A Structural Investigation of Quantitative Easing (w. Felix - - PowerPoint PPT Presentation

A Structural Investigation of Quantitative Easing

(w. Felix Strobel)

Gregor Boehl

IMFS, Goethe University Frankfurt

MMCN Conference, June 2019

Note: All presented results are preliminary

Motivation

◮ Zero-lower bound (ZLB) on nominal interest rates ◮ Conventional monetary policy ineffective ◮ The effects of unconventional monetary policy at ZLB?

So far:

◮ VAR studies: QE affects financial variables ◮ DSGE studies: QE can affect real variables

Issue: Implied effects of QE depend on parameter choice ◮ Bayesian estimation necessary Problem: ZLB is a strong nonlinearity ◮ New methodology (Boehl, 2019)

Boehl & Strobel: A Structural Investigation of Quantitative Easing 1/6

Motivation

◮ Zero-lower bound (ZLB) on nominal interest rates ◮ Conventional monetary policy ineffective ◮ The effects of unconventional monetary policy at ZLB?

This work:

◮ Use estimated DSGE model to quantify QE: Smets & Wouters (2007) + Gertler & Karadi (2013) = NK + bells + whistles + banks + QE ◮ New methodology: OBCs solution & nonlinear filtering (Boehl, 2019)

◮ Bayesian estimation with endogenous ZLB

Boehl & Strobel: A Structural Investigation of Quantitative Easing 1/6

Key parameter estimates (incl. ZLB)

distribution mean sd mean sd hpd 2.5 hpd 97.5 ıp beta 0.500 0.15 0.340 0.084 0.174 0.492 ıw beta 0.500 0.15 0.445 0.142 0.167 0.733 ζp beta 0.500 0.10 0.805 0.044 0.727 0.887 ζw beta 0.500 0.10 0.680 0.058 0.555 0.783 Φp normal 1.250 0.12 1.305 0.119 1.062 1.516 LEV normal 3.000 1.00 1.802 0.457 1.131 2.616 θ beta 0.950 0.05 0.908 0.074 0.762 0.994 λcbl uniform 0.000 10.00 2.694 0.842 0.995 4.134 ρ beta 0.700 0.20 0.784 0.040 0.697 0.849 ρu beta 0.500 0.20 0.766 0.051 0.667 0.862 ρr beta 0.700 0.20 0.488 0.092 0.332 0.683 ρg beta 0.500 0.20 0.838 0.101 0.661 0.983 ρi beta 0.500 0.20 0.816 0.069 0.692 0.944 ρz beta 0.500 0.20 0.583 0.193 0.214 0.888 ρp beta 0.700 0.20 0.260 0.053 0.158 0.363 ρw beta 0.700 0.20 0.455 0.094 0.314 0.660 ρcbl beta 0.500 0.20 0.555 0.064 0.439 0.670 ρqeb beta 0.500 0.20 0.863 0.041 0.778 0.945 ρqek beta 0.500 0.20 0.921 0.033 0.857 0.980 . . . . . . . . . . . . . . . . . . . . . . . .

Boehl & Strobel: A Structural Investigation of Quantitative Easing 2/6

Historical decomposition I.

more ZLB effects no banks

2000 2008 2016 GDP −5

total u non-MP MP (all)

2000 2008 2016 C −5 2000 2008 2016 I −20 2000 2008 2016 inflation −0.5 0.0 0.5

Boehl & Strobel: A Structural Investigation of Quantitative Easing 3/6

Historical decomposition II.

more ZLB effects no banks

2000 2008 2016 nominal interest −1.0 −0.5 0.0 0.5 2000 2008 2016 bond price −10 10 2000 2008 2016 bank equity −40 −20

total u non-MP MP (all)

