or non-monetary


o equilibrium) for the
j
economy summarized in (3.17) consists of sequences Rt ; st such that every agent optimizes
and markets clear, i.e.: !
j 1 j ! j2
st (Rt+1 ) = !1 +
2 Rt+1
and:
N
X
Lj sjt (Rt+1 ) = 0
j=1

Example 11 Suppose log-utility and two types of generation t agents j = 1; 2.with mass Lj
each. Endowments ! 1t ; ! 1t+1 = ( ; 0) and ! 2t ; ! 2t+1 = (0; ) with ; positive constants.
Then in period t; type 1 agents will be lenders and type 2 agents will be borrowers. In period 1
savings functions are:

St2 (Rt+1 ) = L2 s2t (Rt+1 ) = L2


2Rt+1

St1 (Rt+1 ) = L1 s1t (Rt+1 ) = L1


2
so that the interest rate is uniquely determined by Rt+1 = L2 = L1 . Thus there exists a unique
equilibrium without valued money. Naturally, for this exchange to take place, agent type 2 must
issue IOUs for exactly ( =2Rt+1 ) L2 which are repaid with interest in period t + 1: Note that
when L2 < L1 ) R < 1; Samuelsons case, which opens the door for at money.

Next, introduce money through generation zero old agents (as before). For simplicity assume
that, unlike generations t 1; these agents are identical and each of them is endowed with M
units of at currency. Then:

n 3 A monetary
Denition o equilibrium for the economy summarized in (3.17) consists of se-
j
quences Pt ; Rt ; st such that 8t:

0 < Pt < 1
Pt
Rt =
Pt+1
!
1 ! j2
sjt (Rt+1 ) = ! j1 +
2 Rt+1

38
and:
N
X M
Lj sjt (Rt+1 ) =
j=1
Pt

In this economy, a monetary steady state implies P = Pt = Pt+1 which in turn implies
Rt+1 = 1. Thus a monetary equilibrium requires (as was seen before) positive savings at an
interest rate of 1.

Example 12 Continuiung with the previous example, allow for money. Then savings functions
remain:

s2t (Rt+1 ) =
2Rt+1

s1t (Rt+1 ) =
2
but the equilibrium (market clearing) condition is now:

M
= +
2
|{z} Pt 2Rt+1
|{z} | {z }
dem and for assets m on ey supply IO U s supply

Naturally, it is required that agents are indiferent between holding money and holding IOUs (i.e.,
lending) which implies that:
Pt
Rt+1 =
| {z } Pt + 1
return on IO U s
| {z }
return on m oney
:

3.2.6 The real bills doctrine


Consider the following setup. In addition to issuing money, the government purchases private
IOUs in the amount Dt . Current purchases of IOUs are nanced by the repayment of previous
period IOUs and by issuing money:
Mt Mt 1
Dt = Dt 1 Rt +
Pt
so that the loan market clearing condition is now:
N
X Mt Mt M t 1
Lj sjt (Rt+1 ) + Dt = + (3.18)
Pt P
j=1 | {z t }
| {z } dissaving of "olds"+new m oney
private +public savings

however, this can be re-written as:


N
X Mt
Lj sjt (Rt+1 ) + Dt =
j=1
Pt

next, lagging the government budget constraint:


Mt 1 Mt 2
Dt 1 = Dt 2 Rt 1 +
Pt 1

39
and using Rt = Pt 1 =Pt ;replacing in constraint for Dt :

Mt Mt 2
Dt = Dt 2 Rt 1 Rt +
Pt
iterating back this procedure and assuming L0 = 0 one arrives at:

Mt M
Dt =
Pt

where M is the initial stock of money that would have prevailed if no OMO had been carried
out. Thus. (3.18) becomes:
XN
M
Lj sjt (Rt+1 ) =
j=1
Pt

which is the same expression found when there wese no IOUs purchases. Thus, the irrelevance
of open market operations.

40
Part II

Stochastic models

41
Chapter 4

Stochastic Optimal growth

4.1 Uncertainty in the neoclassical OGM


Consider the OGM of Chapter 2 (including all of its assumptions) under uncertainty steming
from the evolution of e.g. technology shocks. Let st denote the state at t as before. Suppose that
the state space is nite, i.e., s (t) 2 fs1 ; s2 ; :::; sN g, and that Pr (s (t) = si ) is the unconditional
probability that the state is si in period t: Let st = (s (1) ; s (2) ; :::; s (t)) be the history of states
up to and including t: Then the problem faced by the planner can be represented as:
( 1 1 X
)
X X
t t t t
max E0 u(ct ) = u(ct s ) Pr(s ) (4.1)
ct ;kt+1
t=0 t=0 st
s:t:
ct + kt+1 yt + (1 )kt (4.2)
yt = At f (kt ) (4.3)
At = At 1 e"t "t iid(0; 1) (4.4)
k0 given

with:

state at t : st = (kt ; At )
control at t : (ct ; kt+1 )

and:

history of realizations at t : st = (s0 ; s1 ; :::st )


uncond. prob of observing st as of 0 : Pr(st ) = Pr(s0 ; s1 ; :::st )

then the BFE for this problem is now:

V (kt ; At ) = max fu(ct ) + Et V (kt+1 ; At+1 )g


ct

s:t:
ct + kt+1 At f (kt ) + (1 )kt

where: X
Et V (kt+1 ; At+1 ) = V (kt+1 ; At+1 ) Pr(st+1 jst )
st+1 jst

42
and Pr(st+1 jst ) is the probability of st+1 conditional on having observed st : Assuming At has
the Markov property:

V (kt ; At ) = max fu(ct ) + E [V (kt+1 ; At+1 )jAt ]g


ct

since the only source of uncertainty stems from At : The solution to this problem will yield again
time-invariant (i.e. stationary) state-dependent policy functions:

ct = h(At ; kt )
kt+1 = (At ; kt )

As before, the F.O.C:


u0 (ct ) = Et V 0 (kt+1 ; At+1 )
and the associated envelope condition:
@L
Vk (kt ; At ) = [At f 0 (kt ) + (1 )] =
@kt
= u0 (ct ) [At f 0 (kt ) + (1 )]

or:
Et Vk (kt+1 ; At+1 ) = Et fu0 (ct+1 ) [f 0 (kt+1 ) + (1 )]g
thus, the Euler equation and accumulation constraint:

u0 (ct ) = Et fu0 (ct+1 ) [f 0 (kt+1 ) + (1 )]g (4.5)


ct + kt+1 = At f (kt ) + (1 )kt (4.6)

are the pair of rst order stochastic dierence equation that characterize the solution to the
1
social planners problem. That is, a sequence fct ; kt+1 gt=0 that satises equations (4.5)-(4.6)
and the transversality condition:
t 0
lim E0 u (ct )kt+1 = 0
t!1

will solve the social planners problem.

4.1.1 Non-stochastic steady state


The non-stochastic SS for the OGM is found by setting "t = 0 8 t and At = At 1 = A =1
which results in precisely the system derived in section (2.5).

4.1.2 Stationary distribution


First note that under the appropriate conditions, At will inherit some well-behaved stationary
distribution.from "t : In fact, if "t iid N (0; 1) and < 1 then:
2
log At = at N (0; )

Moreover, if At 2 A; A , then one can obtain a limiting distribution for kt in which:

(kt ; A) kt+1 kt ; A

That is, if the non-stochastic steady state level for capital is locally asymptotically stable, there
will exist such starionary limiting distribution under the presence of small shocks.

