What Shifts the Beveridge Curve?
Recruiting Intensity and Financial Shocks
Alessandro Gavazza
London School of Economics
Simon Mongey
New York University
Gianluca Violante
New York University
RBA Quantitative Macroeconomics Workshop
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 1 /42
The outward shift in the Beveridge curve
0.035
2001-2002
2003-2007
2008-2013
Vacancy rate
0.03
0.025
0.02
0.015
0.01
0.04
0.05
0.06
0.07
0.08
0.09
0.1
Unemployment rate
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 2 /42
The outward shift in the Beveridge curve
0.035
2001-2002
2003-2007
2008-2013
Vacancy rate
0.03
0.025
0.02
0.015
0.01
0.04
0.05
0.06
0.07
0.08
0.09
0.1
Unemployment rate
It indicates a deterioration in aggregate matching efficiency Φt
Ht = Φt Vtα Ut1−α
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 2 /42
M EASUREMENT
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
OF
Φt
p. 3 /42
Estimating Φt accounting for compositional changes
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 4 /42
Estimating Φt accounting for compositional changes
• Hall and Schulhofer-Wohl → among job-seekers, include
◮ Nonparticipants (Nt )
◮ Employed (Et )
Ht = Φt · 1 +
|
sN
t
1−α
Nt
E Et
+ st
Ut
Ut
{z
}
· Vtα U 1−α
composition factor
E
• Veracierto → estimate (sN
t , st ) through data on worker flows
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 4 /42
Estimating Φt accounting for compositional changes
• Hall and Schulhofer-Wohl → among job-seekers, include
◮ Nonparticipants (Nt )
◮ Employed (Et )
Ht = Φt · 1 +
|
sN
t
1−α
Nt
E Et
+ st
Ut
Ut
{z
}
· Vtα U 1−α
composition factor
E
• Veracierto → estimate (sN
t , st ) through data on worker flows
• Fujita and Moscarini → exclude workers on temporary layoff from
the matching function
• Fix α = 0.5, compute composition factor, and get Φt as a residual
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 4 /42
Measured drop in aggregate matching efficiency
Φ
t
Log deviations from 2001:1
0
Composition
−0.1
−0.2
−0.3
−0.4
−0.5
2002
2004
2006
2008
2010
2012
2014
Year
2001: −10 percent & fast rebound
2007-09: −30 percent & slow recovery
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 5 /42
Explaining the deterioration in matching efficiency
Ht = Φt Vtα Ut1−α
1. Mismatch ↑
◮ Sahin-Song-Topa-Violante 2014; Elsby-Michaels-Ratner 2014
2. Worker’s search effort ↓
◮ Mukoyama-Patterson-Sahin 2014; Hagedorn-Karahan-Manovskii-Mitman 2014
3. Firm’s recruiting intensity ↓
◮ Davis-Faberman-Haltiwanger 2012; Kaas-Kircher 2014
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 6 /42
Firms’ recruiting intensity
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 7 /42
Firms’ recruiting intensity
• Effective vacancies:
Vt∗ =
Z
eit vit di
• vit : max open positions ready to be staffed and costly to create
• eit ∈ [0, 1]: probability of filling an open position —an outcome of
how much firms choose to spend on recruitment activities
◮ advertisement, networking, screening, outsourcing, etc.
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 7 /42
Firms’ recruiting intensity
• Effective vacancies:
Vt∗ =
Z
eit vit di
• vit : max open positions ready to be staffed and costly to create
• eit ∈ [0, 1]: probability of filling an open position —an outcome of
how much firms choose to spend on recruitment activities
◮ advertisement, networking, screening, outsourcing, etc.
