”Can News about the Future Drive the Business Cycle?” by Nir Jaimovich and Sergio Rebelo (2009), AER

Jan Hannes Lang EUI

September 2009

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

1 / 25

Outline

1

Overview and Intuition

2

Details of the model

3

Discussion

4

Appendix

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

2 / 25

Outline

1

Overview and Intuition

2

Details of the model

3

Discussion

4

Appendix

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

3 / 25

Economic motivation of the paper

Positive correlation between output, investment, consumption, hours Pigou: changes in expectations are important in business cycles US boom-bust of late 1990s is often cited as an example Intuition: High expectations of future productivity lead to a boom When expectations don’t materialize, a recession ensues

Result of business cycle research This mechanism of Expectations Driven Business Cycles does not work in most standard neoclassical growth models

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

4 / 25

The news shocks literature

Beaudry & Portier (2004, 2007) Most RBC models cannot produce boom-bust cycles due to news Higher future productivity makes agents wealthier causing consumption and leisure to increase and investment to fall They propose multi-sector models with cost complementarities Floden (2007) Proposes a model with vintage capital, variable capital utilization, and standard preferences This one-sector model can produce expectations driven business cycles

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

5 / 25

Overview of the paper Critical features of the model Adjustment costs to investment Induce immediate investment response to news

Variable capital utilization Makes output flexible in the short-run; pushes up labor’s marg. prod.

Weak short-run wealth effects on the labor supply Help to push up hours in response to news

Main results Positive co-comovement of aggregates to technology and news shocks Business cycle properties are robust to the timing of information Positive sectoral co-movement to technology and news shocks J.H. Lang (EUI)

News and Business Cycles

Sep 2009

6 / 25

Outline

1

Overview and Intuition

2

Details of the model

3

Discussion

4

Appendix

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

7 / 25

Description of the one-sector model

Utility:

θ 1−σ −1 P t (Ct −ψNt Xt ) U = E0 ∞ t=0 β 1−σ 1−γ Xt = Ctγ Xt−1

Output:

At (ut Kt )1−α Ntα = Ct + It /zt

Capital:

It )] + [1 − δ(ut )]Kt Kt+1 = It [1 − ϕ( It−1

Adjustment costs convex:

ϕ(1) = 0, ϕ0 (1) = 0, ϕ00 (1) > 0

Depreciation costs convex:

δ 0 (ut ) > 0, δ 00 (ut ) ≥ 0

KPR set γ = 1 GHH set γ = 0

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

8 / 25

Impulse responses to news shocks For certain parameter ranges, good news lead to booms

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

9 / 25

Main mechanisms in the model I

A future rise in At or zt implies higher future investment Adjustment costs make it optimal to smooth investment Hence, current investment already increases This lowers the cost of future investment Current capital becomes less valuable, as it is cheaper to replace it. It is then efficient to increase utilization This pushes up the marginal product of labor As long as the wealth effect is not too negative, hours will increase

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

10 / 25

Main mechanisms in the model II

Variable capital utilization Assume GHH preferences (γ = 0) with Xt = 1 The FOC for Ct is:

(Ct − ψNtθ )−σ = λt

The FOC for Nt is:

θψNtθ−1 = αAt (ut Kt )1−α Ntα−1

Assume ut = constant, then Nt does not change (See FOC Nt ) Hours and output stay constant, consumption rises, investment falls Variable capital utilization raises the marginal product of labor Incentive for hours to respond to news shock

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

11 / 25

Main mechanisms in the model III

Investment adjustment costs Nt only changes if capital utilization changes The FOC for ut is: If ut ↑ ⇒

(1 − α)At ut−α Kt1−α Ntα = ηt /λt δ 0 (ut )Kt

LHS ↓ and δ 0 (ut )Kt ↑ ⇒

ηt /λt ↓ required

Without adjustment costs ηt /λt = zt ; does not change with news Wealth effect on labor supply Higher future productivity pushes up lifetime wealth If there is a large negative wealth effect, then hours will tend to fall Wealth effect is zero with GHH and negative for KPR preferences

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

12 / 25

Wealth effects on the labor supply

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

13 / 25

Robustness analysis Varying utilization can be costly, labor elasticity can be low

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

14 / 25

Model simulation

Stochastic investment-specific technical progress zt follows a random walk: log (zt+1 ) = log (zt ) + εt+1 They look at three scenarios: no news, perfect news, noisy news Noisy news: agents receive a signal whether output growth at t+2 is going to be below or above average The signal is chosen to have the same precision as output forecasts in the Livingston survey Perfect signal: at t agents get to know the realization of the growth rate at t+2

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

15 / 25

Simulation results BC properties reasonably matched without technological regress

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

16 / 25

Recessions are driven by bad news about the future Fewer and more shallow recessions in the model

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

17 / 25

Outline

1

Overview and Intuition

2

Details of the model

3

Discussion

4

Appendix

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

18 / 25

(My) Conclusion I

Main contribution of Jaimovich & Rebelo (2009) Propose a one-sector model that produces BCs due to news Provide a fully calibrated version of the model Conduct model simulations that roughly match BC moments Model able to generate sectoral co-movement as well Properties are robust to informational assumptions Provides possible explanation for reduction in volatility and increase in persistence of output

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

19 / 25

(My) Conclusion II

News, and changes in expectations important for the BC The model is more tractable than Beaudry & Portier (2004, 2007) Mechanism needed that increases hours and investment Mechanisms that drive investment realistic Mechanism for labor maybe less realistic?

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

20 / 25

Some (provoking) questions for discussion

Have we learned anything about the mechanisms driving BCs? Are business cycles really optimal? Do news or behavioral biases cause expectational changes? Lucas (1987): if unemployment important for the BC, equilibrium models cannot help our understanding

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

21 / 25

Outline

1

Overview and Intuition

2

Details of the model

3

Discussion

4

Appendix

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

22 / 25

The two-sector model I

Consumption production: Ct = At ztc (utc Ktc )1−α (Ntc )α Investment production: Itc + Iti = At zti (uti Kti )1−α (Nti )α c

c Law of motion K c : Kt+1 = Itc [1 − ϕ( I cIt )] + [1 − δ(utc )]Ktc t−1

Law of motion

Ki:

Labor constraint:

J.H. Lang (EUI)

i Kt+1

Ntc

+

=

Nti

Iti [1

Iti

− ϕ( I i )] + [1 − δ(uti )]Kti t−1

= Nt

News and Business Cycles

Sep 2009

23 / 25

The two-sector model II

Can produce aggregate and sectoral co-movements due to contemporaneous and news shocks News shocks require low wealth effects (low γ) to generate sectoral co-movement Also elastic labor supply and low elasticity of utilization costs needed Low wealth effects are needed for labor co-movement Adjustment costs are needed for investment co-movement

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

24 / 25

FOCs

J.H. Lang (EUI)

News and Business Cycles

Sep 2009

25 / 25