43
4.1.3 Log-linear approximation
Approximating the dynamic system (4.5)-(4.6) along with the low of motion of technology yields:

c^t = Et [ c^t+1 at+1 kt+1 ] (4.7)


c k
c^t + k^t+1 = at + k^t (4.8)
y y
at+1 = at + "t+1 (4.9)

where:
cu00 (c) kf 00 (k) kf 0 (k)
= = and =
u0 (c) f 0 (k) f (k)

Remark 4 Notice that because A = 1 ) log A = 0 then a


^t = at

4.2 Solution method 1: Blanchard-Khan


An "old-school" avenue for studying and solving the system (4.7)-(4.9) is to follow Blanchard
and Khan (1980):
Et c^t+1 0 c^t
= c + a
0 k
k^t+1 y y k^t 1 t
or:
c +
Et c^t+1 1 y c^t
= + y at
k^t+1 y
k
c
k
k^t k

Xt+1 = M X t + N at (4.10)

in this case there are two endogenous variables (c; k) and one exogenous a. Of the two endogenous
variables, one is pre-determined (k) and one is forward-looking (c) : Following Blanchard-Khan,
the existence of a unique bounded solution f^ ct ; k^t g1
t=0 to the system above requires that the
number of eigenvalues of M outside the unit circle equal the number of forward-looking variables.
In this case one requires j 1 j < 1 and j 2 j > 1 where 1 ; 2 are the eigenvalues of M:

Remark 5 If the number of eigenvalues lying outside the unit circle is greater than the number
of forward-looking variables, then the system has no bounded solution. On the other hand, if the
number of eigenvalues lying outside the unit circle is less than the number of forward-looking
variables the system may have many bounded solutions.

Remark 6 In larger systems the B-K cannonical representation can be expressed as:

Yt forward-looking endogenous variables


Xt = =
Zt predetermined edendogenous variables

The solution to the system (4.10) above again yields:

k^t k^t
c^t = st = = 1 2 (4.11)
at at
k^t k^t
kt+1 = st = = 1 2 (4.12)
at at

44
for some matrices and . Furthermore note that:

k^t k^t 1 0
s^t = = 1 2
+ "
at 0 at 1 1 t
s^t = P st 1 + Q"t

so one could compute:

0 st s^0t
= E^
0
= E (P s^t 1 + Q"t ) (P s^t 1 + Q"t )
0 2
0 = P 0P +Q Q0

where 2 = V ["t ] = E ["t "0t ]. That is, one can solve this discrete Lyapunov equation to obtain
the contemporaneous variance-covariance matrix of the state vector. From here it is possible to
obtain any two-period var-cov matrix:

j st s^0t+j = P j
= E^ 0

likewise, one can obtain:


ct s^0t =
E^ 0

and:
ct c^0t =
E^ 0
0

4.3 Impulse response functions (IRF)


As an alternative way to investigate how the system adjusts to distrubances, one can generate a
once-and-for-all unit-shock to technology at perior t and study how it reverberates in the system
thereafter:
k 0
st = t =
at 1
and compute:

Et st+1 = P s0
Et st+2 = P 2 s0
.. .
. = ..
Et st+1 = P j s0

From here it is straightforward to generate sequences for f^ct ; k^t+1 gTt=0 by the use of the policy
functions. Moreover, one could generate a series of histories of the variables in the system and
compute "empirical" variances and covariances that can be compared to those found analitically
in the previous section.

45
Chapter 5

RBC models

5.1 The baseline RBC model


5.1.1 The general case
To begin, simply extend the stochastic OGM (with all the assumptions made earlier) from the
previous chapter and relax the assumption that households inelastically supply all the labor they
are endowed with. Instead, explicitely introduce the disutility of labor in the objective function
therefore making labor a choice variable. For simplicity ignore state-dependency notation.
1
X
t
max E0 U (Ct ; 1 Nt ) (5.1)
Ct ;Kt+1 ;Nt
t=0
s:t:
Ct + Kt+1 Yt + (1 )Kt (5.2)
Yt = At F (Kt ; Nt ) (5.3)
At = At 1 e"t "t iid(0; 1) (5.4)
K0 given

the F.O.C. for this problem now include the optimal intra-temporal allocation of labor-consumption
as well as the inter-temporal allocation of consumption-investment. The Lagrangian for this
problem is
1
X
t
L = E0 fU (Ct ; 1 Nt ) t [Ct + Kt+1 At F (Kt ; Nt ) (1 )Kt ]g
t=0

so that:
UC (Ct ; 1 Nt )
Et = Et fAt+1 FK (Kt+1 ; Nt+1 )g (Euler equation)
UC (Ct+1 ; 1 Nt+1 )
UC (Ct ; 1 Nt )
= At FN (Kt ; Nt ) (Labor supply)
UN (Ct ; 1 Nt )
Ct + Kt+1 = At F (Kt ; Nt ) + (1 )Kt (Constraint)

along with the TVC:


t
lim E0 [UC (Ct ; 1 Nt )Kt+1 ] = 0 (TVC)
t!1

is the dynamical system that characterizes the solution to the social planners problem.

46
5.1.2 CRRA utility and Cobb-Douglas production
Consider the following explicit choice of functional forms:
1
X
t Ct1 N 1+'
max E0 (5.5)
ct ;kt+1
t=0
1 1+'
s:t:
Ct + Kt+1 Yt + (1 )Kt (5.6)
Yt = At Kt Nt1 (5.7)
At = At 1 e"t ; "t iid(0; 2
") (5.8)
k0 given

then the equilibrium conditions for this problem are:


( ) ( )
1
Ct Kt+1
Et = Et At+1 (EE)
Ct+1 Nt+1
Ct Kt
= (1 ) At (LS)
Nt' Nt
Ct + Kt+1 = At Kt Nt1 + (1 )Kt (RC)

the non-stochastic SS for this system is:

A = 1
1
K 1 1
= =
N
Y K
= =
N N
C
=
N

5.1.3 The log-linear system


Next, log-linearize the system around the non-stochastic SS:

1. Euler equation:
( 1 1 1
)
K K K
c^t + ct+1 = Et ( 1) k^t+1 ( 1) n
^ t+1 + at+1
N N N

but since: 1 1
K 1 1
K
= ) =
N N
replace in the log-linear Euler:
n o
c^t + Et ct+1 = Et ( 1) k^t+1 + (1 )n
^ t+1 + at+1 (5.9)

2. To log-linearize the labor supply one can proceed in two ways. The approach that is most
useful for solving the model computationally would yield:

C c^t = (' )( 1) N ' AK n


^t + ( 1) N ' AK k^t

47
but note, in steady state:

K
C = (1 )A N' W
N

where W is the SS real wage. Thus, the log-linear equation for labor supply becomes:

c^t = (' ^ t + k^t


)n

An alternative derivation is:


'
C N c^t + 'C N ' n ^t = w
^t
W c^ + 'W n
^t = w
^t

which lends itself to a nice interpretation of the cyclical character of the real wage.
3. Resource constraint:

ct + K k^t+1 = Y a
C^ ^t + K N1 + (1 ) K k^t + (1 ) K N1 n
^t
C K K
c^t + k^t+1 = + (1 ) k^t + (1 )n
^t + a
^t (5.10)
Y Y Y
or, equivalently:

(1 #) c^t + #k^t+1 = ( + (1 ) #) k^t + (1 )n


^t + a
^t

where:
1
1 K
#= =
N

4. Technology process:
at+1 = at + "t+1

5.2 Labor productivity (King, Plosser & Rebelo, 1988)

Suppose that labor productivity is trended. There are two cases to consider. In the rst case,
labor productivity follows a deterministic trend and technology shocks aect overall production.
A typical household then faces the problem:
1
X h i
max E0 t
log C~t + log (1 Nt )
~t ;Nt
C
t=0
s:t:
Ct + I~t = Y~t
~
Y~t = At K~ t (Nt Xt )1
I~t = K~ t+1 (1 )K~t
"t
At = At 1e ; "t iid(0; 1)

where all variables are dened as before, and labor productivity Xt follows:
Xt
=1+
Xt 1