• Aggregate matching function:
α
Ht = (Vt∗ ) Ut1−α = Φt · Vtα Ut1−α
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
with Φt =
Z
eit
vit
Vt
α
di
p. 7 /42
M ECHANISM
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 8 /42
Mechanism
• Fact I: much of job creation is in young firms
◮ Haltiwanger-Jarmin-Miranda 2014
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 9 /42
Mechanism
• Fact I: much of job creation is in young firms
◮ Haltiwanger-Jarmin-Miranda 2014
• Fact II: job-filling rate is steeply increasing with firm’s growth rate
◮ Davis-Faberman-Haltiwanger 2012
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 9 /42
Mechanism
• Fact I: much of job creation is in young firms
◮ Haltiwanger-Jarmin-Miranda 2014
• Fact II: job-filling rate is steeply increasing with firm’s growth rate
◮ Davis-Faberman-Haltiwanger 2012
• Fact III: recent financial shock hit young firms the hardest
◮ Chodorow-Reich 2014; Siemer 2014; Adelino et al. 2013; Fort et al., 2013
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 9 /42
Mechanism
• Fact I: much of job creation is in young firms
◮ Haltiwanger-Jarmin-Miranda 2014
• Fact II: job-filling rate is steeply increasing with firm’s growth rate
◮ Davis-Faberman-Haltiwanger 2012
• Fact III: recent financial shock hit young firms the hardest
◮ Chodorow-Reich 2014; Siemer 2014; Adelino et al. 2013; Fort et al., 2013
• Financial shock : hits start-ups and young firms the most
hR
iα
◮ Recruiting intensity Φt =
eit vVitt di falls
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 9 /42
Mechanism
• Fact I: much of job creation is in young firms
◮ Haltiwanger-Jarmin-Miranda 2014
• Fact II: job-filling rate is steeply increasing with firm’s growth rate
◮ Davis-Faberman-Haltiwanger 2012
• Fact III: recent financial shock hit young firms the hardest
◮ Chodorow-Reich 2014; Siemer 2014; Adelino et al. 2013; Fort et al., 2013
• Financial shock : hits start-ups and young firms the most
hR
iα
◮ Recruiting intensity Φt =
eit vVitt di falls
• TFP shock: more neutral across firms, so smaller effect on Φt
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 9 /42
E CONOMIC E NVIRONMENT
RANDOM - MATCHING MODEL WITH MULTI - WORKER FIRMS
C OOPER -H ALTIWANGER -W ILLIS 07, E LSBY-M ICHAELS 13, ACEMOGLU -H AWKINS 14
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 10 /42
Cast of characters
1. Firms
• Operate a DRS technology y(z, n), z stochastic
• Hire in frictional labor markets: choose (e, v)
• Face non-negative dividend constraint → borrowing
• Endogenous entry and exit/default
◮
Sedlacek 14; Siemer 14; Schott 14
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 11 /42
Cast of characters
1. Firms
• Operate a DRS technology y(z, n), z stochastic
• Hire in frictional labor markets: choose (e, v)
• Face non-negative dividend constraint → borrowing
• Endogenous entry and exit/default
◮
Sedlacek 14; Siemer 14; Schott 14
2. Banks
• Price loans competitively reflecting default prob. (+ wedge)
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 11 /42
Cast of characters
1. Firms
• Operate a DRS technology y(z, n), z stochastic
• Hire in frictional labor markets: choose (e, v)
• Face non-negative dividend constraint → borrowing
• Endogenous entry and exit/default
◮
Sedlacek 14; Siemer 14; Schott 14
2. Banks
• Price loans competitively reflecting default prob. (+ wedge)
3. Households
• Risk neutral representative family (some members unempl.)
