Macroeconomic Dynamics and Policy
F. Langot
Univ. Le Mans (GAINS) & CES & Cepremap & IZA
November, 2009
Part I
Motivation
Motivation : The Lucas Critique (1976)
”Given that the structure of an econometric model consist of optimal decision rules of economic agents, and that optimal decision rule vary systematically with change in the structure of series relevant of the decision maker, it follows that any change in policy will
systematically alter the structure of econometric models”
(Lucas (1976), p.40)
Motivation : The Lucas Critique (1976)
Consider the simple representation of the log–linearized version of a forward-looking model
y
t= aE
ty
t+1+ bx
twhere
|a|< 1 (1) where y
tand x
trespectively denote the endogenous and the forcing variables. x
tis assumed to follow a rule of the form
x
t= ρx
t−1+ σε
twith ε
t ∼iid (0, 1) (2) The solution to this simple model is given by
½
y
t=
1−aρbx
tx = ρx + σε
Motivation : The Lucas Critique (1976)
This reduced form may be used to estimate the parameters of the
model
½y
t= αx
b tx
t= ρx
b t−1+
bσε
tUsing this model, one can predict the impact of an innovation ε
t= 1 in period t + n :
y
t+n= α
bρ
bnThis forecast is true if and only if the policy rule does not change,
ie. if and only if ρ is stable.
Motivation : The Lucas Critique (1976)
At the opposite, assume that the policy rule change. ρ becomes ρ/2. This reduced form of the model becomes
(
y
t= αx
b tx
t=
³bρ 2
´
x
t−1+ σε
b twhereas the true model is
(
y
t=
b1−a
(
ρ2) x
tx
t=
¡ρ2
¢
x
t−1+ σε
tMotivation : The Lucas Critique (1976)
Proposition
There is a gap between the naive prediction y
t+n= α
bµ
ρ
b2
¶n
= b
1
−aρ
µρ
b2
¶n
and the “true” prediction
y
t+n= b 1
−a
¡ρ2
¢ µ
ρ
b2
¶n
Motivation : The Lucas Critique (1976)
Ultimately, it [demanding that one’s structural relations be derived from individual optimization] is the only way in which the ”observational equivalence” of a multitude of alternative possible structural interpretations of the co–movements of aggregate series can be resolved”
(Rotemberg and Woodford (1998), p.1)
PropositionTo estimate or to calibrate the DEEP parameters
{a,b, ρ, σ} using theoretical restrictions.
References : AES special issue (2002)
Motivation : The Lucas Critique (1976)
The implication of the Lucas critique
1
The economist must find micro-foundations to the equilibrium in order to identify both the structural parameters and the policy parameters. This is necessary to forecast the
implications of a policy change.
2
For the policy maker, the Lucas critique implies that the
equilibrium of the economy is the equilibrium of a game : the
private agents expect the policy changes and then the
discretionary policy becomes inefficient (see Kydland and
Prescott [1977] : Rules versus discretion)
Part II
A basic framework : the RBC
model
The RBC model : the production function
All firms have access to the same technology :
Y
t= A
tF (K
t, X
N,tN
t) = A
tK
tα(X
N,tN
t)
1−αwhere
K
tis the capital stock determined at date t
−1. The law of motion of this stock is given by :
K
t+1= (1
−δ)K
t+ I
tN
tis the total hours determined at date t
A
tis the technological shock observed at date t,
X
N,tis the exogenous technical progress which is
The RBC model : the technological shock
If all the markets are walrasian, the technological shock can be measured by the Solow reisual (Solow (1957)) :
Let y
t= Y
t/X
N,tand k
t= K
t/X
N,tthen y
t= A
tF (k
t, N
t) By differentiating this equation, we obtain :
∆A
tA
t= ∆y
ty
t −∂F
t∂k
tk
ty
t∆k
tk
t −∂F
t∂N
tN
ty
t∆N
tN
tIf all the markets are walrasian, then
∂F∂ktt
kt
yt
=
rtyktt
= α
band
∂Ft
∂Nt
Nt
yt
=
wytNtt= 1
−α.