48
~ K;
The rst thing to note from the above formulation, is that in any balanced growth path C; ~ Y~ ; I~
will be trended (and as will be the real wage). Therefore, we rst proceed to de-trend these
variables, by:
Z~t ~ K;
~ Y~ ; I~
Zt = for Z = C;
Xt
The only visible changes are in the production function equation:

Yt = At Kt Nt 1
~ t+1 =Xt+1 we require:
and in the capital accumulation equation; since Kt+1 = K

I~t ~ t+1
K ~t
K
= (1 )
Xt Xt Xt
Xt+1
It = Kt+1 (1 ) Kt
Xt
= Kt+1 (1 + ) (1 ) Kt

From this point the analysis is just as before. The social planners problem is to choose
Ct ; Nt ; Kt+1 taken as given the state of capital (Kt ) and the technology (At ): The equilibrium
conditions can be found to be:
" #
At+1 Kt+11 Nt+1 1 1+
Et = (5.11)
Ct+1 Ct
1 Ct
(1 ) At Kt Nt = (5.12)
Nt
At Kt Nt 1 + (1 ) Kt = Ct + (1 + )Kt+1 (5.13)

as required, there are three equilibrium conditions and three variables whose path are endoge-
nously determined.
Suppose instead that the level of technology is static (A is a constant), but labor productivity
is subject to shocks:
Xt
= (1 + ) (1 + t )
Xt 1
where t is an iid productivity shock with zero mean stationary process. One can rst derive
the equilibrium conditions for the original system:
2 ! 1 3
K~ t+1 1 Et C~t+1
Et 4 A Xt+1 5 =
Nt+1 C~t
!
K~t ~
t Ct
(1 )A Xt1 =
Nt Nt
~ t Nt 1
AK + (1 ~t
)K = C~t + K
~ t+1

and de-trend these conditions. First, multimply both sides of Euler equation by Xt =Xt+1 :
2 ! 13
X K~ t+1 ~
Et 4
t
A 5 = Et Ct+1 Xt
Xt+1 Xt+1 Nt+1 C~t Et Xt+1

49
note that by assumption Et [Xt+1 ] = (1 + ) Xt :
2 ! 3
1
1 K~ t+1 Et Ct+1
Et 4 A 5 =
1+ Xt+1 Nt+1 Ct
" #
1
Kt+1 (1 + ) Et Ct+1
Et At+1 =
Nt+1 Ct

next, divide both sides of the labor suply equation by Xt :


!
~t
K Xt1 ~
t Ct 1
(1 )A =
Nt Xt Nt Xt
Kt+1 t Ct
(1 ) At+11 =
Nt+1 Nt
and nally dividing both sides of the resource constraint by Xt :

At Kt Nt 1 + (1 ) Kt = Ct + (1 + )Kt+1

Notice that the assumptions on the stochastic process t lead to the same equilibrium conditions
as those found for the case of deterministic trended labor productivity.

5.3 Solution method 2: Sims GENSYS


In tis section, the log-linear model presented in section 5.1.3 is solved using Sims GENSYS
procedure. First, re-write the system as:

ct (1 )n
^t at = c^t 1 c;t +( 1) k^t + (1 ) n;t + a;t

c^t (' ^ t = k^t


)n
C K K
c^t + k^t+1 (1 )n
^t a
^t = + (1 ) k^t
Y Y Y
at = at 1 + "t+1
where x;t = x ^t Et 1x
^t . Now, this solution method requires the problem to be cast in the
following form:

0 xt = 1 xt 1 + zt + t (5.14)
h i0
^ t k^t+1 at , t = [ c;t
In this particular case xt = c^t n n;t 0 a;t ]
0
and zt = ["t ]:Therefore,
the input matrices for the numerical solution are:
2 3 2 3
(1 ) 0 1 0 ( 1) 0
6 (' ) 0 0 7 60 0 07
6 7 6 7
0 := 4 C 5 1 := 6 7
Y
(1 ) K
Y
1 40 0 + (1 )K
Y
05
1 0 0 0 0 0 0
2 3 2 3
0 (1 ) 0 1
607 6 0 0 0 07
=6 7
405 := 6
4 0
7
0 0 05
1 0 0 0 0

50
with this system at hand, one can proceed to implement Sims routine GENSYS (in Matlab)
as described in Sims (2002), and then obtain a solution for the model in the form:
1
X
xt = xt 1 + zt + AM s BEt zt+s+1 (5.15)
s 0
h i
^ t k^t+1 at as functions of the states.
which gives, as usual, a system of policy rules for c^t n
Again, one can proceed to either obtain moments analytically using the output matrices or
generate series to compute "empirical" moments.

5.4 Varieties of RBC models


Habbit formation
Capital adjustment costs
Variable K utilization

Campbells mechanism
Hansens indivisible labor

5.4.1 Asset pricing models (Lucas, Shiller)


Single asset enviroment
A simplied version of Lucas (1978) tree. Suppose that there is no production and agents nance
consumption with an exogenous stochastic endowment ! t , capital gains from selling shares of
a single traded asset carried from previous period Pt ( t t 1 ) and exogenous stochastic
dividends from ownership of shares dt t 1 . The planners problem is therefore:
1
X
t
max E0 U (Ct )
t=0
s:t:
Ct + Pt ( t t 1) dt t 1 + !t

The household choice variables at t are fCt ; tg while the states are f t 1 ; dt ; ! t g : The F.O.C
for this problem are:

U 0 (Ct+1 )
Pt = Et (dt+1 + Pt+1 )
U 0 (Ct )
Ct + Pt ( t t 1) = dt t 1 + !t

so that for given sequences of f t 1 ; dt ; ! t g ; a feasible allocation fCt ; t g and price system fPt g
that satises the FOC and TVC for the HH and clear markets will solve the planners problem.

51
Multi-asset enviroment.
Consider a simple extension of the above model in which households can hold shares from k
dierent risky assets and a riskless asset, B. The planners problem now becomes:
1
X
t
max E0 U (Ct )
t=0
s:t:
k
X k
X
Ct + Pjt ( jt jt 1 ) + Bt djt jt 1 + (1 + rt ) Bt 1 + !t
j=1 j=1

now the FOC for this problem are:


t
[Ct ] : U 0 (Ct ) = t
[Bt ] : t = Et t+1 (1 + rt+1 )

and k (one for each jth asset) FOC of the form:

t Pjt = Et [ t+1 (Pjt+1 djt+1 )]

hence the k + 1 Euler equations are:

U 0 (Ct ) = Et [U 0 (Ct+1 )] (1 + rt+1 )


U 0 (Ct+1 ) (Pjt+1 djt+1 )
Pjt = Et for j = 1; :::k
U 0 (Ct )

5.5 Calibration
A calibration exercise is best understood in an example. Consider the RBC model of section
5.1.2. The parameters to be calibrated in this model are:

: = capitals share of output


: = subjective discount factor
: = RRA coe cient
: = technology serial correlation
2
" : = variance of "t

5.6 Estimation methods


5.6.1 Generalized method of moments (GMM)
The theory of RCB gives rise to equations like (5.11)-(5.13) with stochastic content, which can
be expressed as:
E [f (Zt ; 0 )] = 0 (5.16)
where Zt = [Ct Kt Nt At ] and 0 = [ o o o ] is a vector of true parameters. For instance,
if the true parameters were known, the Euler equation in (5.11):
" #
At+1 Kt+11 Nt+1 1 1+
Et =
Ct+1 Ct

