Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Look up keyword
Like this
1Activity
0 of .
Results for:
No results containing your search query
P. 1
Macro Summary

Macro Summary

Ratings: (0)|Views: 12|Likes:
Published by Eduardo Jimenez

More info:

Published by: Eduardo Jimenez on Jun 27, 2012
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

11/01/2014

pdf

text

original

 
Luke SteinStanford graduate economics core (2006–7)Econ 210–211 reference and review (last updated April 11, 2012)
1 Models
Two period intertemporal choice model
(Nir 1-11)
Maximize
(
0
,
1
) subject to
0
+
0
0
and
1
RS 
0
. Theconstraints can be rewritten
0
+
1
R
1
0
.
Neoclassical growth model (discrete time)
(Nir 3-3–5, 11-3,8, 10, 16–28, 12-12–26)
(a.k.a. Ramsey model) Maximize
t
β
t
(
c
t
) subject to:1.
c
t
+
k
t
+1
(
k
t
,n
t
) + (1
δ
)
k
t
for all
t
(RHS can bedenoted
(
k
t
,n
), or
(
k
t
), since the lack of disutility of working ensures that
n
t
= 1);2.
k
t
+1
0 for all
t
;3.
c
t
0 for all
t
;4.
k
0
is given.We further assume
(
·
) satisfies Inada conditions; that theproduction function is constant returns to scale and satis-fies
k
>
0,

kk
<
0,
n
>
0, and

nn
<
0; and TVClim
t
→∞
β
t
(
c
t
)
(
k
t
)
k
t
= 0.The problem can be rewritten in “SL canonical form”as maximizing
t
β
t
(
(
k
t
)
k
t
+1
) subject to
k
t
+1
[0
,
(
k
t
)], the given
k
0
, and conditions on
(
·
) and
(
·
)as above. In functional equation form, we write
(
k
)
max
k
[0
,f 
(
k
)]
(
(
k
)
k
)+
βV 
(
k
) (again, with additionalconditions as above).Steady state satisfies
βf 
(
k
) = 1. Note the utility func-tion does
not 
affect the steady state (although it will affectdynamics).Linearization of the intereuler
u
(
(
k
t
)
k
t
+1
)
βf 
(
k
t
+1
)
u
(
(
k
t
+1
)
k
t
+2
) = 0about steady state gives:
k
t
+2
k
k
t
+1
k
1 +
1
β
+
β
(
c
)

(
c
)

