Banque nationale de Belgique SA · EURONEXT:BNB
Banque nationale de Belgique SA — presentatie antonella trigari
40 slides from Banque nationale de Belgique SA's investor presentation, March 2008
Slide 1: An Estimated Monetary DSGE Model with Unemployment and Staggered Nash Wage Bargaining
An Estimated Monetary DSGE Model with Unemployment and Staggered Nash Wage Bargaining Mark Gertler Luca Sala Antonella Trigari New York University IGIER, Università Bocconi March 2008
Slide 2: What we do
What we do 1. Estimate variant of a conventional monetary DSGE model as in Christiano et al. (CEE, 2005) and Smets and Wouters (SW, 2007) 2. Allow for unemployment via a variant of the Diamond-Mortensen-Pissarides (DMP) search and matching framework 3. Add wage rigidity via staggered Nash bargaining
Slide 3: Why we do it
Why we do it Existing quantitative models • Employment adjusts along the intensive margin + sticky wages • Wage rigidity — is key in accounting for labor market volatility and, also — influences short run output/inflation tradeoff (Blanchard and Galı́, 2006) • Subject to the Barro’s (1977) critique:
Slide 4: In our model
In our model • As employment adjusts along the extensive margin: — address the Barro’s critique — consistent with evidence • Address debate over whether the DMP framework can account for labor market volatility absent wage rigidity (Shimer, 2005, Hall, 2005, Hagedorn/Manovskii, 2006) — Estimate the
Slide 5: Model
Model • As in CEE and SW, basic non-labor market features: price stickiness, habit formation in consumption, investment adjustment costs, variable capital utilization • Unemployment and staggered wage bargaining as in GT, but: wage contracting in nominal terms with indexing for past inflation • Whol
Slide 6: Main results
Main results Wage rigidity improves the quantitative performance More work needed to ensure robust identi cation of key labor market parameters Similar t of the data as the SW model
Slide 7: Households
Households ∞ X Et β sεbt+s log (ct+s − hct+s−1) s=0 • Budget constraint: Bt k p p Bt−1 ct+it+ = wtnt+(1 − nt) bt+rt ztkt−1+Πt−Tt−a (zt) kt−1+ ptrt pt µ µ ¶¶ p p • Capital accumulation: kt = (1 − δ) kt−1 + εit 1 − s i it it t−1 p • Effective capital: kt = ztkt−1
Slide 8: Unemployment, vacancies and matching
Unemployment, vacancies and matching • Each firm i employs nt(i) workers and post vt(i) vacancies R1 R1 • nt = 0 nt(i)di, vt = 0 vt(i)di and ut = 1 − nt−1 1−σ mt mt • mt = σ muσ v t t , qt = and st = vt ut • Exogenous separation 1 − ρ
Slide 9: Firms: setup
Firms: setup κt Ft(i) = pt yt(i)−wt(i)nt(i)− xt(i)2nt−1(i)−rtk kt(i)+βEtΛt,t+1Ft+1(i) w 2 • Technology: yt (i) = [kt (i)]α [atnt (i)]1−α with at/at−1 = εat • Workforce dynamics: nt (i) = (ρ + xt (i)) nt−1 (i) qtvt (i) • Hiring rate: xt (i) = nt−1 (i)
Slide 10: Firms: rental capital decision
Firms: rental capital decision k w yt (i) yt rt = pt α =α kt (i) kt
Slide 11: Firms: hiring decision