52
can be expressed as (5.16) by:
( " # )
1
o At+1 Kt+1
o
Nt+1 1 o
1+ o
E =0
Ct+1 Ct

where E [ ] is the unconditional expectation operator. Likewise, denoting 2Z ( o ) the variance


of variable Z = fC; N; K; Ag, which is naturally a function of o , a series of (second) moment
conditions can be obtained as:

E Ct2 2
C ( o) = 0
E Nt2 2
N ( o) = 0
E Kt2 2
K ( o) = 0
E A2t 2
A ( o) = 0

to obtain the sample analogs of (5.16) suppose that a sample of T observation is at hand so that
one has Z = (Z1 ; Z2 ; :::ZT ) for Z = fC; N; K; Ag. Then since the sample mean of Z converges
in probability to its expectation, the analogue of (5.16) would be:
T
1X
g (Z; ) = f (Zt ; )
T t=1

since E [f (Zt ; 0 )] = 0, the objective becomes to choose so as to minimize the expression above.
Now, assuming for simplicity that = 1, in this particular problem there are exactly as many
parameters as there are moment conditions. Thus using sample averages in this case looks a lot
like a calibration exercise with the only advantage that in this case one obtains standard errors
for the parameters as well. In general, however, this exact identication will not occur. More
often, there would be more moment conditions than parameters to estimate and therefore the
problem becomes:
0
min g (Z; ) g (Z; ) (5.17)

where g (Z; ) is a vector of sample conditions.and is a weighting matrix that, ideally, woudl
give less weight to those moments with higher variance compared to the true parameters. How-
ever, since these true values are unknown, one can proceed recursively:

1. Guess some starting value for , like e.g., =I


2. Estimate ^ using () above.
3. With the estimated variances, choose optimal weights and construct a new . Repeat
from (2)

Hansen (1982) pointed out that in this enviroment central limit theorems apply and therefore:
p d 1
T ^GM M o !W N 0; [D D0 ]

so that for a given parameter vector, this procedure will yield variances for each component
of Z:.It is important to point out that, as in other empirical procedures and solution methods
presented before, it is required that the data for the components of Z is stationary. This can
be achieved by obtaining the "detrended" moment conditions as was done in section 5.2 (in
which case or + must be estimated) and then using detrended data (e.g. applying a
Hodrick-Prescott lter or any other cycle-trend lter).
With the resutls from this estimation procedure hypothesis testing can be carried out as
in any estimation exercise. For instance, t-statistics can be used to evaluate the statistical

53
signi cance of the dierence between a hypothesized value of a given parameter and its sample
estimate counterpart. Likeweise restrictions such as the mapping from one parameter to another
can be tested using chi-square statistics. As an example, note that equation (5.11) implies that:

K
(1 + ) =
Y
a familiar condition. Finally, one can test the stability of a given parameter over the sample
analyzed. .
Full information maximum likelihood...

54
Bibliography

D. Acemoglu. Introduction to Modern Economic Growth. Princeton University Press, 2008.


C. Dave and D. De Jong. Structural Macroeconometrics. Princeton University Press, 2011.
A. Dixit. Optimization in Economic Theory. Oxford University Press, 1990.

A. Levy. Economic Dynamics: Applications of Di erence Equations, Di erential Equations and


Optimal Control. Ashgate Publishing, 1992.
L. Ljungqvist and T. Sargent. Recursive Macroeconomic Theory. MIT Press, 2004.
D. Romer. Advanced Macroeconomics. McGraw-Hill, 2001.

N. Stokey and R. Lucas. Recursive Methods in Economic Dynamics. Harvard University Press,
1989.

55
Part III

Appendixes

56
Appendix A

The dynamic programming


method

This appendix presents the various approaches to solving for the value function in a stationary
dynamic programming problem along with an explicit derivation of the envelope condition.
A typical dynamic programming problem includes:

A state space S Rm
A control space A Rn
Instantaneous objective function: u : A S!R
State transition function: g : A S!S
Feasible control correspondence: :S A
Discount factor:
Initial state: K0 2 S

Example 13 Suppose that S = A = R: Then the simplest "Cake-Eating" version of the Ramsey
problem is therefore:
1
X p
t
max Ct (A.1)
t=0
s:t:
g(Kt ; Ct ) = Kt Ct
Ct 2 [0; Kt ]
K0 =

so the Bellman Functional Equation for this problem is:


hp i
V ( ) = max C + V (K C)
C2[0; ]

57
A.1 Guess and verify
One method to solve the BFE is to try to guess the value function V ( ). In order to come up
with a guess, solve the problem rst as if it was a sigle-period maximization one:
p
max C
s:t: C 2 [0; ]

whose solution is: p


V0 ( ) =
Next, guess that, for the multi-period problem, the solution to the BFE looks something like:
p
V( )=a +b

where a; b are undetermined coe cients. Now solve for a; b using the BFE:
p h p i
V( C) = C + a C +b

so that: p np h p io
V( )=a + b = max C+ a C +b (A.2)
C2[0; ]

at this point one can actually maximize the RHS since under the assumed value function. To
do so, dierentiate w.r.t. C and set equal to zero. Since the term inside the braces is a concave
function, this would be in fact a maximizer. The result of this maximization is:

C= 2
1 + (a )

replacing this maximizer in (A.2) one obtains:


p q p
2
a + b = 1 + (a ) + b

so that the undetermined coe cients are:


q
2
a = 1 + (a )
1
a = p 2
1
and:

b = b
b = 0

so that nally one is ready to verify the initial guess. To do so, one must show that:
1 p hp i
p = max C+ V ( C) = V ( )
2 C2[0; ]
1
" p !#
p C
= max C+ p =V( )
C2[0; ] 2
1

58
2
so, maximizing the RHS again, one obtains Cmax = 1 . Plugging it in the equation
above one obtains:
p s
q 1 2
2
p = 1 + 2 =V( )
2 1
1
p p
p = p =V( )
2 2
1 1
so that the guessed solution the BFE is veried. Now to actually generate an optimal sequence
1
fKt+1 ; Ct gt=0 one uses the system of dierence equations:

Ct = h(Kt )
2
= 1 Kt
Kt+1 = Kt Ct
2
= Kt
K0 =
2 2
which yields Ct = (1 )Kt = (1 ) 2t and Kt+1 = 2t : Finally, note that the transver-
sality condition is satised. To see this recall that the value function is:
p
Kt+1
V (Kt+1 ) = p 2
1
so that:
1
V 0 (Kt+1 ) = p 2
p
2 1 Kt+1
and the transversality condition is:
t+1 p
t+1 0 t+1 Kt+1
lim V (Kt+1 )Kt+1 = lim p p = lim p =0
t!1 t!1 2 t!1 2
2 1 Kt+1 2 1

A.2 Value function iteration


Also known as method of successive approximations from the xed point theory. If the following
conditions are satised:

- The objective function u() is continuous and concave.


- The transition function g() is continuous.
- The feasible control correspondence :S A is non-?, compact, UHC, LHC.

- W 2 Cb (S) where Cb (S) is the set of continuous, bounded, real-valued functions on S

then, if T : Cb (S) ! S is an operator dened as:

T (W ( )) = max fu( ; C) + W [g( ; C)]g


C2 ( )

then T : Cb (S) ! S is a contraction mapping and has a unique xed point T (W ) = W .


Moreover, starting from any initial value, W n ( ) ! W ( ) whenever W n ( ) = W n 1 ( ) and
W = V which is the solution to the BFE.