• Save into bank deposits and mutual fund that owns all firms
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 11 /42
Timeline
(n, a, z)
δ shock
Draw z
Exogenous
Exit
a′ = y(z, n′ ) − b′
n′
b′
w
Exit/Default
Hire
Borrow
V and U meet
Payment of
Production
Entry
Fire
Bargain
prod. costs
y(z, n′ )
and dividends
and consumption
Individual firm’s state variables:
• n: initial employment (pre-hiring)
• a: initial net worth
• z: productivity
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 12 /42
Entry, exit, hire/fire decisions
• Entry decision: λ0 potential entrants drawing z ∼ Γ0 (z)
◮ fraction ε of start-up cost χ0 financed by equity
−εχ0 + Vi (0, −(1 − ε)χ0 , z ∗ ) = 0
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 13 /42
Entry, exit, hire/fire decisions
• Entry decision: λ0 potential entrants drawing z ∼ Γ0 (z)
◮ fraction ε of start-up cost χ0 financed by equity
−εχ0 + Vi (0, −(1 − ε)χ0 , z ∗ ) = 0
• Stay as incumbent (and repay) or exit-repay or exit-default
V(n, a, z) =
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
max V (n, a, z), a, 0
i
p. 13 /42
Entry, exit, hire/fire decisions
• Entry decision: λ0 potential entrants drawing z ∼ Γ0 (z)
◮ fraction ε of start-up cost χ0 financed by equity
−εχ0 + Vi (0, −(1 − ε)χ0 , z ∗ ) = 0
• Stay as incumbent (and repay) or exit-repay or exit-default
V(n, a, z) =
max V (n, a, z), a, 0
i
• Incumbent: fire or hire
f
h
V (n, a, z) = max V (n, a, z), V (n, a, z)
i
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 13 /42
Incumbent firms’ decisions: fire
f
V (n, a, z)
=
f
d + β(1 − δ)
max
′ ′
n ,b
X
V(n′ , a′ , z ′ )Γ(z ′ , z)
z ′ ∈Z
s.t.
n′
≤
n
df
≡
a − w(z, n′ , b′ )n′ − χ + Q(n′ , b′ , z)b′ ≥ 0
a′
=
y(z, n′ ) − b′
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 14 /42
Incumbent firms’ decisions: fire
f
V (n, a, z)
=
f
d + β(1 − δ)
max
′ ′
n ,b
X
V(n′ , a′ , z ′ )Γ(z ′ , z)
z ′ ∈Z
s.t.
n′
≤
n
df
≡
a − w(z, n′ , b′ )n′ − χ + Q(n′ , b′ , z)b′ ≥ 0
a′
=
y(z, n′ ) − b′
Note: wage determined by Nash bargaining (Stole-Zwiebel solution)
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 14 /42
Incumbent firms’ decisions: hire
h
V (n, a, z) =
max
e∈[0,1],v>0,b
d
′
h
+
β(1 − δ)
X
V(n′ , a′ , z ′ )Γ(z ′ , z)
z ′ ∈Z
s.t.
n′ − n =
q(θ ∗ )ev
dh ≡ a − w(z, n′ , b′ )n′ − χ − C(e, v, n) + Q(n′ , b′ , z)b′ ≥ 0
a′
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
=
y(z, n′ ) − b′
p. 15 /42
Choice of functional form for C(e, v, n)
DFH: Log-linear relation btw job-filling rate q(θ ∗ )e and growth rate
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 16 /42
Choice of functional form for C(e, v, n)
DFH: Log-linear relation btw job-filling rate q(θ ∗ )e and growth rate
-1.0
-1.5
Log Daily Fill
Rate
-2.0
-2.5
Data points correspond to
growth rate bins
-3.0
-3.5
Hires-Weighted Least Squares
Slope (s.e.) = 0.820 (0.006)
R-squared = 0.993
-4.0
-4.5
-5.0
-5.0
Log Hires Rate
-4.0
-3.0
-2.0
-1.0
Reverse engineer C(·) that yields the above relationship
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 16 /42
Hiring problem
1. Stage I: Choose target employment level n′ > n
2. Stage II: Choose max new positions v, and recruitment effort e
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 17 /42
Hiring problem
1. Stage I: Choose target employment level n′ > n
2. Stage II: Choose max new positions v, and recruitment effort e
C ∗ (n, n′ )
=
s.t.