bThe RBC model : the households
Preferences
u(C
t, L
t) =
(
θ log(C
t) + (1
−θ) log(L
t) (
CtθL1−θt)
1−σ−11−σ
Ct denotes the consumption Lt denotes the leisure
The time allocation is constrained by : T
0= L
t+ N
tThe budgetary constraint is given in a complete market economy :
C
t+
Zp
t+1q
t+1dA
t+1 ≤q
t+ w
tN
twhere
Rp
t+1q
t+1dA
t+1represents the value of contingent
claims at time t yielding q units of the consumption good
The RBC model : general equilibrium
The general equilibrium is definied by the set of functions
{Ct, K
t+1, N
t, Y
t}, solution of the following system :u0Ct = βEt
·µ
1−δ+αYt+1
Kt+1
¶ uC0t+1
¸
u0Lt
uC0t = (1−α)Yt
Nt
Kt+1 = (1−δ)Kt+Yt−Ct
Yt = AtKtα(XN,tNt)1−α
log(At+1) = ρlog(At) + (1−ρ) log(¯A) +²t
This system gives the equilibrium dynamics of C , K , N , Y ,
The RBC model : resolution method
In order to solve this non-linear dynamic system, we take its first-order Taylor series approximation around the steady-state.
We assume that the utility function is u = θ log(C
t) + (1
−θ) log(L
t)
We obtain the following linear equations :
−bct = Et[−bct+1+a1bkt+1+a2bnt+1+a3bat+1+o(x2)]
b
ct+b1bnt = b2bkt+b3bnt+b4bato(x2)
bkt+1 = c1bkt+c2bnt+c3bat+c4bct+o(x2) bat+1 = ρbat+b²t+1+o(x2)
The RBC model : resolution method
Then, this leads to the first order vector system M
1s
t= M
0s
t+1+ M
2η
t+1where s
t= [b c
t,
bk
t,
ba
t]
0where η
t+1= [b ²
t+1, w
bt+1c, w
bt+1k, w
bt+1a]
0with w
bt+1x= E
t[x
t+1]
−x
t+1, for x = c, k , a.
By premultiplying the preceding equation by the matrix M
1−1, we obtain :
s
t= Js
t+1+ Rη
t+1with
bk
0,
ba
0predetermined
The RBC model : resolution method
Let Q, the matrix of the eigenvectors, such that Q
−1JQ = Λ with dim(Q) = 3
×3 and where Λ is the matrix of the eigenvalues, we denote :
z
t= Q
−1s
tand φ
t= Q
−1Rη
tThis leads to the following system :
z
tfz
tbz
tx
=
λ
f0 0 0 λ
b0 0 0 λ
x
z
t+1fz
t+1bz
t+1x
+
φ
ft+1φ
bt+1φ
xt+1
where, in the general case (dim(Λ) > 3
×3) λ
fis a diagonal
matrix formed of all the eigenvalues less than unity.
The RBC model : resolution method
Then the saddle path is determined by the set of restrictions : z
t,if= λ
f,iz
t+1,if+ φ
ft+1,i ∀iThe rational expectation of this equation leads to : z
t,if= λ
f,iE
th
z
t+1,ifi
=
⇒z
t,if= λ
Tf,iE
t hz
t+Tf ,i i∀i
As z
t,i1<
∞and lim
T→∞λ
Tf,i= 0, we get the saddle path : z
t,if= 0
⇐⇒q
1,k−1bk
t+ q
1,c−1bc
t+ q
1,a−1ba
t= 0 Then, using these restrictions, we get after substitution :
· b
k
t+1 ¸= A
· b
k
t ¸+ Bv
The RBC model : calibration
The unknown parameters are
Ψ
1=
{β, θ,A, α, δ, ρ, σ
²}dim(Ψ
1) = 7 the fraction of time spent working to N = 1/3, Aggregate data : K /Y = 10 and I/C = 0.25,
the capital’s share of output to α = 0.36
⇒ wNY= 1
−α, The following set of steady state restrictions
C =Y −I I =δK 1 =β
·
1−δ+αY K
¸
Y =AKαN1−α θ= (1−α)Y
NC
If we normalize Y = 1, these restrictions give
CY= 0.8,
I
= 0.2, A = 1.2, δ = 0.021 and β = 0.985.