(
k
)
1
β
1 0
k
t
+1
k
k
t
k
The optimal policy function
g
(
k
) satisfies:1.
g
(
·
) is single-valued, since the value function is strictlyconcave.2.
g
(0) = 0 since that is the only feasible choice.3.
g
(
k
) is in the interior of 
γ 
(
k
), since otherwise exactlyone of 
(
c
) and
(
c
) would be infinite, violating theIntereuler condition.4.
g
(
k
)
>
0, since as
k
increases, marginal cost of savinggoes down, while marginal benefit stays same.5. There is unique
k
>
0 such that
g
(
k
) =
k
.6.
k > k
=
k > g
(
k
)
> k
, and
k < k
=
k <g
(
k
)
< k
; i.e., capital moves closer to (without crossingover) the steady-state level.7. The sequence
k
0
,g
(
k
0
)
,g
(
g
(
k
0
))
,...
displays mono-tonic and global convergence.
Endowment economy, Arrow-Debreu
(Nir 13-4–6)
At everyperiod household gets
y
t
units of good; there is no storage,and trading in all commodities (goods delivered in each pe-riod) takes place in period 0. Maximize
β
t
(
c
t
) subjectto
 p
t
c
t
 p
t
y
t
(budget constraint), and ensure
y
t
=
c
t
(market clearing).
Endowment economy, sequential markets
(Nir 13-7–9)
At ev-ery period household gets
y
t
units of good; there is no stor-age, and trading in “assets” (loans across periods) takes placeeach period. Maximize
β
t
(
c
t
) subject to
c
t
+
a
t
+1
y
t
+
R
t
a
t
(budget constraint), and ensure
y
t
=
c
t
,
a
t
+1
= 0(market clearing).
Production economy, Arrow-Debreu
(Nir 13-10–6)
Householdowns capital rented for production by competitive firms,which also rent labor from households. Capital depreciates atrate
δ
. There is no storage, and trading in all commodities(goods delivered, capital rental, and labor in each period)takes place in period 0.1. Households maximize
β
t
(
c
t
) subject to
 p
t
[
c
t
+
k
t
+1
]
 p
t
[
r
t
k
t
+ (1
δ
)
k
t
+
n
t
w
t
].2. Firms maximize (in each period)
p
t
(
k
t
,n
t
)
 p
t
r
t
k
t
 p
t
w
t
n
t
.3. Markets clear:
c
t
+
k
t
+1
=
(
k
t
,n
t
) + (1
δ
)
k
t
.Note the rental rate of capital (
r
t
) and wage rate (
w
t
) aremeasured in units of consumption good.
Production economy, sequential markets
(Nir 13-25)
Asabove, household owns capital rented for production bycompetitive firms, which also rent labor from households.Capital depreciates at rate
δ
. There is no storage, and eachperiod, goods delivered, capital rental, and labor are traded.Note there is no inter-period trading, so we do not need aprice for goods.1. Households maximize
β
t
(
c
t
) subject to
c
t
+
k
t
+1
r
t
k
t
+ (1
δ
)
k
t
+
n
t
w
t
.2. Firms maximize (in each period)
(
k
t
,n
t
)
r
t
k
t
w
t
n
t
.3. Markets clear:
c
t
+
k
t
+1
=
(
k
t
,n
t
) + (1
δ
)
k
t
.Formulated recursively (assuming
n
= 1),1. Households have
(
k,
) = max
c,k
[
(
c
) +
βV 
(
k
,
)subject to
c
+
k
=
R
(
)
k
+
(
); where
is aggre-gate capital, and
R
(
) and
(
) are the rental andwage rates as functions of aggregate capital. We furtherrequire the “forecast” of future capital to be rational:
=
G
(
), where
G
(
·
) is the optimal aggregate policyfunction.2. Firms maximize
(
)
RK 
, which gives FOC
R
(
) =
k
(
) + 1
δ
and
(
) =
n
(
).3. Markets clear:
+
=
(
) + (1
δ
)
.4. Consistency:
G
(
) =
g
(
K,
).
Overlapping generations
(Nir 17-15–6, 17-32)
In an endowmentOLG economy, a competitive equilibrium is Pareto optimaliff 
t
=0
1
/p
t
=
where the interest rate is
R
t
=
p
t
/p
t
+1
.In an an OLG economy with production, a competitive equi-librium is dynamic efficient iff 
K
(
) + 1
δ
1.
Real Business Cycles model
(Nir 18-7–14)
Benchmark RBCmodel is NCGM with endogenous hours, ownership of firmsand capital by households, and stochastic productivity
A
t
multiplying labor (so production function is
(
k
t
,A
t
n
t
)).We specified:1. CRRA utility:
(
c,n
) =
c
1
σ
11
σ
n
1+
χ
1+
χ
;2. Cobb-Douglas production:
y
t
=
A
t
k
αt
n
1
αt
;3. Productivity given by log
A
t
=
ρ
log
A
t
1
+
ε
t
with
ε
t
distributed iid normal.
Optimal taxation—primal approach
(Nir 10)
Solve HH prob-lem and combine FOC and budget constraint with govern-ment budget constraint to get one expression that ties to-gether allocations but has
no prices or taxes
. Then gover-ment maximizes HH utility over
allocations
subject to thisconstraint.Note that in the case with capital,
the problem is not sta-tionary; i.e., we cannot use our typical dynamic program-ming tools, and need to worry about the government’s abilityto commit to a future tax policy.
We generally need to impose restrictions on first-period taxation, since otherwise the government will tax capital very highly then (since it’s supplied perfectly inelastically).
1
 