2000 2008 2016 premium −2.5 0.0 2.5 5.0

Boehl & Strobel: A Structural Investigation of Quantitative Easing 4/6

The net effects of QE Measures I

IRFs QE vs. FG

2008 2016 GDP −0.5 0.0 0.5 2008 2016 C 0.0 0.2 0.4 0.6 2008 2016 I −5.0 −2.5 0.0 2.5

total liquidity

• gov. bonds
• priv. bonds

2008 2016 inflation 0.000 0.025 0.050

Boehl & Strobel: A Structural Investigation of Quantitative Easing 5/6

The net effects of QE Measures II

IRFs QE vs. FG

2008 2016 nominal interest 0.00 0.02 0.04

total liquidity

• gov. bonds
• priv. bonds

2008 2016 bond price 2 2008 2016 bank equity −5 2008 2016 premium −2 −1

Boehl & Strobel: A Structural Investigation of Quantitative Easing 6/6

Thank you for your attention

Key Contributions

◮ Short run: (moderate) stimulating effect of QE ◮ Long run: QE strong recessionary risk ◮ Private assets purchases more effective than gov. bond purchases ◮ Technical challenge: nonlinear estimation (ZLB) of large-scale model

Boehl & Strobel: A Structural Investigation of Quantitative Easing 6/6

Historical decomposition

back

2000 2008 2016 nominal interest −1.0 −0.5 0.0 0.5 2000 2008 2016 bond price −10 10 2000 2008 2016 bank equity −40 −20

total u non-MP MP (all)