The RBC model : the behaviors after a technological shock
1
After a technological shock, the labor productivity
↑ ⇒labor demand
↑.2
The production
↑because the technological frontier shift-up and because the total hours increase.
3
At equilibrium, this adjustment is possible if the raise in labor productivity (the real wage) gives the incentive to supply more labor.
4
This is possible if the shock is transitory : intertemporal substitution
⇔large opportunities today in the labor market, leading the agents to work more today than tomorrow when these opportunities disappear.
5
The saving opportunity allows the households to smooth these
The RBC model : the IRF
0 20 40 60
0 0.5 1 1.5 2
Produit
Trimestres
% Dev.
0 20 40 60
0.2 0.4 0.6 0.8 1
Consommation
Trimestres
% Dev.
0 20 40 60
−2 0 2 4 6
Investissement
Trimestres
% Dev.
0 20 40 60
0 0.5 1 1.5
Capital
Trimestres
% Dev.
0.5 1 1.5
Heures
% Dev.
0.6 0.8 1
Productivit‚
% Dev.
Tab.:US economy versus K/P/82
Correlation between GDP and Variablex σ x(t−1) x(t) x(t+ 1)
GDP 1,8% 0,82 1,00 0,82
Model 1,79% 0,60 1,00 0,60
Consumption
Services 0,6% 0,66 0,72 0,61
Non-durable 1,2% 0,71 0,76 0,59
Model 0,45% 0,47 0,85 0,71
Investment
Total investment 5,3% 0,78 0,89 0,78
Model 5,49% 0,52 0,88 0,78
Employment
Total hours 1,7% 0,57 0,85 0,89 Hour per capita 1,0% 0,76 0,85 0,61
Model 1,23% 0,52 0,95 0,55
Productivity 1,0% 0,51 0,34 -0,04
Model 0,71% 0,62 0,86 0,56
Part III
An extension of the RBC
model : Labor market frictions
RBC and labor market analysis
Puzzles
The magnitude of the employment fluctuations The dynamic of the real wage
The Persistence of the employment dynamics
Models
Employment lotteries : Hansen (1985), Rogerson (1988).
Efficiency wages : Danthine and Donalson (1990), Collard and De La Croix (2000).
Matching and wage bargaining : Merz (1995), Langot (1995 and 1996), Andolfatto (1996)
Endogenous job creation and job destruction : Den Han, Ramey and Watson (2000)
RBC and non-walrasian labor market
N
tis the number of jobs that are matched with a worker, U
t= 1
−N
tis the measure of unemployment rate.
Workers search passively : e is constant,
V
tis the total number of new jobs made available by firms, each of the vacancies incurs a flow cost equal to ω.
s is the exogeneous jobs destructions.
The rate M
tat which new job matches form is governed by an aggregate-matching technology.
The law of motion for aggregate employment is defined by :
N
t+1= (1
−s)N
t+ M
twith M
t= ΥV
tψ(eU
t)
1−ψRBC and non-walrasian labor market : the firms
All firms have access to the same technology : Y
t= A
t(h
tN
t)
αK
t1−αwhere Y
t, N
th
tand K
trespectively denote the output, the total hours, the capital stock. The stochastic process of the
technological shock, A
t, is :
log A
t= ρ
Alog A
t−1+ (1
−ρ
A) log A + ²
A,twhere
|ρA|< 1 and ²
A → N(0, σ
²A).
As firms are owned by households, they discount expected future
values at rate β
λλt+1.
RBC and non-walrasian labor market : the firms
Let
V(ΩFt) be the maximum expected value of the firm in state Ω
Ft. This value function must satisfy the following recursive relationship,
V
(Ω
Ft) = max
{Vt,It,Nt+1,Kt+1}
©
A
tK
tα(N
th
t)
1−α−w
th
tN
t−I
t−ωVt
+ βE
t·
λ
t+1λ
t V(ΩFt+1)
¸¾
s.t. N
t+1= (1
−s)N
t+ Φ
tV
t(3)
K
t+1= (1
−δ)K
t+ I
t(4)
where Ω
F=
{K, N , A , Φ , w , h
}summarizes each firm’s state.