2 Imperfect competition
Imperfect competition model
(Nir 21)
Consider two stages of production: final good manufacturer operates competitively,purchasing inputs from intermediate good producers, whoare small enough to take general price levels and aggregatedemand as given, but whose products are imperfect substi-tutes to the production process.
Imperfect competition: final goods
(Nir 21)
Final good pro-ducer has Dixit-Stiglitz production technology:
=
 
10
Q
ρj
dj
1
which is constant elasticity of substitution (
11
ρ
between allpairs of inputs) and constant returns to scale.
ρ
(0
,
1),with
ρ
1 corresponding to a competitive market where allinputs are perfect substitutes.Conditional factor demands are therefore
Q
j
(
) =
 p
j
λ
1
ρ
1
=
 p
j
1
ρ
1
Y,
where the Lagrange multiplier on [
 
10
Q
ρj
dj
]
1
also sat-isfies
λ
=
/Y 
, the (aggregate) average cost. Solvingfor this gives
λ
=
=
 
10
 p
ρρ
1
j
ρ
1
ρ
.
The own-price elasticity of demand is
η
1
ρ
1
, ignoring theeffects through
.
Imperfect competition: intermediate goods
(Nir 21)
Inter-mediate goods producers have identical production technolo-gies (affected by same shocks):
Q
j
=
zk
αj
n
1
αj
φ
, where
φ >
0 is overhead cost (since there aren’t really economicprofits). Has constant marginal cost, and therefore increas-ing returns to scale (because of 
φ
).Monopolistic problem is to maximize over prices
pQ
(
 p
)
c
(
Q
(
 p
)); first-order condition gives markup
p/
MC =
η
1+
η
=
1
ρ
.
3 General concepts
Cobb-Douglas production function
(Macro P.S. 1)
(
K,
) =
zK 
α
1
α
. Constant returns to scale. Fraction
α
of outputgoes to capital, and 1
α
to labor.
Constant relative risk aversion utility function
(Nir 14-15,18-72–4)
(
c
) = (
c
1
σ
1)
/
(1
σ
), with
σ >
0 (reduced tolog utility with
σ
= 1;
empirically we expect
σ
[1
,
5]).Relative risk aversion is
σ
. Higher risk aversion also cor-responds to a higher desire to smooth consumption overtime.
Additively separability with stationary discounting
(Nir 1-24)
General assumption that
(
0
,
1
) =
u
(
0
) +
βu
(
1
)with
β
(0
,
1).
Intertemporal Euler equation
(Nir 1-28, 11-5)
u
(
c
t
) =
βRu
(
c
t
+1
), where
R
is the return on not-consuming (e.g.,
(
k
t
+1
)
k
(
k
t
+1
,n
t
+1
)+(1
δ
)). This first-order differ-ence equation is given by FOCs of the maximization problemfor two consecutive time periods.
Intertemporal elasticity of substitution
(Nir 1-31, Max notes)
d
log
C
0
C
1
/d
log
R
=
d
log
C
0
C
1
/d
log
(
C
0
)
(
C
1
)
.For CRRA utility function
(
c
) =
c
1
σ
/
(1
σ
), intereuleris
σ
0
=
βC 
σ
1
R
so IEoS =
1
σ
= RRA
1
.
General equilibrium
(Nir 2-11)
A competitive equilibrium is a setof 
allocations
and
prices
such that1. Actors (i.e., households, firms, ...) maximize their ob- jectives;2. Markets clear; and3. Resource constraints are satisfied.
Inada conditions
(Nir 3-5)
We generally assume utility satisfiesInada conditions, which imply that the nonnegativity con-straint on consumption will not bind:1.
(0) = 0;2.
is continuously differentiable;3.
(
x
)
>
0 (
is strictly increasing);4.