n′ − n =
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
min
e∈[0,1],v>0
κ2 v γ2
κ1 γ1
e +
v
γ1
γ2 + 1 n
q(θ ∗ )ev
p. 17 /42
Hiring problem
1. Stage I: Choose target employment level n′ > n
2. Stage II: Choose max new positions v, and recruitment effort e
C ∗ (n, n′ )
=
s.t.
n′ − n =
min
e∈[0,1],v>0
κ2 v γ2
κ1 γ1
e +
v
γ1
γ2 + 1 n
q(θ ∗ )ev
• The solution yields the job filling-rate:
′
h
γ2
n −n
log
log
≡ q(θ ∗ )e = Ω (κ1 , κ2 , θ ∗ ) +
v
γ1 + γ2
n
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 17 /42
Why is recruiting effort increasing in the growth rate?
v
n′ − n
= q(θ ∗ ) · e ·
n
n
1. e and v/n are both inputs in the production of employment growth
κ1 γ1
κ2
C
=
e +
v
γ1
γ2 + 1
v γ2 n
2. cost of creating a new position is increasing in both e and v/n
3. relative curvature of cost function C with respect to e and v/n
determines their elasticity with respect to the desired growth rate
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 18 /42
Banks
• Competitive sector with free entry
• Intermediate funds at cost ϕ > 0 (financial wedge)
• Pay risk-free return Q̄−1 = β −1 on deposits
• Upon firm’s default, i.e., xD (n′ , a′ , z ′ ) = 1, recover nothing
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 19 /42
Banks
• Competitive sector with free entry
• Intermediate funds at cost ϕ > 0 (financial wedge)
• Pay risk-free return Q̄−1 = β −1 on deposits
• Upon firm’s default, i.e., xD (n′ , a′ , z ′ ) = 1, recover nothing
• Equilibrium price of a loan to a firm of type (n′ , b′ , z):
′
′
"
Q(n , b , z) = Q̄(1 − ϕ)(1 − δ) 1 −
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
X
z ′ ∈Z
#
xD (·)Γ (z ′ , z)
p. 19 /42
(Inverse of) price of debt for start-ups
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 20 /42
Representative household
W(U, T, M )
=
s.t.
′
′
′
C
+
βW(U
,
T
,
M
)
max
′
′
M ,T
C + Q̄M ′ + P T ′
=
Z
U′
=
U + δ(1 − U ) − ΦV α U 1−α
w(z, n′ , b′ )n′ dλ + ωU + (D + P )M + T
• T : household deposits
• M : shares of the mutual fund owning all firms
• D: average dividends paid by firms
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 21 /42
P RELIMINARY
PARAMETERIZATION
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 22 /42
Externally calibrated
Parameter
Value
Target
Discount factor (monthly)
β
0.9967
Risk-free rate
Potential entrants
λ0
0.02
Meas. of incumbents = 1
Size of labor force
Λ
18.7
Average firm size = 17.5
Nash bargaining share of workers
η
0.5
—
Elasticity of matching function wrt Vt
α
0.5
Empirical estimates
Financial intermediation wedge (monthly)
ϕ
0.002
Excess bond premium
External equity share of start-up cost
ε
0.50
Kauffman Firm Survey
Model period is 1 month
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 23 /42
Externally calibrated
Parameter
Value
Target
Discount factor (monthly)
β
0.9967
Risk-free rate
Potential entrants
λ0
0.02
Meas. of incumbents = 1
Size of labor force
Λ
18.7
Average firm size = 17.5
Nash bargaining share of workers
η
0.5
—
Elasticity of matching function wrt Vt
α
0.5
Empirical estimates
Financial intermediation wedge (monthly)
ϕ
0.002
Excess bond premium
External equity share of start-up cost
ε
0.50
Kauffman Firm Survey
Model period is 1 month
Addition to the model: heterogeneity in DRS across firms
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 23 /42
Internally calibrated
Parameter
Value
Target
Model
Data
Flow of home production
ω
0.62
Monthly separ. rate
0.03
0.02
Scaling of match. funct.
