Macro II

Professor Griffy

UAlbany

Spring 2025
Introduction

I Today: the Hosios Condition

I Efficiency in search and matching models.
I (Note: largely derived from Christine Braun’s lecture on the
DMP model).
I Homework on my website.
I Code for Aiyagari w/ labor-leisure choice on the cluster.
Efficiency

I Is zero unemployment efficient? No

I higher unemployment incentivizes firms to post vacancies

I but high unemployment is costly, less production

I Is a high vacancy rate efficient?

I vacancy creation is costly

I but lots of vacancies reduces unemployment

I So what is the efficient level of ✓?

Efficiency

I Congestion externality

I one more hiring firm makes unemployed workers better off and
makes all other hiring firms worse off

I one more searching worker makes hiring firms better off and
makes all other searching workers worse off

I Appropriability

I firm pays a cost  to post vacancy but does not get to keep
the entire output p
Efficiency

I What value of ✓ would the social planer choose to maximize

total output/utility if he is constrained by the same matching
frictions?

I does not care about wage b/c it’s a linear transfer from the
firm to the worker

I Does there exist a wage such that job creation is the same in
the decentralized equilibrium as in the social planners
outcome?

I Can we achieve this wage with the Nash solution?

The DMP Model (“Ch. 1 of Pissarides (2000)”)
I Agents:
1. Employed workers;
2. unemployed workers;
3. vacant firms;
4. matched firms.
I Linear utility (u = b, u = w) and production y = p > b.
I Matching function:
1. Constant returns to scale (L is lab. force):
v
M(uL, vL) = uL ⇥ M(1, ) = uL ⇥ p(✓)
u
2. where ✓ = vu is “labor market tightness”
3. Match rates:

p(✓) = ✓ q(✓)
|{z} |{z}
Worker Firm

I Social planner: pick ✓ optimally, no need to respect free entry

condition.
Social Planner’s Problem

Z 1
rt
e [p(1 u) + bu ✓u] dt
0
s.t. u̇ = (1 u) p(✓)u
I

separation rate

I Social planner’s problem

I p(1 u): social output of employment

I bu: leisure enjoyed by unemployed workers

I ✓u: cost of jobs

I Social planner is subject to the same transition equation for

unemployment
Social Planner’s Problem

I The Hamiltonian
rt
H=e [p(1 u) + bu ✓u] + µ(t)[ (1 u) p(✓)u]

*
I FOCs
↓ k
! U
PEO O

rt
Hu = µ̇ + r µ ) e (p b + ✓) [ + r + p(✓)]µ + µ̇ = 0
rt
H✓ = 0 ) e u µu(q(✓) + ✓q 0 (✓)) = 0

I µ: marginal value of an extra unemployed worker.

Social Planner’s Problem
I Optimal ✓

rt ✓q 0 (✓)
H✓ = 0 ) e u µuq(✓)(1 + )=0
q(✓)
✓q 0 (✓)
I What is q(✓) ?

m(u, v) = vq(✓)
@m(u, v) v
! = vq 0 (✓) 2
@u u
@m(u, v)
! = ✓2 q 0 (✓)
@u
✓2 q 0 (✓)
@m(u,v)
@u
! =
m(u, v) vq(✓)
@m(u,v)
✓q 0 (✓)
!u @u
=
m(u, v) q(✓)
✓q 0 (✓)
I q(✓) is the elasticity of the matching function wrt u.
Social Planner’s Problem

I The Hamiltonian
rt
H=e [p(1 u) + bu ✓u] + µ(t)[ (1 u) p(✓)u]

I FOCs
rt
Hu = µ̇ + r µ ) e (p b + ✓) [ + r + p(✓)]µ + µ̇ = 0
rt
H✓ = 0 ) e u µuq(✓)(1 ⌘(✓)) = 0

I ⌘(✓): elasticity of match fun. wrt u.

Optimal ✓ -

r+ p b + Ro
-

Hu = -

C
6+ r+ P(0)

I Using p(✓) = ✓q(✓) and solving in steady state (µ̇ = 0):

p b + ✓ 
=
+ r + p(✓) q(✓)(1 ⌘(✓))
p(✓) ( + r + p(✓))
(p b)(1 ⌘(✓)) + (1 ⌘(✓)) =
q(✓) q(✓)

+ r + ⌘(✓)p(✓)
! (1 ⌘(✓))(p b) =0 (1)
q(✓)
I This is optimal ✓
Decentralized solution

I Can the decentralized solution achieve the same level of ✓?

I i.e., can the decentralized level of unemployment be efficient?
Decentralized ✓
I Free entry V = 0:

rJ(w) = (p w) + [V J(w)]
(r + )J(w) = (p w)

I Vacancy creation condition (i.e., free entry imposed):


q(✓) =
E [J(w)]
(r + )
q(✓) =
(p w)
(r + )
✓ = q 1( )
(p w)
I Thus, mapping between wages and ✓. 1 equation, 2
unknowns.
I Need equation to determine wages in equilibrium.
Wage Determination
I Recall Nash Bargained wages:

w = argmaxw (W (w) U) (J(w) V )1

| {z }| {z }
Net Utility Net Profits

1 @W
0 = (W (w) U) (J(w) V )1
@w
@J
+ (1 )(J(w) V) (W (w) U)
@w
I @W
@w = 1, @J
@w = 1:

J(w) W (w) U
( )1 = (1 )( )
W (w) U J(w)
(J(w) + W (w) U) = W (w) U
S(w) = W (w) U
Wage Determination
I Note that S(w) = [W (w) U]

(1 )(w b) = (p w J(w))
+ (1 )(p(✓) + ) S(w)

J(w)
I And (1 )S(w) = J(w) ! S(w) = 1

(1 )(w b) = (p w J(w))
J(w)
+ (1 )(p(✓) + )
1
w = (1 )b + p + p(✓) J(w)

I Free entry condition: q(✓) = 

J(w) ! p(✓) = ✓
J(w)

w = (1 )b + p + ✓
Decentralized free entry

I Job creation curve:

(r + )J(w) = (p w)

q(✓) =
J(w)
(r + )
q(✓) =
(p w)
(r + )
p w =0
q(✓)
I Now, plug in using wages we just found:

w = (1 )b + p + ✓
Decentralized free entry

I Job creation curve:

(r + )
p ((1 )b + p + ✓) =0
q(✓)
p(✓)
I identities: p(✓) = ✓q(✓) ! ✓ = q(✓)

p(✓) (r + )
!p ((1 )b + p + ) =0
q(✓) q(✓)
p(✓) (r + )
(1 )(p b) ) =0
q(✓) q(✓)
r + + p(✓)
(1 )(p b) =0
q(✓)
I Looks familiar?
Social Planner’s Problem

I Using p(✓) = ✓q(✓) and solving in steady state (µ̇ = 0)

+ r + ⌘(✓)p(✓)
(1 ⌘(✓))(p b) =0 (2)
q(✓)

I From the decentralized solution, plug the wage curve into the
Job creation curve
+ r + p(✓)
(1 )(p b) =0 (3)
q(✓)
Efficiency

I Comparing (1) and (2) we see that we have efficiency in the

decentralized market if = ⌘(✓). The workers bargaining
power is equal to the elasticity of the matching function with
respect to u.

I This is a general result: we have efficiency when

⌘(✓) =

I This is called the Hosios (1990) condition

A couple more classes

I Two more classes, likely Directed Search and Block Recursive

Equilibrium.
Cats!