RBC and non-walrasian labor market : the firms
The optimal decisions for the firm give conditions that govern the intertemporal allocation of investment in both physical capital and vacancies. They are given respectively by
λ
t= βE
t·
λ
t+1µ
α Y
t+1K
t+1+ 1
−δ
¶¸
ω
Φ
t= βE
t·
λ
t+1λ
tµ
F
2,t+1h
t+1−w
t+1h
t+1+ (1
−s )ω Φ
t+1¶¸
Φ
t=
MVttis the exogenous probability for each firm that a vacant
position becomes a filled position.
RBC and non-walrasian labor market : the firms
Given the production function, we have F
2,t ≡(1
−α)[Y
t/N
th
t]
Since the ratio of capital stock to labor input K
t/N
th
t ≡k
tis determined by the rental cost, r
t= αA
tk
tα, it is straightforward to deduce that
F
2,t= (1
−α)α
1−ααA
1 1−α
t
r
α α−1
t
It is assumed that the weight of a worker is ‘small’, F
2,tthus being
taken as given by the firm during the bargaining process.
RBC and non-walrasian labor market : insurance
There is a distinction between jobs and workers flows in the model :
H1)
jobs flows are governed by the matching/separation process,
H2)workers flows are determined exogenously by a game of
“musical chairs”.
⇒
At the beginning of each period, the entire workforce is
“shuffled” randomly across the given set of jobs (before any
trading occurs). There are N
tproductive jobs at the beginning of
period t, which represents the probability of employment for each
household in any period.
RBC and non-walrasian labor market : insurance
Risk-adverse workers choose to purchase B
tunits of
unemployment insurance at a price τ
t, that delivered B
tunits of the consumption good whenever she is unemployed within the period.
In a perfectly competitive unemployment insurance markets, the profit flow for an insurance company is given by :
Π
At= τ
tB
t−(1
−N
t) B
tAssuming zero profit condition for the unemployment
insurance company implies τ
t= 1
−N
t.
RBC and non-walrasian labor market : insurance
In each period the household chooses a level of insurance, B
t, and makes its consumption C
tzand saving choices q
t+1z, contingent upon its employment status (z = n, u ).
Let denoted the vector of control variables
Ct=
©B
t, C
tz, q
t+1z ªand Ω
H,ztthe vector of state variables, the problem faced by each household can be stated recursively by :
W(ΩHt) = max
Ct
n Nt
³
Utn+βEt
h
W(ΩH,nt+1) i´
+ (1−Nt)
³
Utu+βEt
h
W(ΩH,ut+1) i´o
RBC and non-walrasian labor market : insurance
This maximization is subject to the constraints referring to each state on the labor market :
Ctn+τtBt+ Z
pt+1qnt+1dAt+1 ≤ qt+πt+wtht (5) Ctu+τtBt+
Z
pt+1qut+1dAt+1 ≤ qt+πt+Bt (6)
where
Rp
t+1q
t+1zdA
t+1represents the value of contingent claims
at time t yielding q
t+1zunits of the consumption good at time
t + 1, in a given state A
t+1. π
trepresents pure profits due to the
non-walrasian structure of the labor market.