2000 2008 2016 premium −2.5 0.0 2.5 5.0

Boehl & Strobel: A Structural Investigation of Quantitative Easing 1/0

Historical decomposition

back

2000 2004 2008 2012 2016

kb

15 10 5 5 total u non-MP MP (all) 2000 2004 2008 2012 2016

l

10 5 5 2000 2004 2008 2012 2016

q

15 10 5 5 10 2000 2004 2008 2012 2016

premk

2 2 4

Boehl & Strobel: A Structural Investigation of Quantitative Easing 2/0

Historical decomposition

back

2000 2004 2008 2012 2016

z

0.2 0.1 0.0 0.1 2000 2004 2008 2012 2016

u

2 2 4 2000 2004 2008 2012 2016

g

1.5 1.0 0.5 0.0 0.5 2000 2004 2008 2012 2016

eps_i

0.5 0.0 0.5

Boehl & Strobel: A Structural Investigation of Quantitative Easing 3/0

QE vs. FG I

back

2000 2004 2008 2012 2016

y

1 1 2 total QE eps_r 2000 2004 2008 2012 2016

c

0.0 0.5 1.0 1.5 2000 2004 2008 2012 2016

i

4 2 2 4 2000 2004 2008 2012 2016

l

1 1 2

Boehl & Strobel: A Structural Investigation of Quantitative Easing 4/0

QE vs. FG II

back

2000 2004 2008 2012 2016

r

0.4 0.2 0.0 0.2 2000 2004 2008 2012 2016

rr

0.6 0.4 0.2 0.0 0.2 2000 2004 2008 2012 2016

Pi

0.00 0.05 0.10 2000 2004 2008 2012 2016

w

0.0 0.5 1.0

Boehl & Strobel: A Structural Investigation of Quantitative Easing 5/0

QE vs. FG III

back

2000 2004 2008 2012 2016

kb

0.0 0.5 1.0 1.5 2.0 total QE eps_r 2000 2004 2008 2012 2016

n

5 5 2000 2004 2008 2012 2016

q

1 1 2 3 4 2000 2004 2008 2012 2016

premk

2.0 1.5 1.0 0.5 0.0

Boehl & Strobel: A Structural Investigation of Quantitative Easing 6/0

IRF to a shock to CB capital purchases

back

10 20 30

n

2.5 0.0 2.5 5.0 7.5 10 20 30

premk

2.0 1.5 1.0 0.5 0.0 10 20 30

q

2 4 6 10 20 30

i

2 2 4 6 10 20 30

y

0.2 0.0 0.2 0.4 0.6 10 20 30

eps_qe_k_GDP

0.0 2.5 5.0 7.5 10.0

Boehl & Strobel: A Structural Investigation of Quantitative Easing 7/0

IRF to a shock to CB gov. bond purchases

back

10 20 30

n

2 1 1 2 3 10 20 30

premk

1.0 0.8 0.6 0.4 0.2 0.0 10 20 30

q

0.5 0.0 0.5 1.0 1.5 2.0 10 20 30

i

2 1 1 10 20 30

y

0.2 0.1 0.0 0.1 10 20 30

eps_qe_b_GDP

2 4 6 8 10

Boehl & Strobel: A Structural Investigation of Quantitative Easing 8/0

Parameter estimates SW07 model

go back

Prior Posterior mean sd mean sd hpd 2.5 hpd 97.5 ζp 0.500 0.10 0.758 0.044 0.664 0.835 ζw 0.500 0.10 0.630 0.052 0.533 0.732 Φ 1.250 0.12 1.803 0.081 1.658 1.973 φπ 1.700 0.25 1.517 0.257 1.033 2.018 φy 0.125 0.05 0.199 0.032 0.141 0.263 φdy 0.125 0.05 0.181 0.044 0.092 0.265 ρ 0.700 0.20 0.790 0.047 0.699 0.879 ρr 0.700 0.20 0.688 0.114 0.518 0.897 ρi 0.500 0.20 0.782 0.139 0.413 0.964 ρz 0.500 0.20 0.722 0.165 0.345 0.934 ρu 0.500 0.20 0.761 0.044 0.653 0.839 ρp 0.700 0.20 0.341 0.091 0.181 0.515 ρw 0.700 0.20 0.300 0.056 0.192 0.405 σu 0.100 2.00 1.769 0.429 0.925 2.620 σz 0.100 2.00 0.214 0.131 0.058 0.505 σr 0.100 2.00 0.150 0.078 0.076 0.245 σi 0.100 2.00 0.323 0.300 0.119 1.039 σp 0.100 2.00 0.284 0.104 0.112 0.496 σw 0.100 2.00 1.482 0.313 0.848 2.059

