On the Sources of the Great Moderation Jordi Gal and Luca Gambetti - PowerPoint PPT Presentation

On the Sources of the Great Moderation Jordi Galí and Luca Gambetti May 2008 Jordi Galí and Luca Gambetti () Great Moderation May 2008 1 / 11

The Great Moderation Basic Evidence (Table 1) Two Broad Hypotheses (i) good luck (ii) structural change (policy or non-policy related) ) di¤erent implications for second moments Our paper (i) evidence on changes in second moments of output, hours and labor productivity around the time of the volatility break (ii) identi…cation of the sources of those changes using time-varying SVAR ) time-varying conditional second moments and IRFs Such evidence may shed light on the merits of alternative explanations for the Great Moderation Jordi Galí and Luca Gambetti () Great Moderation May 2008 2 / 11

Figure 1 U.S. GDP Growth 4 3 2 1 0 -1 -2 -3 -4 1948 1953 1958 1963 1968 1973 1978 1983 1988 1993 1998 2003

The Great Moderation Basic Evidence (Table 1) Two Broad Hypotheses (i) good luck (ii) structural change (policy or non-policy related) ) di¤erent implications for second moments Our paper (i) evidence on changes in second moments of output, hours and labor productivity around the time of the volatility break (ii) identi…cation of the sources of those changes using time-varying SVAR ) time-varying conditional second moments and IRFs Such evidence may shed light on the merits of alternative explanations for the Great Moderation Jordi Galí and Luca Gambetti () Great Moderation May 2008 2 / 11

Our Approach Focus on output, hours and labor productivity Changes in unconditional second moments Identi…cation and estimation of conditional second moments and their changes over time Main tool: Time-varying VAR with stochastic volatility - Cogley-Sargent, Primiceri, Gambetti, Benati-Mumtaz - identi…cation based on Galí AER 99 technology and non-technology shocks technology shocks only source of unit root in labor productivity - extension to Fisher’s JPE 05 two technology shock model Jordi Galí and Luca Gambetti () Great Moderation May 2008 3 / 11

Main Findings Increase in the volatility of hours and labor productivity relative to that of output. Decline in the cyclicality of labor productivity (relative to both hours and output) Main source of output volatility decline: fall in contribution of non-technology shocks Large decline in the correlation of labor productivity with both output and hours conditional on non-technology shocks, accelerating in the 1990s. Large negative correlation of hours with both output and labor productivity conditional on technology shocks. Exception: the second half of the 1970s (oil shocks) and 1990s (the dotcom boom period). ) Picture more complex picture than suggested by good luck hypothesis ) Structural changes in labor market, timing close to GM. Causality? Jordi Galí and Luca Gambetti () Great Moderation May 2008 4 / 11

The Labor Market and the Great Moderation Focus on output, hours and labor productivity Quarterly U.S. data Nonfarm business sector Sample period: 1948:I-2005:IV Changes in Unconditional Volatilities (Table 2) Changes in Unconditional Comovements (Table 3) Jordi Galí and Luca Gambetti () Great Moderation May 2008 5 / 11

A Time-Varying Structural VAR Let x t � [ ∆ ( y t � n t ) , n t ] x t = A 0 , t + A 1 , t x t � 1 + A 2 , t x t � 2 + ... + A p , t x t � p + u t (1) where E t f u t u 0 t g = Σ t .and E t f u t x 0 t � k g = E t f u t u 0 t � k g = 0 for k = 1 , 2 , 3 , ... Let A t � [ A 0 , t , A 1 , t ..., A p , t ] and θ t � vec ( A 0 t ) , we assume θ t = θ t � 1 + ω t (2) where ω t � N ( 0 , Ω ) is serially uncorrelated and independent of f u t g . Jordi Galí and Luca Gambetti () Great Moderation May 2008 6 / 11

Let Σ t � F t D t F 0 where F t is lower triangular with ones on the diagonal t and D t is a diagonal matrix. De…ne γ t = vec ( F � 1 ) and σ t = vec ( D t ) . t We assume γ t = γ t � 1 + ζ t log σ t = log σ t � 1 + ξ t where ζ t � N ( 0 , Ψ ) and ξ t � N ( 0 , Ξ ) are serially and mutually uncorrelated. Jordi Galí and Luca Gambetti () Great Moderation May 2008 7 / 11

Identi…cation Structural shocks: ε t � [ ε a t , ε d t ] 0 , satisfying E f ε t ε 0 t g = I ε a t : technology shock ε d t : non-technology shock Assumption : u t = K t ε t for all t , for some non-singular matrix K t satisfying K t K 0 t = Σ t . Identifying restriction: only technology shocks have a long-run e¤ect on labor productivity Resulting decomposition: ∞ ∞ x i , t = µ i C ia t , k ε a C id t , k ε d ∑ ∑ t + t � k + t � k k = 0 k = 0 Jordi Galí and Luca Gambetti () Great Moderation May 2008 8 / 11