RBC and non-walrasian labor market : insurance
The first order conditions with respect to B
t, C
tz, and q
zt+1are respectively :
(1−Nt)λut(1−τt) = Ntλntτt (7)
∂Utz
∂Ctz = λzt z = n,u (8) β∂W(ΩH,zt+1)
∂qt+1z f(At+1|At) = λztpt+1 z = n,u (9)
The envelop theorem implies that :
∂W(Ω
Ht)
∂q
t= λ
t ∀z = n, u
Because τ
t= 1
−N
t, we have λ
t ≡λ
ut= λ
nt(Rational
RBC and non-walrasian labor market : insurance
Then, concavity and differentiability of the value function imply :
∂U
tn∂C
tn= ∂U
tu∂C
tu= λ
tand q
t+1n= q
t+1u= q
t+1This result shows that, given the complete set of insurance markets, the worker’s saving choice does not depend on its state on the labor market. We have
Ω
H,nt+1= Ω
H,ut+1= Ω
Ht+1RBC and non-walrasian labor market : insurance
Then, the difference between the two budget constraints (equations (5) and (6)) gives the optimal choice of insurance :
B
t= w
th
−(C
tn−C
tu)
Consequently, we get the following budget constraint for a representative household :
N
tC
tn+ (1
−N
t)C
tu+
Zp
t+1q
t+1dA
t+1= q
t+ π
t+ N
tw
th Remark that the equilibrium price of asset is given by :
β λ
t+1f (A
t+1|At) = p
t+1RBC and non-walrasian labor market : Workers
Then, the program solved by the representative household can be summarized by the following dynamic problem :
W(ΩHt) = max {Ctn,Ctu}
n
NtUtn+ (1−Nt)Utu+βEt
h
W(ΩHt+1) io
NtCtn + (1−Nt)Ctu=Ntwth+ Πt
Nt+1 = (1−s)Nt+ Ψt(1−Nt)
where Ψ
t= M
t/(1
−N
t), denotes the transition rate from E
→U.
In addition, we can abstract from the securities market as it plays no role in our analysis (given the representative framework, these claims are not traded in equilibrium). Then, let Π
tdenotes the financial income :
Πt = V(ΩFt)−βEt
·λt+1
λt V(ΩFt+1)
¸
Z
RBC and non-walrasian labor market : Consumption dynamics
Consequently, the worker’s optimal behavior is completely characterized by
λ
t= 1
C
tzz = n, u (10)
where λ
tis multiplier associated with the representative household’s budget constraint.
The optimal decision for the firm gives conditions that govern the intertemporal allocation of investment in physical capital. This decision gives the intertemporal allocation of consumption :
λ
t= βE
t[λ
t+1(1
−δ + r
t+1)] (11)
The Wage Setting Rule : Constant Hours per worker
The marginal value of employment for the representative household is given by (Ψ
t= M
t/(1
−N
t)) :
∂W (Ω
Ht)
∂N
t= λ
tB
t+ U
tn−U
tu+ (1
−s
−Ψ
t) βE
t"
∂W(Ω
Ht+1)
∂N
t+1#
B
t= w
th
−(C
tn−C
tu) refers to the optimal choice of unemployment insurance.
λ
tis the budget constraint Lagrange multiplier.
The utility function is given by (Andolfatto [AER,96]) : U
tz= log (C
tz) + Γ
z,twith
½
Γ
n,t= Γ(h
t)
Γ = Γ(e )
The Wage Setting Rule : Constant Hours per worker
At equilibrium, if the utility function is additively separable in consumption and leisure, we have :
C
tn= C
tuand U
tn−U
tu= Γ
n−Γ
uif Γ
n< Γ
u ⇔employed workers are worse off Then we have :
∂W (Ω
Ht)
∂N
t= λ
tw
th + (Γ
n−Γ
u) + (1
−s
−Ψ
t) βE
t"
∂W(Ω
Ht+1)
∂N
t+1#
(12)
The Wage Setting Rule : Constant Hours per worker
Symmetrically, for the firm this value is given by :
∂V(Ω
Ft)
∂N
t= F
2,th
−w
th + (1
−s)βE
t"
λ
t+1λ
t∂V(Ω
Ft+1)
∂N
t+1#
(13) From the firm’s decision problem, we have also :
βE
t"
λ
t+1λ
t∂V (Ω
Ft+1)
∂N
t+1#
= ω
Φ
t(14)
The Wage Setting Rule : Constant Hours per worker
We define the total marginal value of employment measured in units of the consumption good :
St= ∂V(ΩFt)
∂Nt
+∂W(ΩHt)/∂Nt
λt
(15)
Let 0 < ξ < 1 denotes the firm’s share of S
t. This sharing rule implies :
ξ∂W(ΩHt )/∂Nt
λt = (1−ξ)∂V(ΩFt)
∂Nt (16)
Thus, equations (16) and (14) yield :
ξβEt
·λt+1
λt
∂W(ΩHt+1)/∂Nt+1
λt+1
¸
= (1−ξ)Et
·λt+1
λt
∂V(ΩFt+1)
∂Nt+1
¸
= (1−ξ)ω
(17)
The Wage Setting Rule : Constant Hours per worker
Finally, combining conditions (12), (13), (16) and (17), the wage equation is given by :
wth= (1−ξ)
·
F2,th+ω Vt
1−Nt
¸ +ξ
·Γu−Γn
λt
¸
The real wage is a weighted average of
(i)
a procyclical component (labor productivity and average hiring costs)
(ii)
the outside option, which is a decreasing function of λ
t, with Γ
u> Γ
n.