Boehl & Strobel: A Structural Investigation of Quantitative Easing 9/0

Historical decomposition (no Banks) I

back

2000 2004 2008 2012 2016

c

6 4 2 2 2000 2004 2008 2012 2016

y

7.5 5.0 2.5 0.0 2.5 2000 2004 2008 2012 2016

i

20 10 10 total e_u e_i e_z e_r e_g e_p e_w 2000 2004 2008 2012 2016

l

10 5 5

Boehl & Strobel: A Structural Investigation of Quantitative Easing 10/0

Historical decomposition (no Banks) II

back

2000 2004 2008 2012 2016

Pi

0.50 0.25 0.00 0.25 0.50 2000 2004 2008 2012 2016

kb

10 5 5 total e_u e_i e_z e_r e_g e_p e_w 2000 2004 2008 2012 2016

w

5.0 2.5 0.0 2.5 5.0 2000 2004 2008 2012 2016

r

1.0 0.5 0.0

Boehl & Strobel: A Structural Investigation of Quantitative Easing 11/0

Historical decomposition (no Banks) III

back

2000 2004 2008 2012 2016

eps_i

0.75 0.50 0.25 0.00 0.25 0.50 2000 2004 2008 2012 2016

eps_r

0.4 0.3 0.2 0.1 0.0 0.1 2000 2004 2008 2012 2016

eps_p

0.2 0.1 0.0 0.1 0.2 2000 2004 2008 2012 2016

eps_w

0.5 0.0 0.5 1.0

Boehl & Strobel: A Structural Investigation of Quantitative Easing 12/0

Historical decomposition (no Banks) IV

back

2000 2004 2008 2012 2016

y_gap

7.5 5.0 2.5 0.0 2.5 total e_u e_i e_z e_r e_g e_p e_w 2000 2004 2008 2012 2016

z

0.2 0.1 0.0 0.1 0.2 2000 2004 2008 2012 2016

u

2 2 4 6 8 2000 2004 2008 2012 2016

g

2 1

Boehl & Strobel: A Structural Investigation of Quantitative Easing 13/0

(Counterfactual) Effects of hitting the ZLB I

back

2008 2012 2016

c

0.0 0.5 1.0 1.5 2008 2012 2016

y

0.0 0.5 1.0 1.5 2.0 2008 2012 2016

r

0.8 0.6 0.4 0.2 0.0 2008 2012 2016

Pi

0.00 0.05 0.10 0.15

Boehl & Strobel: A Structural Investigation of Quantitative Easing 14/0

(Counterfactual) Effects of hitting the ZLB II

back

2008 2012 2016

kb

0.0 0.5 1.0 1.5 2.0 2008 2012 2016

l

0.0 0.5 1.0 1.5 2.0 2008 2012 2016

i

2 4 2008 2012 2016

w

0.0 0.5 1.0 1.5

Boehl & Strobel: A Structural Investigation of Quantitative Easing 15/0

(Counterfactual) Effects of MP / Forward Guidance

back

2008 2012 2016

y

0.5 0.0 0.5 1.0 1.5 2008 2012 2016

l

0.5 0.0 0.5 1.0 1.5 2.0 2008 2012 2016

r

0.6 0.4 0.2 0.0 2008 2012 2016

Pi

0.0 0.1 0.2

Boehl & Strobel: A Structural Investigation of Quantitative Easing 16/0

Smoothing & Path-adjustment (Boehl, 2019)

go back

◮ Smoothing (Transposed-Ensemble Rauch-Tung-Striebel smoother): Xt|T = Xt|t + ¯ Xt|t ¯ X

+ t+1|t

• Xt+1|T − Xt+1|t
• (1)

◮ Extraction (Iterative path-adjustmend):

◮ Fully reflects the nonlinearity of the transition function ◮ Interested in shocks {εt}T −1

t=0 that fully recover smoothened states

(historical decomposition!) ◮ Initialize ˆ x0 = EX0|T , define Pt|T = Cov{Xt|T }. ◮ For each t: ˆ εt = arg max

ε

• log fN
• g(ˆ

xt−1, ε)|¯ xt|T , Pt|T

• ,

(2) ˆ xt =g(ˆ xt−1, ˆ εt), (3)

Boehl & Strobel: A Structural Investigation of Quantitative Easing 17/0

The (flat) Phillips Curve

Prior Posterior mean sd mean sd hpd 2.5 hpd 97.5 ıp 0.500 0.15 0.247 0.094 0.072 0.419 ıw 0.500 0.15 0.452 0.143 0.196 0.723 ζp 0.500 0.10 0.758 0.044 0.664 0.835 Φ 1.250 0.12 1.803 0.081 1.658 1.973

πt = β 1 + ıpβ Etπt+1 + κˆ xt + ıp 1 + ıpβ πt−1 κ = (1 − ζpβ)(1 − ζp) (1 + βıp)ζp(ǫp(Φ − 1) + 1) (slope of PC) ˆ xt =wt − zt + α(lt − kt) (marginal costs) ◮ SW07: κ ≈ 0.02 ◮ here: κ ≈ 0.007 ! ◮ Key-incredient of NK model?

Boehl & Strobel: A Structural Investigation of Quantitative Easing 18/0

Equilibrium conditions

Definition (transition equilibrium)

A rational expectation solution S(l∗, k∗) is a rational expectations equilibrium iff bLs(l∗, k∗) ≥ ¯ r ∀s < l∗ ∧ s ≥ k∗ + l∗ (4) and bLs(l∗, k∗) < ¯ r ∀l∗ ≤ s < k∗ + l∗. (5)

go back Boehl & Strobel: A Structural Investigation of Quantitative Easing 19/0

Linearized equilibrium

back

ct = 1/γ (1 + 1/γ)ct−1 + 1 1 + 1/γ Et[ct+1] − (1 − 1/γ) (1 + 1/γ)σc (rt − Et[πt+1] + vd,t) it = 1 1 + β [it−1] + β 1 + β Et[it+1] + 1 (1 + β)γ2S′′ qk

t

kt =(1 − δ)/γkt−1 + (1 − (1 − δ)/γ)ˆ it + (1 − (1 − δ)/γ)(1 + β)γ2S′′vi,t Rt − Et[πt+1] + vd,t = Rk Rk + (1 − δ)Et[rk

t+1] +

(1 − δ) Rk + (1 − δ)Et[qk

t+1] − qk t

kt =1 − ψ ψ rk

t + kt−1

kt =wt − rk

t + lt

yt =Φ(αkt + (1 − α)lt + zt)