Firms: hiring decision κt+1 w κtxt(i) = pt fnt−wt(i)+βEtΛt,t+1 xt+1(i)2+ρβEtΛt,t+1κt+1xt+1(i) 2 with yt (i) yt fnt (i) = (1 − α) = (1 − α) = fnt nt (i) nt
Slide 12: Staggered Nash bargaining: basics
Staggered Nash bargaining: basics • Each period only a subset of firms/workers negotiate a wage contract • Each firm negotiates with its existing workforce including new hires • Workers hired in-between contract settlements receive existing wage • Form of the contract: fixed nominal wage per period
Slide 13: • Tractable generalization of the period-by-period Nash bargaining • Differences from conventional time-dependent stag…
• Tractable generalization of the period-by-period Nash bargaining • Differences from conventional time-dependent staggered wage setting — No unexploited bilateral gains from renegotiating the wage — General-equilibrium spillovers of average wages on contract wages ⇒ extra stickiness in addition to t
Slide 14: Staggered Nash bargaining: the problem
Staggered Nash bargaining: the problem • For the fraction 1 − λ of firms that are renegotiating at t: wtn∗ = arg max Ht(i)ηt Jt (i)1−ηt η with η t = ηεt • For the fraction λ that are not: γ wtn (i) = γ̄wt−1 n (i) π t−1
Slide 15: Firm's marginal value of a worker (firm's surplus)
Firm’s marginal value of a worker (firm’s surplus) κt+1 2 nt+1 Jt (i) = pw f t nt −wt (i)−βE Λ t t,t+1 xt+1 (i) +Et βΛt,t+1 (i) Jt+1 (i) 2 nt
Slide 16: Worker's surplus
Worker’s surplus • Worker’s employment value Vt (i) = wt (i) + βEtΛt,t+1 [ρVt+1 (i) + (1 − ρ) Ut+1] • Worker’s unemployment value h i Ut = bt + βEtΛt,t+1 stVx,t+1 + (1 − st) Ut+1 Z 1 xt (i) nt−1 (i) Vx,t = Vt (i) di 0 xt−1nt−1 • Worker surplus ³ ´ Ht (i) = Vt (i) − Ut = wt (i) − bt + βEtΛt,t+1 ρHt+1
Slide 17: Staggered Nash bargaining: solution
Staggered Nash bargaining: solution • The Nash foc is χt (i) Jt (i) = (1 − χt (i)) Ht (i) with the “effective bargaining power” given by η χt (i) = η + (1 − η) Σt (i) /∆t and Σt (i) , ∆t ≡ firm and worker cumulative discount factors • “Horizon effect” ⇒ χ < η
Slide 18: • As in GT, the solution to the bargaining problem leads to: • Absent
• As in GT, the solution to the bargaining problem leads to: ∆twt∗ = wto (i) + ρλβEtΛt,t+1∆t+1wt+1 ∗ as in time-dependent contracting, wt∗ is a weighted sum of a future expected “target” • Absent the horizon effect, the “target” is the period-by-period Nash wage: µ ¶ κt+1 o w wt (i) = χ pt fnt + βEtΛ
Slide 19: Retail firms
Retail firms p • Monopolistically competitive, set prices á la Calvo. Price markup shocks εt . Monetary and fiscal policy s s • Taylor rule: rt/r = (rt−1/r)ρ [(π t/π)rπ (yt/ynt)ry ](1−ρ ) εrt • Government balances budget: gt + (1 − nt) bt + Bt−1/pt = Tt + Bt/ptrt ³ ´ g where gt = 1 − 1/εt yt
Slide 20: Aggregate wage dynamics
Aggregate wage dynamics • The aggregate wage is given by: b t = γ b (w w b t−1 − π b t + γπ ba b t−1 − εt) +γ ow b to ³ ´ +γ f Et w b t+1 + π b t+1 − γ π bt + εa bt+1 b to = period-by-period Nash bargained wage economy-wide w γb + γo + γf = 1 as λ → 0, γ b, γ f → 0 and γ o → 1 • Tractable generaliza