If the wealth increases (⇔ ∆λ
t< 0)
⇒
the outside option increases. Thus, it clearly shows that
dynamics of the real wage is a combination of two procyclical
components, (i) labor productivity and hiring costs, and (ii) the
General Equilibrium
The general equilibrium is defined by the set of functions
{Ct
, V
t, K
t+1, N
t+1, w
t, M
t, Y
t}, solution of the following system :1
Ct = βEt
·µ
1−δ+αYt+1
Kt+1
¶ 1 Ct+1
¸
ωVt
Mt
= βEt
· Ct
Ct+1
µ
(1−α)Yt+1
Nt+1
−wt+1h+ (1−s)ωVt+1
Mt+1
¶¸
Kt+1 = (1−δ)Kt+Yt−Ct−ωVt
Nt+1 = (1−s)Nt+Mt
wth = (1−ξ)
· αYt
Nt +ω Vt
1−Nt
¸
+ξCt(Γu−Γn)
Mt = ΥVtψ(eUt)1−ψ Yt = AtKtα(hNt)1−α
log(At+1) = ρAlog(At) + (1−ρA) log(¯A) +²A,t
The LMS model : the IRF
0 10 20 30
0.5 1 1.5 2 2.5
OUTPUT (Y)
QUARTERS
%DEV.
LMS1 LMS2
0 10 20 30
0 2 4 6 8
INVESTMENT (I)
QUARTERS
%DEV.
−5 0 5 10 15 20
VACANCIES (V)
%DEV.
−0.5 0 0.5 1 1.5 2 2.5
TOTAL HOURS (Nh)
%DEV.
The LMS model : the IRF
0 10 20 30
0 0.2 0.4 0.6 0.8 1
LABOR PRODUCTIVITY (Y/Nh)
QUARTERS
%DEV.
LMS1 LMS2
0 10 20 30
0 0.2 0.4 0.6 0.8 1 1.2 1.4
WEALTH (1/λ)
QUARTERS
%DEV.
0.1 0.2 0.3 0.4 0.5 0.6 0.7
REAL WAGE (w)
%DEV.
−0.4
−0.3
−0.2
−0.1 0 0.1 0.2
LABOR SHARE (whN/Y)
%DEV.
US RBC LMS
σ(Y) 1.91 1.08 1.80
X (1) (2) (3) (1) (2) (3) (1) (2) (3)
C 0.43 0.82 0 0.26 0.90 0 0.60 0.86 +1
I 2.70 0.97 0 4.59 0.99 0 3.44 0.93 0
H 0.95 0.67 +1 0.26 0.99 0 0.93 0.95 +1 LP 0.54 0.53 -1 0.75 0.99 0 0.31 0.38 -1 W 0.48 0.43 -1 0.75 0.99 0 0.16 0.46 -1
WH/Y 0.52 -0.15 +4 0 0 0 0.16 -0.30 +3
corr(H,LP) 0.17 0.99 0.08
corr(H,W) 0.06 0.99 0.17
The Social Welfare Problem
In a labor market search model, are the fluctuation Pareto optimal, as in the basic RBC model ?
If no, structural policies can have an impact at business cycle frequencies.