Changing Labor Market Dynamics and the Great Moderation Unconditional Second Moments (F2-3) Conditional Volatilities: What are the Forces behind the Great Moderation? (F4, T4) var ( ∆ y t ) = var ( ∆ y a t ) + var ( ∆ y d t ) Conditional Correlations and Structural Change (F5, T5) corr ( x t , z t ) = λ a corr a ( x t , z t ) + λ d corr d ( x t , z t ) where λ i � σ i ( x t ) σ i ( z t ) σ ( z t ) . σ ( x t ) Time-Varying Impulse Responses (F6-8) Extension: Fisher Three-Variable Model Jordi Galí and Luca Gambetti () Great Moderation May 2008 9 / 11

Figure 2a Time-Varying Standard Deviations 1.5 1.4 1.3 1.2 1.1 1 0.9 0.8 0.7 1965 1970 1975 1980 1985 1990 1995 2000 Output Hours Productivity

Figure 2b Time-Varying Relative Standard Deviations 0.9 0.85 0.8 0.75 0.7 0.65 0.6 1965 1970 1975 1980 1985 1990 1995 2000 Hours Productivity

Figure 3 Time-Varying Unconditional Correlations 1 2 0.5 1.5 0 1 -0.5 0.5 1960 1960 1965 1965 1970 1970 1975 1975 1980 1980 1985 1985 1990 1990 1995 1995 2000 2000 2005 2005 (n,y) (y-n,y) (y-n,n) s.d.(y)

Changing Labor Market Dynamics and the Great Moderation Unconditional Second Moments (F2-3) Conditional Volatilities: What are the Forces behind the Great Moderation? (F4, T4) var ( ∆ y t ) = var ( ∆ y a t ) + var ( ∆ y d t ) Conditional Correlations and Structural Change (F5, T5) corr ( x t , z t ) = λ a corr a ( x t , z t ) + λ d corr d ( x t , z t ) where λ i � σ i ( x t ) σ i ( z t ) σ ( z t ) . σ ( x t ) Time-Varying Impulse Responses (F6-8) Extension: Fisher Three-Variable Model Jordi Galí and Luca Gambetti () Great Moderation May 2008 9 / 11

Figure 4a Conditional Standard Deviations: Output 1.5 1.4 1.3 1.2 1.1 1 0.9 0.8 0.7 0.6 0.5 1965 1970 1975 1980 1985 1990 1995 2000 Technology Non-Technology Unconditional

Figure 4b Conditional Standard Deviations: Hours 1.1 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 1965 1970 1975 1980 1985 1990 1995 2000 Technology Non-Technology Unconditional

Figure 4c Conditional Standard Deviations: Labor Productivity 0.9 0.8 0.7 0.6 0.5 0.4 0.3 1965 1970 1975 1980 1985 1990 1995 2000 Technology Non-Technology Unconditional

Changing Labor Market Dynamics and the Great Moderation Unconditional Second Moments (F2-3) Conditional Volatilities: What are the Forces behind the Great Moderation? (F4, T4) var ( ∆ y t ) = var ( ∆ y a t ) + var ( ∆ y d t ) Conditional Correlations and Structural Change (F5, T5) corr ( x t , z t ) = λ a corr a ( x t , z t ) + λ d corr d ( x t , z t ) where λ i � σ i ( x t ) σ i ( z t ) σ ( z t ) . σ ( x t ) Time-Varying Impulse Responses (F6-8) Extension: Fisher Three-Variable Model Jordi Galí and Luca Gambetti () Great Moderation May 2008 9 / 11

Figure 5a Conditional Correlations: Hours - Output 1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 1965 1970 1975 1980 1985 1990 1995 2000 Technology Non-Technology Unconditional

Figure 5b Conditional Correlations: Labor Productivity - Hours 1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 1965 1970 1975 1980 1985 1990 1995 2000 Technology Non-Technology Unconditional

Figure 5c Conditional Correlations: Labor Productivity - Output 1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 1965 1970 1975 1980 1985 1990 1995 2000 Technology Non-Technology Unconditional

Changing Labor Market Dynamics and the Great Moderation Unconditional Second Moments (F2-3) Conditional Volatilities: What are the Forces behind the Great Moderation? (F4, T4) var ( ∆ y t ) = var ( ∆ y a t ) + var ( ∆ y d t ) Conditional Correlations and Structural Change (F5, T5) corr ( x t , z t ) = λ a corr a ( x t , z t ) + λ d corr d ( x t , z t ) where λ i � σ i ( x t ) σ i ( z t ) σ ( z t ) . σ ( x t ) Time-Varying Impulse Responses (F6-8) Extension: Fisher Three-Variable Model Jordi Galí and Luca Gambetti () Great Moderation May 2008 9 / 11

Figure 6a Non-Technology Shocks: Output Response

Figure 6c Figure 6b 0.2 1.2 0 1 -0.2 0.8 -0.4 -0.6 0.6 -0.8 0.4 -1 0.2 2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20 Pre-84 Post-84 minus Pre-84 Post-84

Figure 7a Non-Technology Shocks: Labor Productivity Response

Figure 7c Figure 7b 0.2 0.1 0.3 0 -0.1 0.2 -0.2 0.1 -0.3 -0.4 0 -0.5 -0.6 -0.1 -0.7 -0.8 -0.2 2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20 Pre-84 Post-84 minus Pre-84 Post-84

Figure 8a Technology Shocks: Hours Response