Boehl & Strobel: A Structural Investigation of Quantitative Easing 20/0

Linearized equilibrium II

back

yt =G Y gt + C Y ct + I Y it + RkK Y 1 − ψ ψ rk

t

πt = β 1 + ıpβ Etπt+1 + ıp 1 + ıpβ πt−1 + (1 − ζpβ)(1 − ζp) (1 + βıp)ζp((Φ − 1)ǫp + 1)(wt − zt + αlt − αkt) wt = 1 1 + βγ (wt−1 + ıwπt−1) + βγ 1 + βγ Et[wt+1 + πt+1] − 1 + ıwβγ 1 + βγ πt + (1 − ζwβγ)(1 − ζw) (1 + βγ)ζw((λw − 1)ǫw + 1)(wh

t − wt)

wh

t =

σc (1 − h)(ct − hct−1) + L 1 − Llt rt =max{0, ρrt−1 + (1 − ρ)(φππt + φy yt + φdy( yt − yt−1)) + vrt}

Boehl & Strobel: A Structural Investigation of Quantitative Easing 21/0

Linearized equilibrium III

back

vd,t =ρdvd,t−1 + ǫd

t ,

zt =ρzzt−1 + ǫz

t ,

gt =ρggt−1 + ǫg

t ,

vr,t =ρrvr,t−1 + ǫr

t,

vi,t =ρivi,t−1 + ǫi

t,

vp,t =ρpvp,t−1 + ǫp

t ,

vw,t =ρwvw,t−1 + ǫw

t ,

Boehl & Strobel: A Structural Investigation of Quantitative Easing 22/0

Households

Ut = E0

∞

• t=0

βt (Ct − hCt−1)1−σc − 1 1 − σc − ν log(1 − Lt)

• PtCt +

Dt vd,tRt = Dt−1 + WtLt − Tt + Pt. ◮ Ct - consumption ◮ Lt - labor ◮ Wt - real wage ◮ Dt - bank deposits ◮ vd,t - risk premium shock ◮ Rt - real rate on deposits ◮ Tt - lump sum taxes ◮ Pt- profits from firms and banks

back Boehl & Strobel: A Structural Investigation of Quantitative Easing 23/0

Unions

max

Wt(i) Et ∞

• s=0

(βζw)s Λt,t+s Πt,t+s

• Wt(i)Πs

l=1(Πıw t+l−1Π1−ıw) − MRSt+s

• Lt+s(i)

s.t.Lt+s(i) Lt+s = G

′−1

w

• Wt(i)Πs

l=1(Πıw t+l−1Π1−ıw)

Wt+s τ w

t+s

• .

Wt(i) - wage set by union i; ζw - Calvo parameter; Λt,t+s - SDF; ıw - the degree of wage indexation; MRSt - mraginal rate of substitution; Gw - Kimball aggregator.

Wt = [(1 − ζw)(W ∗

t )G

′−1

w

• W ∗

t τ w t

Wt

• +

ζwΠıw

t−1Π(1−ıw))Wt−1G

′−1

w

Πıw

t−1Π(1−ıw))Wt−1τ w t

Wt

• ,

W ∗

t - optimal wage

back Boehl & Strobel: A Structural Investigation of Quantitative Easing 24/0

Firms

◮ Intermediate good producers

details

◮ Cobb Douglas production function; employ labor and capital ◮ perfect competition ◮ buy and re-sell entire capital stock each period ◮ capital purchases are financed with bank loans

◮ Capital good producers

details

◮ perfect competition ◮ buys and re-sells capital to intermediate good producer ◮ repairs used capital, invests in new capital ◮ subject to investment adjustment costs.