(
x
)
0 (
is strictly concave);5. lim
x
0+
(
x
) =
;6. lim
x
→∞
(
x
) = 0.
Transversality condition
(Macro P.S. 2)
In NCGM,lim
t
→∞
β
t
(
c
t
)
(
k
t
)
k
t
= 0.
No Ponzi condition
(Max notes)
Credit constraint on agents,which should never bind but suffice to prevent the optimalityof a “doubling” strategy. For example,
t
,
a
t
¯
k
for some¯
k
.
Dynamic systems
(Nir 12-4–11, Macro P.S. 5-4)
Let the sequence
{
x
t
}
evolve according to
x
t
+1
=
Wx
t
. Using eigen de-composition
=
Λ
1
, giving the “decoupled system”
1
x
t
+1
= Λ
1
x
t
. Then if ˜
x
t
1
x
t
, we have˜
x
ti
=
λ
ti
˜
x
0
i
and hence
x
t
=
Λ
t
˜
X
0
.For 2
×
2 case, if 1.
|
λ
1
|
>
1 and
|
λ
2
|
>
1, then the system “explodes” un-less ˜
x
0
,
1
= ˜
x
0
,
2
= 0 (source).2.
|
λ
1
|
<
1 and
|
λ
2
|
<
1, then the system converges forany ˜
x
0
,
1
and ˜
x
0
,
2
(sink).3.
|
λ
1
|
<
1 and
|
λ
2
|
>
1, then the system converges forany ˜
x
0
,
1
as long as ˜
x
0
,
2
= 0 (saddle path).The speed of convergence for a state variable equals one mi-nus the slope of the optimal policy function (1
g
(
k
)).
First Welfare Theorem
(Nir 13-27)
The competitive equilibriumis Pareto optimal. Then if we know our solution to the socialplanner problem is the unique Pareto optimal solution, weknow it must equal the competitive equilibrium.
Ramsey equilibrium
(Nir 20-3, final solutions)
A Ramsey equilib-rium is allocations, prices, and taxes such that:1. Households maximize utility subject to their budgetconstraints, taking prices and taxes as given, and2. Government maximizes households’ utility while financ-ing government expenditures (i.e., meeting its budgetconstraint).
4 Dynamic programming mathematics
Metric space
(Nir 6-3–4)
A set
and a distance function
ρ
:
×
R
such that:1.
x
,
y
,
ρ
(
x,y
)
0;2.
x
,
y
,
ρ
(
x,y
) = 0
x
=
y
;3.
x
,
y
,
ρ
(
x,y
) =
ρ
(
y,x
);4. Triangle inequality:
x
,
y
,
z
,
ρ
(
x,z
)
ρ
(
x,y
) +
ρ
(
y,z
).
Normed vector space
(Nir 6-5–6)
A vector space
and a norm
·
:
R
such that:1.
x
,
x
0;2.
x
,
x
= 0
x
= 0;3. Triangle inequality:
x
,
y
,
x
+
y
x
+
y
.4. Scalar multiplication:
x
,
α
R
,
αx
=
|
α
|
x
.Note any normed vector space induces a metric space with
ρ
(
x,y
)
x
y
.
Supremum norm (
R
n
)
(Nir 6-7–8)
·
s
:
R
n
R
with
x
s
sup
i
=1
,...,n
|
x
i
|
.
Euclidean norm
(Nir 6-9)
·
E
:
R
n
R
with
x
E
n
 
n
i
=1
|
x
i
|
n
.
Note the Taylor approximation of 
c
1
σ
1 about
σ
= 1 is
c
1
σ
1
(
c
0
1)
c
0
(log
c
)(
σ
1) = (1
σ
)log
c
. Thus as
σ
1, we have
(
c
) = (
c
1
σ
1)
/
(1
σ
)
log
c
.
2
 