The central planner solves the following problem, where
Ct=
{Ct, V
t, K
t+1, N
t+1}:
P
(Ω
t) = max
Ct
{log(Ct
) + N
tΓ
n+ (1
−N
t)Γ
u+ βE
t[P(Ω
t+1)]}
K
t+1= (1
−δ)K
t+ A
t(h
tN
t)
αK
t1−α−C
t−ωV
tN
t+1= (1
−s)N
t+ ΥV
tψ(1
−N
t)
1−ψThe Social Welfare Problem
The FOC and the envelope theorem give :
1 Ct
= βEt
·µ
1−δ+αYt+1
Kt+1
¶ 1 Ct+1
¸
ωVt
Mt = βEt
· Ct
Ct+1
µ
ψ(1−α)Yt+1
Nt+1 −ψ(Γu−Γn)Ct+1
−(1−ψ)ω Vt+1
1−Nt+1 + (1−s)ωVt+1
Mt+1
¶¸
Kt+1 = (1−δ)Kt+AtKtα(hNt)1−α−Ct−ωVt
Nt+1 = (1−s)Nt+ ΥVtψ(eUt)1−ψ log(At+1) = ρAlog(At) + (1−ρA) log(¯A) +²A,t
In order to have an Equilibrium Allocation which correspond to the
Pareto Optimal one, the wage must internalize the search
Pareto Optimal Allocation
The Equilibrium path of the labor market is such that
ωVt
Mt = βEt
· Ct
Ct+1
µ
(1−α)Yt+1
Nt+1 −wt+1h+ (1−s)ωVt+1
Mt+1
¶¸
wth = (1−ξ)
· αYt
Nt
+ω Vt
1−Nt
¸
+ξCt(Γu−Γn)
or, after substitution,
ωVt
Mt = βEt
· Ct
Ct+1
µ
ξ(1−α)Yt+1
Nt+1 −ξ(Γu−Γn)Ct+1
−(1−ξ)ω Vt+1
1−Nt+1 + (1−s)ωVt+1
Mt+1
¶¸
Pareto Optimal Allocation
The Optimal path of employment is such that
ωVt
Mt
= βEt
· Ct
Ct+1
µ
ψ(1−α)Yt+1
Nt+1
−ψ(Γu−Γn)Ct+1
−(1−ψ)ω Vt+1
1−Nt+1 + (1−s)ωVt+1
Mt+1
¶¸
whereas the equilibrium allocation solves :
ωVt
Mt
= βEt
· Ct
Ct+1
µ
ξ(1−α)Yt+1
Nt+1
−ξ(Γu−Γn)Ct+1
−(1−ξ)ω Vt+1
1−Nt+1 + (1−s)ωVt+1
Mt+1
¶¸
Thus Equilibrium level of employment, and of vacancies,
correspond to the Social-Optimal level iff ξ = ψ, the
Part IV
Real and nominal shocks : a
DSGE model
Unemployment and price rigidities : the Phillips and Beveridge curves
At each date t a final consumption good is produced by a perfectly competitive firm that combines a continuum of intermediate goods using a C.E.S. technology.
The demand for each intermediate good is given by :
Yj,td = µPj,t
Pt
¶−1+γ
γ Yt with Pt= µZ 1
0
Pj,t−γ1dj
¶−γ
, Yt= µZ 1
0
Yj,td 1+γ1 dj
¶1+γ
where γ
≥0, and
1+γγis the elasticity of substitution
between different goods.
Unemployment and price rigidities : the Phillips and Beveridge curves
The dynamic problem of each intermediate good firm is written recursively as follows :
W(Sj,tF ) = max
Cj,t
½Pj,t
Pt
³
At(hj,tNj,t)1−αKj,tα −F´
−wj,thj,tNj,t+ (1−δ)Kj,t
−Kj,t+1−φ
2 µ Pj,t
Pj,t−1
−1
¶2
−ωVj,t+βEt
· Ct
Ct+1
W(Sj,t+1F )
¸)
s.t. Yj,t ≤ Yj,td
Nj,t+1 = (1−s)Nj,t+ ΘtVj,t