◮ Retailers

details

◮ monopolistic competition, Calvo pricing

Boehl & Strobel: A Structural Investigation of Quantitative Easing 25/0

Intermediate goods producers

max

Kt,Lt,Ut Et[βΛt,t+1(−Rk,t+1QtKt(i) + Pm,t+1(i)Ym,t+1(i) − Wt+1Lt+1(i)...

... − a(Ut)Kt(i) + (1 − δ)Qt+1Kt(i))] s.t. Ym,t(i) = eztKt(i)α γtLt(i) 1−α − γtΦ, (6) Ymt - intermediate good; Pmt - price of intermediate good; zt - technology shock; δt - depreciation rate; Kt - physical capital stock; Kt - effective capital; Ut - utilization rate; Qt - price of capital; Rk,t+1

• real return of capital; Φ - fixed cost; γ - growth trend;

back Boehl & Strobel: A Structural Investigation of Quantitative Easing 26/0

Capital Goods Producers

Capital accumulation Kt = (1 − δ)Kt−1 + vi,t

• 1 − S

It It−1

• It,

Objective of Capital Good producer max

It

E0

∞

• t=0

βtΛ0,t

• Qt
• 1 − S

It(k) It−1(k)

• vi,t − 1
• It.

First-order condition for optimal investment: 1 = Qtvi,t

• 1 − S

It It−1

• − S′

It It−1 It It−1

• +Et
• Qt+1vi,t+1S′

It+1 It It+1 It 2 ◮ It - Investment; vi,t - investment specific technology shock

back Boehl & Strobel: A Structural Investigation of Quantitative Easing 27/0

Retailers

max

Pt(i) Et ∞

• s=0

(βζp)s Λt,t+s Πt,t+s

• Pt(i)Πs

l=1(Πıp t+l−1Π1−ıp) − MCt+s

• Yt+s(i)

s.t. Yt+s(i) Yt+s = G

′−1

• Pt(i)Πs

l=1(Πıp t+l−1Π1−ıp)

Pt+s τt+s

• .

Aggregate price index

Pt = [(1 − ζp)(P ∗

t )G

′−1

• P ∗

t τt

Pt

• + ζpΠ

ıp t−1Π(1−ıp)Pt−1G

′−1

Π

ıp t−1Π(1−ıp))Pt−1τt

Pt

• Pt(i) - price set by firm i; Πt,t+s - gross inflation, Λt,t+s -SDF, Yt -

demand for intermediate goods; MCt - marginal cost; ıp - degree of price indexation; G - Kimball aggregator; P ∗

t - optimal price;

back Boehl & Strobel: A Structural Investigation of Quantitative Easing 28/0

Banks

Banks’ balance sheet QtKb,t + Qb

tBb,t = Dt + Nt

Law of motion of net worth Nt = Rk,tQt−1Kb,t−1 + Rb,tQb

t−1Bb,t−1 − vd,t−1Rt−1Dt−1

◮ Qt, Qb

t - prices of capital asset and government bonds

◮ Kb,t - claims on capital stock held by banks ◮ Bb,t - government bond held by banks ◮ Dt deposits ◮ Nt - net worth ◮ Rk,t, Rb,t, Rt - interest rates on capital, bonds and deposits ◮ vd,t - risk premium shock (AR(1)-process)

back Boehl & Strobel: A Structural Investigation of Quantitative Easing 29/0

Banks

Each period a fraction of bankers, (1-θ), exits the business with a fixed probability. When they exit they consume their accumulated net

• worth. Hence, bankers maximize the terminal value of their net worth

Vt = max

{Kb,t},{Bb,t},{Dt}

EtΛt,t+1[(1 − θ)Nt+1 + θVt+1], subject to an incentive constraint. Vt ≥ λQtKb,t + λbQb

tBb,t.