Continuous function
(Nir 6-10; Micro math 3)
:
R
is contin-uous at
x
iff 
y
and
ε >
0, there exists
δ >
0 such that
y
x
< δ
=
|
(
y
)
(
x
)
|
< ε
.Equivalently, iff for all sequences
x
n
converging to
x
, thesequence
(
x
n
) converges to
(
x
).
Supremum norm (real-valued functions)
(Nir 6-7–8)
LetC(
X
) denote the set of bounded, continuous functions from
X
to
R
. Then
·
s
: C(
X
)
R
with
s
sup
x
X
|
(
x
)
|
.
Convergent sequence
(Nir 7-4)
The sequence
{
x
i
}
i
=0
in
con-verges to
x
(or equivalently, the sequence has limit
x
) iff 
ε >
0, there exists
n
such that
i > n
=
x
i
x
< ε
. Inother words, there is a point in the sequence beyond whichall elements are arbitrarily close to the limit.
Cauchy sequence
(Nir 7-9)
The sequence
{
x
i
}
i
=0
in
is Cauchyiff 
ε >
0, there exists
n
such that
i > n
and
j > n
togetherimply that
x
i
x
j
< ε
. In other words, there is a point inthe sequence beyond which all elements are arbitrarily closeto each other. Every convergent sequence is Cauchy (shownvia triangle inequality), but not every Cauchy sequence isconvergent.
Complete metric space
(Nir 7-10–16)
A metric space (
S,ρ
) iscomplete iff every Cauchy sequence in
converges
to somepoint in 
. Importantly, if C(
X
) is the set of bounded, con-tinuous functions from
X
to
R
, and
·
s
: C(
X
)
R
is thesup norm, then (C(
X
)
,
·
s
is a complete metric space.
Contraction
(Nir 8-3–4)
If (
S,ρ
) is a metric space, the operator
:
is a contraction of modulus
β
(0
,
1) iff 
x
,
y
, we have
ρ
(
Tx,Ty
)
βρ
(
x,y
). That is,
brings anytwo elements closer together. Every contraction is a contin-uous mapping.
Contraction Mapping Theorem
(Nir 8-5)
[a.k.a. Banach fixedpoint theorem.] If 
is a contraction on a
complete
metricspace, then1.
has exactly one fixed point
such that
T
=
;and2. The sequence
{
i
}
where
i
+1
=
T
i
converges to
from any starting
0
.
Contraction on subsets
(Nir 8-11)
Suppose
is a contraction ona complete metric space (
S,ρ
), with fixed point
=
T
.Further suppose
is a closed set, that
, and that
y
,
Ty
. Then
.
Blackwell’s sufficient conditions for a contraction
(Nir 8-12–3)
Blackwell gives sufficient (not necessary) conditions foran operator to be a contraction on the metric space
B
(
R
n
,
R
)(the set of bounded functions
R
n
R
) with the sup norm.An operator
is a contraction if it satisfies1. Monotonicity: if 
x
,
(
x
)
g
(
x
), then
x
,
T
(
x
)
Tg
(
x
); and2. Discounting: there exists
β
(0
,
1) such that for all
α
0,
(
·
), and
x
, we have
(
(
x
)+
α
)
(
(
x
))+
βα
[slight abuse of notation].
Principle of Optimality
(Nir 9)
Under certain conditions, thesolutions to the following two problems are the same:1. Sequence problem:
(
x
0
) = max
{
x
t
+1
}
t
=0
β
t
(
x
t
,x
t
+1
)such that
x
0
given and
t
,
x
t
+1
Γ(
x
t
).2. Functional equation:
(
x
) = max
x
Γ(
x
)
[
(
x,x
) +
βV 
(
x
)].Assume1. Γ(
x
) is nonempty for all
x
;2. For all initial conditions
x
0
and feasible plans
{
x
t
}
, thelimit
u
(
{
x
t
}
)
lim
n
→∞
β
t
(
x
t
,x
t
+1
) exists (althoughit may be
±∞
).Then:1. If 
(
x
0