Slide 21: Aggregate hiring dynamics
Aggregate hiring dynamics ³ ´ w b bt = γ fn pbt + fnt − γ w w b b t + γ λEtΛ bt+1 x t,t+1 + βEtx
Slide 22: Estimation strategy
Estimation strategy • Seven quarterly series as in SW: Yt, Ct, It, Wt/Pt, Nt, rt, π t • Sample period: 1960Q1-2005Q1 • One shock for each series • Log-linearize the model • Model-consistent detrending for non-stationary variables
Slide 23: • Labor market → 5 new parameters + λ • Calibrate 3 for which
• Labor market → 5 new parameters + λ • Calibrate 3 for which there is independent evidence: — separation rate: ρ = 0.895 — matches elasticity to unemployment: σ = 0.5 — job finding rate: s = 0.95 • Estimate — relative flow value of unemployment b̄ — bargaining power η — wage rigidity λ
Slide 24: • Bayesian methods (see, e.g., An and Schorfheide, 2006) • Combine the likelihood function
• Bayesian methods (see, e.g., An and Schorfheide, 2006) • Combine the likelihood function of the model with priors for the parameters to be estimated to obtain the posterior distribution • Let the data speak as much as possible
Slide 25: Table 1: Calibrated parameters
Table 1: Calibrated parameters β δ α ξ g/y σ ρ s 0.99 0.025 0.33 10 0.2 0.5 0.895 0.95
Slide 26: Table 2: Prior and posterior distribution of structural parameters
Table 2: Prior and posterior distribution of structural parameters Prior Posterior Parameter Distribution Max Mean 5% 95% λ wage stickiness Beta (0.75,0.1) 0.717 0.725 0.670 0.780 γ wage indexation Uniform (0,1) 0.816 0.815 0.713 0.918 b̄ unem. flow value Beta (0.5,0.1) 0.726 0.723 0.664 0.786 η bar
Slide 27: Table 4: Prior and posterior distribution of structural parameters, λ = 0
Table 4: Prior and posterior distribution of structural parameters, λ = 0 Prior Posterior Parameter Distribution Max Mean 5% 95% b̄ unem. flow value Beta (0.5,0.1) 0.983 0.982 0.975 0.987 η bargaining power Beta (0.5,0.1) 0.616 0.589 0.451 0.726 ψz utilization rate elast. Beta (0.5,0.1) 0.861 0.852
Slide 28: Table 6: Log Marginal Likelihood
Table 6: Log Marginal Likelihood Baseline Model Flex Wage Model -1191 -1234
Slide 29: Y(t) − Y(t+k) , Y(t) − W(t+k) , Y(t) − H(t+k) , Y(t) −
Y(t) − Y(t+k) , Y(t) − W(t+k) , Y(t) − H(t+k) , Y(t) − π(t+k) , Y(t) − i(t+k) , Y(t) − u(t+k) , 3 0 10 2 0.3 0 −0.1 0 0.2 2 −0.2 1 Y 0.1 −0.2 −10 1 −0.3 −0.4 −20 0 0 −0.6 0 W(t) − Y(t−k) , 5 0 W(t) − W(t+k) , 5 0 W(t) − H(t+k) , 5 0 W(t) − π(t+k) , 5 0 W(t) − i(t+k) , 5 0 W(t) − u(t+k) , 5 0.3 0.6 0
Slide 30: Y(t) − Y(t+k) , Y(t) − W(t+k) , Y(t) − H(t+k) , Y(t) −
Y(t) − Y(t+k) , Y(t) − W(t+k) , Y(t) − H(t+k) , Y(t) − π(t+k) , Y(t) − i(t+k) , Y(t) − u(t+k) , 4 10 0.4 0.5 0 1.5 8 1 0.2 2 6 0 Y 0.5 −50 4 0 0 0 2 −0.2 −0.5 −100 0 W(t) − Y(t−k) , 5 0 W(t) − W(t+k) , 5 0 W(t) − H(t+k) , 5 0 W(t) − π(t+k) , 5 0 W(t) − i(t+k) , 5 0 W(t) − u(t+k) , 5 5 0.3 0.3 1.5 4
Slide 31: Table 7: Variance decomposition for output at different horizons