Φ̄
0.42
Monhtly job finding rate
0.40
0.30
Midpoint DRS in prod.
ν
0.70
Dividend share
0.06
0.05
High-Low DRS in prod.
∆ν
0.20
Empl. share 500+
0.25
0.47
Persistence of z shocks
Γ
0.99
Annual |g| < 0.05
0.50
0.42
SD of z shocks
Γ
0.11
SD (g)
0.32
0.42
Cost elasticity wrt e
γ1
3.74
Recr. int. small/large firms
1.23
1.50
Cost elasticity wrt v
γ2
22.97
Elasticity of job. fill rate wrt g
0.86
0.82
Cost shifter wrt e
κ1
4.88
Hiring cost/monthly wage
0.03
0.15
Cost shifter wrt v
κ2
0.08
Vac. share. of small firms
0.18
0.34
Entry cost
χ0
0.28
Annual entry rate
0.09
0.11
(Exponential) distrib. of z0
ξ
11.82
Share of JC by entrants
0.33
0.32
Operating cost
χ
0.08
Survive ≥ 5 years
0.62
0.50
Exogenous exit shock
δ
0.008
Share of JD by exit
0.24
0.35
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 24 /42
“Up or out” dynamics of young firms
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 25 /42
“Up or out” dynamics of young firms
A. Cohort job creation and destruction
0.12
1. Job creation rate
2. Job destruction rate
0.10
B. Type of destruction
1.0
1. Job destruction by continuing firms
2. Job destruction due to exit
0.8
0.08
0.6
0.06
0.4
0.04
0.1
0.02
0
0
5
10
Age (years)
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
15
0
0
5
10
Age (years)
15
p. 25 /42
Age distribution of recruiting intensity and vacancies
0.04
Recruitment intensity
Vacant positions
0.03
0.02
0.01
0
0
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
5
Age (years)
10
15
p. 26 /42
Firms’ life-cycle (averages)
B. Growth Rate
A. Size
0.15
40
30
0.10
20
0.05
10
0
0
5
10
15
0
0
5
10
Age (years)
Age (years)
C. Recruitment Effort
D. Debt to Output Ratio
0.5
15
1.5
0.4
1.0
0.3
0.2
0.5
0.1
0
0
5
10
Age (years)
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
15
0
0
5
10
15
Age (years)
p. 27 /42
DATA
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
TO BE
E XPLAINED
p. 28 /42
Data: 2001 vs 2008
B. Data 2008:01−2014:01
1.0
0.8
0.8
0.6
0.6
Log deviation from date 0
Log deviation from date 0
A. Data 2001:01−2007:01
1.0
0.4
0.2
0
−0.2
−0.4
0.4
0.2
0
−0.2
−0.4
−0.6
−0.6
−0.8
−0.8
−1.0
0
Vacancies
1
2
3
Years
Vacancy yield
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
4
5
6
Unemployment
−1.0
0
1
Job finding rate
2
3
Years
4
5
6
Aggregate Recruitment Intensity
p. 29 /42
E XPERIMENTS
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 30 /42
Experiments
Trace transitional dynamics of the economy in response to:
• 2001 recession
◮ Aggregate productivity Z ↓ by 4% and recovers in 6 years
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 31 /42
Experiments
Trace transitional dynamics of the economy in response to:
• 2001 recession
◮ Aggregate productivity Z ↓ by 4% and recovers in 6 years
• 2007-09 recession
◮ Same aggregate productivity Z ↓ combined with:
◮ Financial shock