◮ λ, λb - fraction of assets that banker can divert Assumption: Incentive constraint is always binding

back Boehl & Strobel: A Structural Investigation of Quantitative Easing 30/0

Banks - First Order Conditions

Guess: value function is linear in loans, government bonds and net worth: Vt = νktQtKt + νbtQb

tBt + νntNt

FOC for Kt, Bt. µt: νkt = λ µt 1 + µt (7) νbt = λb µt 1 + µt (8) QtKt = νbt − λb (λ − νkt)Qb

tBt +

νnt λ − νkt Nt (9) ◮ µt - Lagrange Multiplier of the incentive constraint

back Boehl & Strobel: A Structural Investigation of Quantitative Easing 31/0

Solution to the Bank’s Problem

Guess: value function is linear in loans, government bonds and net worth: Vt = νktQtKt + νbtQb

tBt + νntNt

solution for the coefficients: νk,t = βEtΩt+1(Rk,t+1 − vd,tRt), (10) νb,t = βEtΩt+1(Rb,t+1 − vd,tRt), (11) νn,t = βEtΩt+1vd,tRt. (12) where the stochastic discount factor of the banker is defined as: Ωt ≡ Λt−1,t[(1 − θ) + θ (νnt(1 + µt))] (13) with Λt−1,t being the SDF of the household

back Boehl & Strobel: A Structural Investigation of Quantitative Easing 32/0

Banks - Aggregation

Each period a fraction of bankers, (1-θ), exits the business with a fixed probability, and is replaced by new bankers, which are given a fraction ω of the total assets. net worth by existing and new bankers: Nt = Nnt + Net (14) Net = θ[(RktQt−1Kt−1 + RbtQb

t−1Bt−1 − vd,t−1Rt−1Dt−1]

(15) Nnt = ω[(RktQt−1Kt−1 + RbtQb

t−1Bt−1]

(16) aggregate balance sheet QtKt + Qb

tBt = Dt + Nt

(17) leverage ratio: φt = (QtKt + Qb

tBt)/Nt

(18)

back Boehl & Strobel: A Structural Investigation of Quantitative Easing 33/0

Fiscal Policy

budget constraint Gt + Rb,tQb

t−1Bt−1 = Qb tBt + Tt

government spending Gt = G · egt government spending gt = ρggt−1 + εg

t

tax revenues Tt = T + κb(Bt−1 − B) return on bonds Rbt = rc+ρcQb

t

Qb

t−1

Gt - government spending; Tt - tax revenues; gt - government spending shock; rc - coupon on bond; ρc - decay rate of consol;

back Boehl & Strobel: A Structural Investigation of Quantitative Easing 34/0

Data and Filtered series I

go back

2000 2004 2008 2012 2016

GDP

2 1 1 2000 2004 2008 2012 2016

Cons

3 2 1 1 2000 2004 2008 2012 2016

Inv

7.5 5.0 2.5 0.0 2.5 2000 2004 2008 2012 2016

Lab

5 10

Observables 1

Boehl & Strobel: A Structural Investigation of Quantitative Easing 35/0

Data and Filtered series II

go back

2000 2004 2008 2012 2016

Infl

0.25 0.00 0.25 0.50 0.75 1.00 2000 2004 2008 2012 2016

Wage

2 2 2000 2004 2008 2012 2016

FFR

0.0 0.5 1.0 1.5 2000 2004 2008 2012 2016

CBL

2 4 6 8

Observables 2

2000 2004 2008 2012 2016

CB_Loans

5 10 2000 2004 2008 2012 2016

CB_Bonds

5 10 15

Observables 3

Boehl & Strobel: A Structural Investigation of Quantitative Easing 36/0

Calibrated parameter

trend 0.344 pre-crisis average meanL 6.5415 pre-crisis average meanΠ 0.5 2% inflation target meanLSAP,B 5.65 pre-crisis average meanLSAP,K pre-crisis average meanCBL pre-crisis average λw 1.1 10% markup in labor market ǫ 10 as in SW (2007) h 0.72 as in SW (2007) α 0.19 as in SW (2007) ψ 0.79 mean value for trial estimations

go back Boehl & Strobel: A Structural Investigation of Quantitative Easing 37/0