59
Suppose that S = A = [0; K] for some K > 0 and that the problem is the same cake-eating
as the example above. Then the method of successive approximation requires to generate a
1
sequence of functions fV n gn=0 so that V n ! V , the solution to the BFE. As a starting point
for the sequence, let V 0 = 0 :
n = 0)V0 =0 hp i p
n = 1 ) V 1 ( ) = max C+ V0 =0 =
C2[0; ]
hp i q p
n = 2 ) V 2 ( ) = max C + V 1( C) = 1 + 2
C2[0; ]
hp i q p
n = 3 ) V 3 ( ) = max C + V 2( C) = 1 + 2 2
C2[0; ]
.. .. .. ..
. . . s .
P
1
2t
p
n = n ) V n( ) = =
n=1
p
n ! 1 ) V n( ) ! p 2
1

A.3 Solving for the policy functions


This is very similar to guessing the value function. Recall the F.O.C. from the problem in (2.4)
and assuming for simplicity full depreciation:
u0 (ct ) = u0 (ct+1 )f 0 (kt+1 )
ct + kt+1 = f (kt )
if one assumes Cobb-Douglas technology and log-utility the Euler equation becomes:
1 1 1
= kt+1
ct ct
at this point one can make a guess about the policy function (instead of guessing the value
function). Suppose that, as in the Solow model, it is optimal to set savings and therefore
consumption, as a constant share of output, that is, guess:
ct = (kt )
and using the resource constraint:
kt+1 = (1 ) (kt )
for some undetermined coe cient : Then replacing in the Euler equation and solving for the
undetermined coe cient yields:
1 1 1
= ((1 ) (kt ) )
(kt ) ((1 ) (kt ) )
1
=
(kt ) (1 ) (kt )
1 =
so that the "true" policy functions are:
kt+1 = (kt )
ct = (1 ) (kt )

60
which are of course, the same policy functions found in Example 2 of section 2.3 in the main
text.

A.4 Properties of the BFE


1. Claim 14 The mapping:

(T ) (x) = max fu (x; a) + E [f (x; a; ")]g (A.3)


a2 (x)

is a contraction.
Proof (Blackwell conditions). First, to prove monotonicity: Let (x) ' (x) for
every x, then for every f ( ) is continuous and (x) closed-graph:

[f (x; a; ")] ' [f (x; ; ")]


max fu (x; a) + [f (x; a; ")]g max fu (x; a) + ' [f (x; a; ")]g
a2 (x) a2 (x)

moreover, since the operator E ( ) preserves inequalities:

max fu (x; a) + E [f (x; a; ")]g max fu (x; a) + E' [f (x; a; ")]g


a2 (x) a2 (x)

(T ) (x) (T ') (x)

so that T is a monotone operator. Next, to prove discounting: Let c 2 R, and note:

[T ( + c)] (x) = max fu (x; a) + E [ (f (x; a; ")) + c]g


a2 (x)

= max fu (x; a) + E f (x; a; ") + Ecg


a2 (x)

= max fu (x; a) + E f (x; a; ")g + c


a2 (x)

= (T ) (x) + c
(T ) (x) + c whenever 2 [0; 1]

so that T ( ) satises discounting. Therefore, T ( ) satises both of Blackwell conditions


and thus is a contraction.

Claim 15 The operator T : D ! D in (A.3) maps the space of continuous, bounded,


real-valued, increasing and concave functions into itself.

Proof. (step 0) Assumptions:

- For x 2 X, X is a convex subset of Rn


- ( ) is non-?-valued; convex-valued, compact-valued, U.H.C. and L.H.C.
00
- For x0 > x00 ) (x) (x0 )
- u ( ) is continuous, bounded, real-valued, increasing and concave on both arguments
- f ( ) is continuous, bounded in all its arguments
- ( ) is bounded and continuous.
- f (:; :; ") is an increasing mapping in x and concave in x; a.
- 2 (0; 1)

61
(step 1: continuity and boundedness) First, note that f ( ) ; u ( ) bounded ( ) ) T
bounded. Next, continuity of f ( ) ; ( ) ; u ( ), linearity of E ( ) and the assumptions
on ( ) ) T continuous by Berges theorem (or the Continuous Maximum Theo-
rem).
(step 2: increasingness) Let ( ) be an increasing function. Next, suppose that x0 >
x00 .and that a0 = arg max0 ( ) and a00 = arg max00 ( ) :Then:
a2 (x ) a2 (x )

(x00 ) (x0 ) and f ( ) ; ( ) ; u ( ) increasing ) max0 ( ) max ( )


a2 (x ) a2 (x00 )

or, using the maximizers a0 and a00 :

(T ) (x00 ) = max (f ( ) ; ( ) ; u ( ))
a2 (x00 )

= u (x00 ; a00 ) + E [f (x00 ; a00 ; ")]


u (x0 ; a0 ) + E [f (x0 ; a0 ; ")]
= max0 ( )
a2 (x )

= (T ) (x0 )

so that:
x0 > x00 ) (T ) (x0 ) (T ) (x00 )
so that T ( ) is an increasing operator.
(step 3: concavity) Now, if ( ) is concave, the concavity of f ( ) ; u ( ) imply that the
objective function:
u (x; a) + E [f (x; a; ")]
is concave. Furthermore, since ( ) is convex-valued, we can invoke the Concave
Maximum Theorem1 and conclude that T is concave.

Claim 16 The operator T : D1 ! D1 in (A.3) maps the space of continuous, bounded,


real-valued, strictly increasing and strictly concave functions into itself.

Proof. Because the spaces of strictly increasing and of strictly concave functions are not
complete, this proof requires a 2-step procedure. TBC

A.5 The Envelope Theorem: an application


Heres an application of the Envelope theorem to a more general model in which there are several
choice and state variables. Consider the RH problem:
1
X
max u (ct ; 1 nt ; m t )
t=0
s:t:
mt 1
yt = ct + It + mt + bt = zt f (kt ; nt ) + + Rt bt 1
1+ t
kt+1 = (1 )kt + It
k0 given
1 See Carter, M., 2001, Foundations of Mathematical Economics, MIT Press, Theorem 3.1 pp. 343.

62
where, as in a shopping time or MIU model, only cash provides direct utility. In this case, zt is
a technology process with (for simplicity) deterministic constant growth:

zt = (1 + ) zt 1

and let at be the stock of nancial assets at time t:


mt 1
at = + R t bt 1
1+ t

The state variables for this problem are fkt ; bt 1 ; mt 1 ; zt g or simply fkt ; at ; zt g while the choice
variables are fct ; nt ; bt ; mt ; kt+1 g. As usual, the BFE can be written as:

V (kt ; at ; zt ) = max fu (ct ; 1 nt ; mt ) + V (kt+1 ; at+1 ; zt+1 )g


ct ;nt ;bt ;mt

s:t:
mt 1
ct + kt+1 (1 )kt + mt + bt = zt f (kt ; nt ) + + R t bt 1
1+ t

but bear in mind that:


mt 1 mt
V (kt+1 ; at+1 ; zt+1 ) = V zt f (kt ; nt ) + + R t bt 1 + (1 )kt mt bt ct ; + Rt+1 bt ; zt+1
1+ t 1 + t+1

The F.O.C.s are:

[ct ] : uc (ct ; 1 nt ; m t ) Vk (kt+1 ; at+1 ; zt+1 ) = 0


[nt ] : un (ct ; 1 nt ; mt ) + Vk (kt+1 ; at+1 ; zt+1 ) zt fn (kt ; nt ) = 0
1
[mt ] : um (ct ; 1 nt ; m t ) Vk (kt+1 ; at+1 ; zt+1 ) + Va (kt+1 ; at+1 ; zt+1 ) =0
1+ t+1
[bt ] : Vk (kt+1 ; at+1 ; zt+1 ) + Va (kt+1 ; at+1 ; zt+1 ) Rt+1 = 0