Table 7: Variance decomposition for output (∆ log y ) at different horizons (in percentage) Horizons Shocks on impact 1 year 4 years long run a shock - technology 16.7 32.5 31.0 31.0 r shock - monetary 6.1 5.0 5.4 5.4 b shock - preferences 11.1 9.2 9.5 9.5 i shock - investment 54.8 41.9 42.4 42.4 g s
Slide 32: Data GST GST
Y(t) − Y(t+k) , Y(t) − W(t+k) , Y(t) − H(t+k) , Y(t) − π(t+k) , Y(t) − i(t+k) , 3 0 2 Data 0.3 0 GST 0.2 −0.1 GST 2 −0.2 Y 1 −0.2 SW 0.1 1 −0.4 SW 0 −0.3 0 −0.6 0 W(t)2− Y(t−k) , 4 0 W(t)2− W(t+k) , 4 0 W(t)2− H(t+k) , 4 0 W(t)2− π(t+k) , 4 0 W(t)2 − i(t+k) , 4 0.3 0.6 0 0.2 1 0 0.1 0.4 −0.1 W 0.5 −
Slide 33: Conclusions
Conclusions • Develop and estimate a quantitative macro model with — wage rigidity — efficient bargaining • Wage rigidity improves the quantitative performance • More work needed on identification of labor market parameters (more variables: unemployment, vacancies, employment?) • Similar fit as existi
Slide 34: Percent Percent
(a) Unemployment 10 Model Data 8 Percent 6 4 2 0 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 (b) Labor market tightness 100 50 0 Percent −50 −100 −150 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
Slide 35: Percent
10 Unemployment rate Natural rate 9 8 7 6 Percent 5 4 3 2 1 0 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
Slide 36: Percent Percent
(a) Unemployment gap 4 2 Percent 0 −2 −4 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 (b) Output gap 6 4 2 Percent 0 −2 −4 −6 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
Slide 37: Y C flex W
Y C 1.5 2.5 1.5 2 1 1 flex W 1.5 π 0.5 0.5 1 0 W 0.5 0 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 NT 1.5 1 0.02 1 0 0.5 0.5 −0.02 R 0 0 −0.04 −0.5 −0.06 −0.5 −1 −0.08 −1.5 −1 −0.1 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 a shock 0.05 1.5 0 1 −0.05 0.5 −0.1 −0.15 0 0 10 20 30 40 0 10 20 30 40
Slide 38: C Y I
C Y I 1 4 0.4 flex W 3 0.2 0.5 2 0 1 −0.2 0 0 −1 −0.4 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 W NT π 4 1 0.04 0.03 3 0.5 0.02 2 0.01 0 0 1 −0.01 0 −0.5 −0.02 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 R i shock 0.2 0.2 0.15 0.15 0.1 0.1 0.05 0.05 0 −0.05 0 0 10 20 30 40 0 10 20 30 40
Slide 39: Table 3: Prior and posterior distribution of shock processes
Table 3: Prior and posterior distribution of shock processes Prior Posterior Parameter Distribution Max Mean 5% 95% ρa Beta (0.5,0.2) 0.140 0.096 0.036 0.167 ρr Beta (0.5,0.2) 0.207 0.179 0.096 0.266 ρb Beta (0.5,0.2) 0.713 0.724 0.660 0.788 ρi Beta (0.5,0.2) 0.605 0.599 0.520 0.675 ρp Beta (0.5,0.2
Slide 40: Table 5: Prior and posterior distribution of shock processes, λ = 0
Table 5: Prior and posterior distribution of shock processes, λ = 0 Prior Posterior Parameter Distribution Max Mean 5% 95% ρa Beta (0.5,0.2) 0.287 0.282 0.193 0.378 ρr Beta (0.5,0.2) 0.249 0.250 0.162 0.334 ρb Beta (0.5,0.2) 0.351 0.363 0.231 0.507 ρi Beta (0.5,0.2) 0.865 0.852 0.812 0.891 ρp Beta (
Banque Nationale De Belgique — presentatie antonella trigari on 6ix