Financial wedge ϕ ↑ and recovers in 6 years
Initial equity at start-up ε ↓ and recovers in 6 years
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 31 /42
Model: 2001 vs. 2008
B. Model 2007:01−2014:01
1
0.8
0.8
0.6
0.6
Log deviation from date 0
Log deviation from date 0
A. Model 2001:01−2007:01
1
0.4
0.2
0
−0.2
−0.4
0.4
0.2
0
−0.2
−0.4
−0.6
−0.6
−0.8
−0.8
−1
0
1
Vacancies
Vacancies
2
3
Years
Vacancy
Vacancy
yield
yield
4
5
6
−1
0
1
2
3
Years
4
5
6
Unemployment
Unemployment Job finding
Job
rate
finding rate
Aggregate Recruitment
Aggregate Recruitment
Intensity
Intensity
Entry
A: TFP Shock
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 32 /42
Model: 2001 vs. 2008
B. Model 2007:01−2014:01
1
0.8
0.8
0.6
0.6
Log deviation from date 0
Log deviation from date 0
A. Model 2001:01−2007:01
1
0.4
0.2
0
−0.2
−0.4
0.4
0.2
0
−0.2
−0.4
−0.6
−0.6
−0.8
−0.8
−1
0
1
Vacancies
Vacancies
2
3
Years
Vacancy
Vacancy
yield
yield
4
5
6
−1
0
1
2
3
Years
4
5
6
Unemployment
Unemployment Job finding
Job
rate
finding rate
Aggregate Recruitment
Aggregate Recruitment
Intensity
Intensity
Entry
B: TFP Shock + increase in intermediation wedge ϕ
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 33 /42
Model: 2001 vs. 2008
A. Model 2001:01−2007:01
1
0.8
0.8
0.6
0.6
Log deviation from date 0
Log deviation from date 0
1
0.4
0.2
0
−0.2
−0.4
−0.6
Vacancies
0.4
0.2
0
−0.2
−0.4
−0.6
−0.8
−0.8
−1
0
B. Model 2008:01−2014:01
1
2
3
Years
Vacancy yield
4
5
6
Unemployment
−1
0
1
Job finding rate
2
3
Years
4
5
6
Aggregate Recruitment Intensity
A: TFP shock
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 34 /42
Model: 2001 vs. 2008
A. Model 2001:01−2007:01
1
0.8
0.8
0.6
0.6
Log deviation from date 0
Log deviation from date 0
1
0.4
0.2
0
−0.2
−0.4
−0.6
Vacancies
0.4
0.2
0
−0.2
−0.4
−0.6
−0.8
−0.8
−1
0
B. Model 2008:01−2014:01
1
2
3
Years
Vacancy yield
4
5
6
Unemployment
−1
0
1
Job finding rate
2
3
Years
4
5
6
Aggregate Recruitment Intensity
B: TFP shock + fall in share of entry cost financed by equity (ε)
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 35 /42
Model: 2001 vs. 2008
B. Entry 2008:01−2014:01
A. Entry 2001:01−2007:01
1
0.8
Data
0.8
0.6
Model
0.6
Log deviation from date 0
Log deviation from date 0
1
0.4
0.2
0
−0.2
−0.4
−0.6
−0.8
−1
0
Data
Model
0.4
0.2
0
−0.2
−0.4
−0.6
−0.8
1
2
3
Years
4
5
6
−1
0
1
2
3
Years
4
5
6
A: TFP Shock
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 36 /42
Model: 2001 vs. 2008
B. Entry 2008:01−2014:01
A. Entry 2001:01−2007:01
1
0.8
Data
0.8
0.6
Model
0.6
Log deviation from date 0
Log deviation from date 0
1
0.4
0.2
0
−0.2
−0.4
−0.6
−0.8
−1
0
Data
Model
0.4
0.2
0
−0.2
−0.4
−0.6
−0.8
1
2
3
Years
4
5
6
−1
0
1
2
3
Years
4
5
6
B: TFP shock + fall in share of entry cost financed by equity (ε)
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 37 /42
T HANK YOU !