Now, recall that, at the optimum ( ) one will have policy functions for all the choice variables,
so fct ; nt ; mt ; bt g are all functions of the state variables fkt ; at ; zt g : To save on notation, let
uj (ct ; 1 nt ; mt ) = uj (t) and Va (kt+1 ; at+1 ; zt+1 ) = Va (t + 1) : To derive explicitely the enve-
lope condition totally dierentiate the value function:

@c @n @m @kt+1 @a
Vk (kt ; at ; zt ) = uc (t) + un (t) + um (t) + + Va (t + 1) t+1 Vk (t + 1)
@kt @kt @kt @kt @kt
(A.4)
and note that a depends on kt through mt and bt , but the exogenous technology process zt
@z
does not (i.e. @kt
= 0). Next, note:
mt 1
kt+1 = zt f (kt ; nt ) + (1 )kt + + R t bt 1 mt bt ct
1+ t

so:
@kt+1 @n @c @m @b
= zt fk (kt ; nt ) + zt fn (kt ; nt ) + (1 ) (A.5)
@kt @kt @kt @kt @kt
while:
mt
at+1 = + Rt+1 bt
1 + t+1
so:
@at+1 1 @m @b
= + Rt+1 (A.6)
@kt 1+ @kt @kt

63
combining (A.6)-(A.5) into (A.4):
2 3 2 3
uc (t) @c zt fk (kt ; nt ) + (1 ) " #
@kt 1
@m
6 7
Vk (kt ; at ; zt ) = 4 +un (t) @kt 5+ Vk (t + 1) zt fn (kt ; nt ) @n
@n 4 @kt
@c
@kt
@m
@kt
5+ Va (t + 1) 1+ @kt
t+1

@m @b +Rt+1 @b
@kt
+um (t) @kt @kt

rearranging:
8 9
> @c >
>
> @kt [uc (t) Vk (t + 1)] + >
>
>
< + @n [un (t) >
=
@kt h Vk (t + 1) zt fn (kt ; nt ]i
Vk (kt ; at ; zt ) = Vk (t + 1) (zt fk (kt ; nt ) + (1 ))+ @m Va (t+1)
>
> + @kt um (t) Vk (t + 1) + 1+ >
>
>
> t+1 >
>
: + @b [ V (t + 1) R V (t + 1)] ;
@kt a t+1 k

now, the F.O.C.s tell that each one of the terms in brackets is zero, so that the whole term in
braces vanishes:
Vk (kt ; at ; zt ) = Vk (t + 1) [zt fk (kt ; nt ) + (1 )]
but at the optimum (using the F.O.C.s again) uc (t) = Vk (t + 1) so replacing one nally arrives
at the rst envelope condition:

Vk (kt ; at ; zt ) = uc (t) [zt fk (kt ; nt ) + (1 )]

updating it one period and replacing in the F.O.C. for consumption, one arrives at the Euler
equation:
uc (ct ; 1 nt ; mt )
= [zt+1 fk (kt+1 ; nt+1 ) + (1 )] (A.7)
uc (ct+1 ; 1 nt+1 ; mt+1 )
next, replacing on the F.O.C. for labor:

un (ct ; 1 nt ; mt )
= [zt+1 fk (kt+1 ; nt+1 ) + (1 )] zt fn (kt ; nt ) (A.8)
uc (ct+1 ; 1 nt+1 ; mt+1 )

but note that from the consumption Euler equation (A.7):

uc (ct ; 1 nt ; mt )
uc (ct+1 ; 1 nt+1 ; mt+1 ) =
[zt+1 fk (kt+1 ; nt+1 ) + (1 )]

so replacing in (A.8) one arrives at the intra-temporal Euler equation:

un (ct ; 1 nt ; m t )
= zt fn (kt ; nt )
uc (ct ; 1 nt ; m t )

which, if one regards zt fn (kt ; nt ) as the real wage is a labor supply equation. Next, as a shortcut
to the the second envelope condition, use the F.O.C. w.r.t bonds to obtain:

Vk (kt+1 ; at+1 ; zt+1 )


Va (kt+1 ; at+1 ; zt+1 ) =
Rt+1

so that using the rst envelope condition:

uc (t + 1) [zt+1 fk (t + 1) + (1 )]
Va (kt+1 ; at+1 ; zt+1 ) =
Rt+1

64
Now using this condition in the F.O.C. for money:

uc (t + 1) [zt+1 fk (t + 1) + (1 )]
um (t) + = uc (t + 1) [zt+1 fk (t + 1) + (1 )]
Rt+1 (1 + t+1 )
1
um (t) = uc (t + 1) [zt+1 fk (t + 1) + (1 )] 1
Rt+1 (1 + t+1 )
um (t) it+1
=
uc (t + 1) [zt+1 fk (t + 1) + (1 )] 1 + it+1
um (t) it+1
=
uc (t) 1 + it+1

which is the usual money demand or LM curve equation.

65
Appendix B

The Maximum Principle

The Maximum Principle gives an approach to dynamic optimization that is alternative to the Dy-
namic Programming approach. It also exploits the concepts of states, controls, state-transition
functions and the Envelope Theorem. This section follows closely Dixit (1990).

B.1 Discrete time


Let:
zt be the control variable and yt the state variable.
The objective function be dened by:
F (yt ; zt )

The transition function be dened by:


Q(yt ; zt ) = yt+1 yt

The additional constraints:


G(yt ; zt ) 0

The dynamic optimization problem is therefore:


T
X
max F (yt ; zt )
t=0
s:t:
Q(yt ; zt ) = yt+1 yt
G(yt ; zt ) 0
with y0 0 given adn a terminal condition on yT +1 : The Lgrangian is:
T
X
L= fF (yt ; zt ) + t+1 [Q(yt ; zt ) yt+1 + yt ] + t G(yt ; zt )g
t=0

where t+1 and t are the multipliers associated with each constraint. The F.O.C. for the control
variable is easy to obtain:
@L
= 0 =) Fz (yt ; zt ) + t+1 Qz (yt ; zt ) + t Gz (yt ; zt ) (B.1)
@zt

66
but the condition for the state variable is not so straightforward since each yt appears in two
terms of the innite sum. To circumvent this issue, re-write the relevant part of the Lagrangean
as:
T
X
t+1 [yt yt+1 ] = 1 [y0 y1 ] + 2 [y1 y2 ] + ::: + T +1 [yT yT +1 ]
t=0
= 1 y0 1 y1 + 2 y1 2 y2 ::: T +1 yT T +1 yT +1
T
X
= yt ( t+1 t) + y0 1 yT +1 T +1
t=1

so that the problem becomes:


T
X
L = [F (yt ; zt ) + t+1 Q(yt ; zt ) + yt ( t+1 t) + t G(yt ; zt )]
t=1
+ F (y0 ; z0 ) + 1 Q(y0 ; z0 ) + y0 1 yT +1 T +1 )
| {z }

and note that the terms in braces pertain to t = 0; T + 1 whose values are given by initial and
terminal conditions so no need to worry about them. From this formulation, it is clear why t+1
is given the name "co-state". Now, the F.O.C. for the state variable can be derived more easily:
@L
= 0 =) t+1 t + [Fy (yt ; zt ) + t+1 Qy (yt ; zt ) + t Gy (yt ; zt )] = 0 8 t 6= 0; T + 1 (B.2)
@yt
This optimality condition states that, at the optimum, the overall marginal return from increas-
ing yt is zero; that is, the shadow prices prevent pure or excess return from holding yt : Now,
rearranging:
t+1 t = [Fy (yt ; zt ) + t+1 Qy (yt ; zt ) + t Gy (yt ; zt )] (B.3)
next dene the Hamiltonian:

H(yt ; zt ; t) = F (yt ; z)t + t+1 Q(yt ; zt ) (B.4)

and note that the optimization problem does not consist simply in maximizing the instantaneous
reward function F ( ) since future reward depends upon future values of the state variable, which
in tunr is related to its current value and the choice variable via the state-transition function
Q( ): Next, dene the Lagrangian, L; for the single-period problem:

L = H(yt ; zt ; t+1 ) + t G(yt ; zt ) (B.5)

and here, H(yt ; zt ; t ) is the objective function of this single-period problem. Following F.O.C.
(B.1), it is clear that zt is chosen so as to maximize (B.4), so let H(yt ; zt ; t+1 ) = H (yt ; t+1 ):
Next, notice that:
@L
= Fy (yt ; zt ) + t+1 Qy (yt ; zt ) + t Gy (yt ; zt )
@yt
so we can replace this in (B.3):

@L
t+1 t = = Ly
@yt

But notice, in the static problem (B.5) the Envelope Theorem applies and thus Ly = Hy so:

t+1 t = Hy (yt ; t) (B.6)

67
and a similar envelope condition for the co-state variable gives L = H (yt ; t) = Q(yt ; zt )
which replaced in the denition of the state-transition equation yields:

yt+1 yt = H (yt ; t) (B.7)

so the Maximum Principle states that rst order necessary and su cient conditions for the
optimization problem above are:

1. For each t, zt maximizes the Hamiltonian H(yt ; zt ; t) subject to the single preiod con-
straint(s) G(yt ; zt ):
2. The changes in yt ; t over time are governed by the pair of dierence equations (B.6)-(B.7).

B.2 Continuous time


State the problem above in continuous time:

ZT
max F (y(t); z(t))dt
0
s:t:
Q(y(t); z(t)) = y(t)
_
G(y(t); z(t)) 0

so that the (rearranged) Lagrangean is:

ZT
L = [F (y(t); z(t)) + (t)Q(y(t); z(t)) + y(t)( _ (t)) + (t)G(y(t); z(t))]
0
+ F (y(0); z(0)) + 1 Q(y(0); z(0)) + y(0) (0) y(T ) (T )

The condition for zt to maximize the Hamiltonian is (assuming it is legitimate to dierentiate


under the integral sign):

@L
= 0 =) Fz (y(t); z(t)) + (t)Qz (y(t); z(t)) + (t)Gz (y(t); z(t)) = 0
@z(t)

while the Hamiltonian itself is dened as:

H(y(t); z(t); (t)) = F (y(t); z(t)) + (t)Q(y(t); z(t))

and the pair of dierential equations (Pontryagin conditions) governing the behavior of the state
and co-state variables:

y(t)
_ = H (y(t); (t)) (B.8)
_ (t) = Hy (y(t); (t)) (B.9)

B.2.1 Current value vs. present value Hamiltonian


Using the notation above, suppose that:
t
F (y(t); z(t)) = e f (y(t); z(t))

68
so that the underlying objective function is the present-discounted value of the stream of instan-
taneous utility functions f (y(t); z(t)): Then the present value hamiltonian above can be written
as:
H pv (y; z; ) = e t f (y(t); z(t)) + (t)Q(y(t); z(t))
Now suppose that it is desirable to state the problem in current value terms; the Hamiltonian
would be:
H cv (y; z; ) = f (y(t); z(t)) + q(t)Q(y(t); z(t))
where:
t
q(t) = (t)e (B.10)
is the current-value shadow multiplier. Now revisit the Pontryagin conditions for the present
value problem:
@H pv
= 0 =) e t fz (y(t); z(t)) + (t)Qz (y(t); z(t)) = 0
@z(t)
y(t)
_ = H pv (y; z; ) = Q(y(t); z(t))
_ (t) = Hypv (y; z; ) = e t fy (y(t); z(t)) + (t)Qy (y(t); z(t))

Only the rst and last of these conditions involve discounting so, rewrite the rst in current-value
terms:

fz (y(t); z(t)) + (t)e t Qz (y(t); z(t)) = 0


fz (y(t); z(t)) + q(t)Qz (y(t); z(t)) = 0 (B.11)

and rewrite the last condition, still in present-value terms as:


t
_ (t)e = fy (y(t); z(t)) + (t)e t Qy (y(t); z(t)) (B.12)

now, since from (B.10):

q(t)
_ = _ (t)e t + (t)e t

= _ (t)e t + q(t)
t
_ (t)e = q(t)
_ q(t)

one can replace in (B.12) and:

q(t)
_ q(t) = fy (y(t); z(t)) + (t)e t Qy (y(t); z(t))
q(t)
_ q(t) = fy (y(t); z(t)) + q(t)Qy (y(t); z(t)) (B.13)

is the Pontryagin condition for the costate variable corresponding to the current-value optimiza-
tion problem.

69
Appendix C

First-Order Dierence Equations


and AR(1)

C.1 The AR(1) process


C.1.1 Representation and properties
Let "t N (0; 1) iid shock. Then zt follows an AR(1) process if we can write it as:
zt = (1 ') + 'zt 1 + "t
this is the recursive formulation of the AR(1) process because it recurs in the same form at each
t. To go from the recursive formulation, to the innite order MA formulation, rst replace zt 1
in the expression for zt :
zt = (1 ') + ' [(1 ') + 'zt 2 + "t 1] + "t
next, repeat this recursive replacing and after k + 1 times one obtains:
k
X1 k
X1
zt = (1 ') ' j + ' k zt k + 'j " t j
j=0 j=0

which can be rearranged as follows:


k
X1 k
X1 k
X1
zt = 'j ' ' j + ' k zt k + 'j " t j
j=0 j=0 j=0
2 3
k
X1 k
X1 k
X1
= 4'0 + 'j+1 5 ' ' j + ' k zt k + 'j " t j
j=0 j=0 j=0
k
X1 k
X1 k
X1
= + 'j+1 'j+1 + 'k zt k + 'j " t j
j=0 j=0 j=0
k
X1
k
= + ' zt k + 'j " t j
j=0

and if we let k ! 1 one gets:


1
X
zt = + 'j " t j
j=0

70
that is, the innite-order MA representation of the AR(1) process, saying that the AR(1) process
can be written as an innite sum of past shocks. If j'j = 1 we have a unit root or say that zt
has innite memory.