) is the supremum over feasible
{
x
t
}
of 
u
(
{
x
t
}
),then
satisfies the FE.2. Any solution
to the FE that satisfies boundednesscondition lim
n
→∞
β
n
(
x
n
) = 0 is a solution to theSP.3. Any feasible plan
{
x
t
}
that attains the supremum inthe SP satisfies
(
x
t
) =
(
x
t
,x
t
+1
) +
βw
(
x
t
+1
) for
t
.4. Any feasible plan
{
x
t
}
that satisfies
(
x
t
) =
(
x
t
,x
t
+1
) +
βw
(
x
t
+1
) for
t
andlim
n
→∞
sup
β
t
(
x
t
)
0 attains the supremum inthe SP.
[?]
So approach is: solve FE, pick the unique solution that sat-isfies boundeness condition, construct a plan from the policycorresponding to this solution, check the limit condition tomake sure this plan is indeed optimal for the SP.
Dynamic programming: bounded returns
(Nir 10)
Given theSP/FE as above, assume:1.
x
takes on values in a convex subset of 
R
l
, and Γ(
x
) isnonempty and compact-valued for all
x
. This impliesassumption1above.2. The function
is bounded and continuous;
β
(0
,
1).Together with assumption1,this implies assumption2 above.3. For each
x
, the function
(
·
,x
) is increasing in eachof its first arguments.4. Γ(
·
) is monotone; i.e.,
x
1
x
2
=
Γ(
x
1
)
Γ(
x
2
).5.
is strictly concave.6. Γ is convex; i.e., if 
x
1
Γ(
x
1
) and
x
2
Γ(
x
2
), then
θx
1
+ (1
θ
)
x
2
Γ(
θx
1
+ (1
θ
)
x
2
).7.
is continuously differentiable on the interior of the seton which it is defined.Then1. Under assumptions1and2,the operator
where
T
(
x
)
max
x
Γ(
x
)
[
(
x,x
)+
βV 
(
x
)] maps boundedcontinuous functions into bounded continuous func-tions, is a contraction (and therefore has a unique fixedpoint), and is the value function for the correspondingFE.2. Under assumptions14, the value function
(uniquefixed point of 
as defined above) is strictly increasing.3. Under assumptions12and56,the value function
isstrictly concave, and the corresponding optimal policycorrespondence
g
is a continuous function.4. Under assumptions12and57,the value function
isdifferentiable (by Benveniste-Scheinkman) and satisfiesenvelope condition
(
x
0
) =
∂F ∂x
|
(
x
0
,g
(
x
0
)
.
Benveniste-Scheinkman Theorem
(Nir 10-25)
Suppose1.
X
R
l
is a convex set;2.
:
X
R
is a concave function;3.
x
0
interior(
X
); and
D
is a neighborhood of 
x
0
, with
D
X
;4.
Q
:
D
R
is a concave differentiable function with
Q
(
x
0
) =
(
x
0
) and
x
D
,
Q
(
x
)
(
x
).Then
(
·
) is differentiable at
x
0
with
(
x
0
) =
Q
(
x
0
).The envelope condition of a value function is sometimescalled the Benveniste-Scheinkman condition.
5 Continuous time
Dynamic systems in continuous time
(Nir 14-2–3, 7)
In
R
, thesystem˙
x
t
=
ax
t
dx
t
dt
=
ax
dx
t
x
=
adt
log
x
t
=
at
+
c
x
t
=
x
0
e
at
converges iff real(
a
)
<
0.In
R
n
, the system ˙
x
t
=
Ax
t
˙˜
x
t
= Λ˜
x
t
where ˜
x
t
=
1
x
t
and the eigen decomposition is
A
=
Λ
1
. Therefore˜
x
t
=
e
Λ
t
˜
x
0
=
e
λ
1
t
...
e
λ
n
t
˜
x
0
.
For 2
×
2 case (recall det
A
=
λ
i
and tr
A
=
λ
i
),det
A <
0: Saddle pathdet
A >
0, tr
A >
0: Unstabletr
A <
0: Sink3

You're Reading a Free Preview

Download
scribd
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->