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 38 /42
State-level regressions combining HWOL ads and QWI hires
Log Vacancy Yield
Log Vacancy Yield
Share of Hires - Firm Age 0-1
0.233***
(0.702)
1.838**
(0.908)
Share of Hires - Firm Age 2-3
0.417
(1.648)
0.355
(1.786)
Share of Hires - Firm Age 4-5
1.434
(1.835)
1.307
(1.883)
Share of Hires - Firm Age 6-10
-0.946
(1.312)
-1.090
(1.279)
Share of Hires - Firm Size 0-19
-0.395
(0.815)
0.231
(1.231)
Share of Hires - Firm Size 20-49
0.773
(1.900)
1.428
(2.087)
Share of Hires - Firm Size 250-499
-1.330
(1.648)
-1.815
(1.789)
Share of Hires - Firm Size 500+
0.244
(0.860)
0.469
(0.935)
Constant
0.369
(0.770)
0.183
(0.818)
Observations
1,606
1,606
R-squared
0.934
0.934
State FE
Yes
Yes
Quarter FE
Yes
Yes
Seasonally Adjusted
No
Yes
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 39 /42
The Christiano-Eichenbaum-Trabandt critique
Ut+1 = Ut − Φt Vtα Ut1−α + δ(1 − Ut )
• One can explain joint dynamics of {Ut , Vt } w/o any change in Φt
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 40 /42
The Christiano-Eichenbaum-Trabandt critique
Ut+1 = Ut − Φt Vtα Ut1−α + δ(1 − Ut )
• One can explain joint dynamics of {Ut , Vt } w/o any change in Φt
• Estimation yields {V̂t , Ĥt }, with Ĥt = V̂tα Ut1−α
• Look at model’s implications for:
1. Job-finding rate (Ĥt /Ut )
2. Vacancy yield (Ĥt /V̂t )
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 40 /42
The Christiano-Eichenbaum-Trabandt critique
• One can explain the “shift” without any change in Φt
0.035
Vacancies
0.03
Job−finding rate
Data
Φ constant
0.025
0.02
0.015
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.01
0.04
0.06
0.08
0.05
0.1
Unemployment
Data
Φ constant
2002 2004 2006 2008 2010 2012 2014
Year
1.2
1
1
0.8
0.6
0
−0.5
0.4
0.2
0.5
Log of Φ
Vacancy Yield
Φ
Data
Φ constant
2002 2004 2006 2008 2010 2012 2014
Year
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
−1
2002 2004 2006 2008 2010 2012 2014
Year
p. 41 /42
The Christiano-Eichenbaum-Trabandt critique
• One can explain the “shift” without any change in Φt
0.035
Vacancies
0.03
Job−finding rate
Data
Φ constant
0.025
0.02
0.015
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.01
0.04
0.06
0.08
0.05
0.1
Unemployment
Data
Φ constant
2002 2004 2006 2008 2010 2012 2014
Year
1.2
1
1
0.5
Log of Φ
Vacancy Yield
Φ
Data
Φ constant
0.8
0.6
−0.5
0.4
0.2
0
2002 2004 2006 2008 2010 2012 2014
Year
−1
2002 2004 2006 2008 2010 2012 2014
Year
• Fit for the vacancy yield is poor
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 41 /42
The Christiano-Eichenbaum-Trabandt critique
• With Φt time-varying:
0.035
Vacancies
0.03
Job−finding rate
Data
Φ varying
0.025
0.02
0.015
0.4
0.35
0.3
0.25
0.2
0.15
Data
Φ varying
0.1
0.01
0.04
0.06
0.08
0.05
0.1
2002 2004 2006 2008 2010 2012 2014
Unemployment
Year
1
0.2
0.8
0.6
Φ
0.1
Data
Φ varying
Log of Φ
Vacancy Yield
1.2
0
−0.1
−0.2
−0.3
0.4
0.2
−0.4
2002 2004 2006 2008 2010 2012 2014
−0.5
2002 2004 2006 2008 2010 2012 2014
Year
Year
• Fit for the vacancy yield much better
Gavazza-Mongey-Violante, ”What Shifts the Beveridge Curve?”
p. 42 /42