C.1.2 Conditional Distribution


The distribution of zt conditional on knowing zt 1 : Recall that a linear function of a normal RV
is itself a normal RV. Since at t the quantity zt 1 is known, it can be treated as a constant and
therefore zt , conditional on zt 1 is just a normal RV with its mean shifted by (1 ') +'zt 1 :To
obtain the conditional mean and variance of zt rst note that the variance remains unchanged
as 2 while the mean:

Et 1 [zt ] = Et 1 [(1 ') + 'zt 1 + "t ]


= Et 1 [(1 ') + 'zt 1 + Et 1 [ "t ]
]
= (1 ') + 'zt 1

so the conditional (on t 1) distribution of zt :


2
zt t 1 N ((1 ') + 'zt 1; )

C.1.3 Unconditional Distribution


The distribution of zt presuming no knowledge of zt 1 ; zt 2 :::This is equivalent to the distrib-
ution of zt conditional on knowing zt k for a very large k, or, equivalently, the distribution of
zt+k for a very large k with information on t: This is why the unconditional distribution is also
called the long-run distribution. To obtain this, use the innite order MA representation:
2 3
X1
Ezt = E 4 + 'j " t j 5
j=0
2 3
1
X
= + E4 'j " t j
5
j=0

since j'j < 1 and each "t j N (0; 1). Likewise we could compute the unconditional mean as:

E [zt ] = E [(1 ') + 'zt 1 + "t ]


= (1 ') + 'E [zt 1 ]
= (1 ') + 'E [zt ] (* E [zt ] = E [zt 1 ])

so that solving:

(1 ')E [zt ] = (1 ')


E [zt ] =

71
The unconditional variance is:
2 3
1
X
V ar [zt ] = V ar 4 + 'j " t j
5
j=0
2 3
1
X
= V ar 4 'j " t j
5
j=0
0 1
1
X
= @ 2
'2j A V ar ["t j]
j=0
2
=
1 '2
note that in the last step the following expansion is used:
1
X 1
('1 1) if :' 2 f 1; 1g
'2j = '2 1
1 if ' 2 f 1; 1g
j=0

so that the unconditional distribution of zt is:


2
zt N ;
1 '2

Ntaurally, as long as 0 < j'j < 1 the unconditional variance is greater than the conditional
variance.

C.2 Linear First-Order Dierence Equations (FODE)


Consider the linear FODE:
xt xt 1 =b
can be solved using various techniques.

C.2.1 Induction & Geometric Series


Given an initian value x0 :

x1 = x0 + b
x2 = x1 + b = ( x0 + b) + b = 2 x0 + b + b
x3 = x2 + b = ( ( x0 + b) + b) + b = 3 x0 + 2
b+ b+b
.. .. .. ..
. = . . .
t t 1 t 2
xt = x0 + b+ b + ::: + 2 b + b + b
t 1
X
t k
= x0 + b
k=0

which using the geometric series arithmetic:


t 1
X t
k 1
=
1
k=0

72
we obtain:
t
t 1
xt = x0 + b
1
so the solution to the LFODE takes the form:
t b b
xt = x0 + +
1 1

C.2.2 Homogeneous part and General solution


An alternative 5-step approach is as follows:
1. Find the steady state of x (i.e., xt = xt 1 = x):
b
x=
1
2. Find the solution to the homogeneous part:
xh;t xh;t 1 =0 (C.1)
and let:
t
xh;t =
be a solution to the homogeneous part for each t, so that:
t 1
xh;t 1 =
and (C.1) becomes:
t t 1
= 0
t 1
( ) = 0
t 1
this equation has two solutions, the trivial solution where = 0 and the non-trivial
solution ( ) = 0 so using the latter:
=
and replacing in the hypothesized solution:
t
xh;t =
is the solution to the homogeneous part.
3. Find the General Solution: The general solution is a linear combination of the homogeneous
solution and the steady state:
b
xt = +a t
1
where a is the lin-comb undetermined coe cient.
4. Find a for some initial x0 :
b 0 b
x0 = +a =) a = x0
1 1
5. Replacing in the general solution:
t b b
xt = x0 +
1 1
which is, of course, the same result obtained by use of the induction technique.

73
C.2.3 Asymptotic Stability
The above solution can be expressed as:

t b b
f (x0 ) = x0 +
1 1
b
which has, as xed point, x0 = 1 , i.e.:

b b
f =
1 1

so 1 b can be interpreted as a stationary point ("steady-state"). We say that the above FODE
is asymptotically stable if:
lim (xt x) = 0
t!1

which will be the case only if:


j j<1
and note that if 0 < < 1 we observe a monotonic convergence towards x while if 1< <0
the convergence is via oscilations around x:

Example 17 A version of the Cagan (1956) model. Suppose that supply is based on expected
prices while demand is based on actual prices:
d
yt+1 = + pt+1
s
yt+1 = + Et pt+1

and equilibrium is therefore given by:

+ pt+1 = + Et pt+1

expectations are formed according to the adaptive behavior:

Et pt+1 Et 1 pt = (pt Et 1 pt )

which can be written using the Lag operator as:

Et pt+1 = pt + (1 )LEt pt+1


[1 (1 )L] Et pt+1 = pt
pt
Et pt+1 =
[1 (1 )L]

nd replacing in the equilibrium condition:

pt
+ pt+1 = +
[1 (1 )L]
+ pt+1 (1 )pt = + pt
( ) 1
pt+1 = + 1 pt

74
and now we have a FODE without expectation terms which we
h can solve using
i the abovementioned
techniques. Using the 5 step method. Let ( )
= and 1 1
= then:

1) : Sty Sate =) p =
1
t
2) : Ph;t+1 = =) t 1 ( )=0
: =) ph;t+1 = t
t
3) : Gral:Sol: =) pt+1 = +a
1
0
4) : F ind coef =) p0 = +a
1
: =) a = p0
1
t
5) : Replace =) pt+1 = + p0
1 1
so that the solution to the price equation is:
t
1
pt+1 = + p0 1
1 1
with stationary price level equal to p = 1 and asymptotic stability condition:

1
1 <1

C.3 Systems of linear FODE (or VDE)


Suppose that Xt is a 2-dimensional vector and:
Xt M Xt 1 =B
where M is a 2 2 matrix known as the state-transition matrix and B is 2 1:This system can
be solved in a similar fashion as single equations:
1. Solve for the steady state:
1
X = (I M) B
2. Find the solution to the homogeneous part:
Xh;t M Xh;t 1 =0 (C.2)
and again, suppose it has the shape:
t
Xh;t = A
so that:
t 1
Xh;t = A
where A is a column vector of unknown parameters and is some scalar. Replacing in
(C.2) implies:
t
A MA t 1 = 0
t 1
A( I M ) = 0
A(M I) = 0 (C.3)

75
Again, this equaton will have two solutions. However, we are only interested in the non-
trivial solution so A 6= 0 which in turn implies that (M I) must be singular, i.e.:

det(M I) = 0
m11 m12
det = 0
m21 m22

which yields a quadratic equation in :


2
(m11 + m22 ) + (m11 m22 m12 m21 ) = 0
| {z } | {z }
2
tr(M ) + det(M ) = 0

the roots of this equation are called characteristic roots or eigenvalues. The two solutions
to this equation are:
1h p i
1; 2 = tr(M ) tr(M )2 4 det(M )
2
and substituting 1; 2 in (C.3) results in the two eigenvectors of A :

1
A1 = m11 1
m12

and:
1
A2 = m11 2
m12

and the solution to the homogeneous system is given by the linear combination:

Xh;t = c1 A1 t
1 + c2 A2 t
2

where c1 ; c2 are undetermined coe cients.


3. Obtain the General Solution as in the previous section:

Xt = X + Xh;t
Xt = (I M ) 1 B + c1 A1 t
1 + c2 A2 t
2 (C.4)

4. As before, solve for the undetermined coe cients using some initial value vector X0 :
1
X0 = (I M) B + c1 A1 0
1 + c2 A2 0
2

this is a system of linear equations from which the coe cients c1 ; c2 can be determined.
Then substituting into (C.4) gives the solution to the VDE.

C.3.1 Asymptotic stability


It is easily seen that the asymptotic stability of the stationary point X will depend upon the
eigenvalues 1 ; 2 of the stae-transition matrix, A:Consider only the real-valued ones. If they
both lie within the unit circle:
j i j < 1 for i = 1; 2
the critical (stationary) point is asymptotically stable. If only one of them is inside the unit
circle, this implies that there are only two possible converging trajectories (i.e., a saddle-point).

76

You might also like