Equilibrium Unemployment
Transcription
Equilibrium Unemployment
Equilibrium Unemployment Joao Gomes, Jeremy Greenwood and Sergio Rebelo FACTS In the US economy the following variables move countercyclically: I Unemployment rate. FACTS In the US economy the following variables move countercyclically: I I Unemployment rate. Average duration of unemployment. FACTS In the US economy the following variables move countercyclically: I I I Unemployment rate. Average duration of unemployment. Flows into and out of unemployment. GOAL I To construct a general equilibrium model in which individual job opportunities are affected by both aggregate and idiosyncratic shocks and that is consistent with these key facts about unemployment. GOAL I To construct a general equilibrium model in which individual job opportunities are affected by both aggregate and idiosyncratic shocks and that is consistent with these key facts about unemployment. Why? GOAL I To construct a general equilibrium model in which individual job opportunities are affected by both aggregate and idiosyncratic shocks and that is consistent with these key facts about unemployment. Why? I Such a model would be an ideal laboratory to examine such questions as the impact of unemployment insurance and the cost of business cycles fluctuations. The Model (1) I Heterogeneous agent economy. The Model (1) I I Heterogeneous agent economy. Agents maximize expected value of lifetime utility: E0 ∞ X t=0 β t U (˜ ct − D (lt )) The Model (1) I I Heterogeneous agent economy. Agents maximize expected value of lifetime utility: E0 ∞ X β t U (˜ ct − D (lt )) t=0 I Employed agents pay taxes The Model (1) I I Heterogeneous agent economy. Agents maximize expected value of lifetime utility: E0 ∞ X β t U (˜ ct − D (lt )) t=0 I I Employed agents pay taxes Unemployed agents receive unemployment insurance and do not pay taxes The Model (1) I I Heterogeneous agent economy. Agents maximize expected value of lifetime utility: E0 ∞ X β t U (˜ ct − D (lt )) t=0 I I I Employed agents pay taxes Unemployed agents receive unemployment insurance and do not pay taxes Incomplete markets (there is only a riskfree bond) The Model (1) I I Heterogeneous agent economy. Agents maximize expected value of lifetime utility: E0 ∞ X β t U (˜ ct − D (lt )) t=0 I I I I Employed agents pay taxes Unemployed agents receive unemployment insurance and do not pay taxes Incomplete markets (there is only a riskfree bond) exogenous borrowing constraint The Model (2) I Every period an agent receives a job opportunity O (k, l; ε, λ) The Model (2) I I Every period an agent receives a job opportunity O (k, l; ε, λ) If he accepts it he earns labor income O (k, l; ε, λ) − (r + δ) k and pays taxes τ . Also ε0 is drawn from G (ε0 |ε) The Model (2) I I I Every period an agent receives a job opportunity O (k, l; ε, λ) If he accepts it he earns labor income O (k, l; ε, λ) − (r + δ) k and pays taxes τ . Also ε0 is drawn from G (ε0 |ε) If he rejects it he gets UI benefits µ and his next ε0 is drawn from H (ε0 ) The Problem of a Worker I A worker first chooses k and l as follows Y (ε, λ; Z ) = max [O (k, l; ε, λ) − (R(λ; Z ) + δ)k − d(l)] k,l The Problem of a Worker I A worker first chooses k and l as follows Y (ε, λ; Z ) = max [O (k, l; ε, λ) − (R(λ; Z ) + δ)k − d(l)] k,l The Choice problem of a worker is Z W (a, ε, λ; Z ) s.t c + a0 a0 Z0 = max {U (c) + β 0 c,a max [W (a0 , ε0 , λ0 ; Z 0 ) , S (a0 , λ0 ; Z 0 )] ×dG (ε0 |ε) dF (λ0 |λ) dε0 dλ0 } = Y (ε, λ; Z ) + [1 + R(λ; Z )] a − T (λ; Z ) , ≥ ¯a = TZ The Problem of a Searcher The Choice problem of a searcher is : Z max W a0 , ε0 , λ0 ; Z 0 , S a0 , ε0 ; Z 0 c,a ×dH (ε) dF λ0 |λ dε0 dλ0 } S (a, λ; Z ) = max0 {U (c) + β s.t c + a0 = [1 + R(λ : Z )] a + µ, a0 ≥ ¯a Z 0 = TZ The decision rule governing whether someone works or not is: Ω (a, ε, λ; Z ) = 1 if W (a, ε, λ; Z ) ≥ S (a, ε; Z ) 0 otherwise. The decision rule governing whether someone works or not is: Ω (a, ε, λ; Z ) = 1 if W (a, ε, λ; Z ) ≥ S (a, ε; Z ) 0 otherwise. The government maintains a balanced budget each period: Z µ Z [1 − Ω (a, ε, λ; Z )] dZ (a, ε) dadε = τ Ω (a, ε, λ; Z ) dZ (a, ε) dadε The decision rule governing whether someone works or not is: Ω (a, ε, λ; Z ) = 1 if W (a, ε, λ; Z ) ≥ S (a, ε; Z ) 0 otherwise. The government maintains a balanced budget each period: Z Z [1 − Ω (a, ε, λ; Z )] dZ (a, ε) dadε = τ µ Ω (a, ε, λ; Z ) dZ (a, ε) dadε The capital market clears Z Z K (ε, λ; Z ) Ω (a, ε, λ; Z ) dZ (a, ε) dadε = adZ (a, ε) dadε The law of motion for the economy-wide distribution of wealth, or Z’=TZ, is described by: 0 0 0 Z (a , ε ) = Z {I (A (a, ε, λ; Z ) − a0 ) [Ω (a, ε, λ; Z ) G (ε0 |ε) + (1 − Ω (a, ε, λ; Z )) dH (ε0 )]dZ (a, ε) dadε}, where I (x)=1 if x ≤ 0 and I (x)=0 otherwise. Calibration The instantaneous utility function is: 1−σ c˜ − l 1+θ / (1 + θ) −1 U (˜ c − D (l)) = , θ > 0, σ > 0 1−σ The production function is: O (k, l; ε, λ) = exp (λ + ε) k α l 1−α Income Process Calibration Aggregate shocks. λ0 = ρλ λ + ξ, ξ ∼ N 0, σλ2 Income Process Calibration Aggregate shocks. λ0 = ρλ λ + ξ, ξ ∼ N 0, σλ2 Idiosyncratic shocks. The worker shock ε evolves according to ε0 = ρε ε + η, η ∼ N 0, σε2 and the searcher draws a value of ε with ε = υ, υ ∼ N 0, συ2 Income Process Calibration Aggregate shocks. λ0 = ρλ λ + ξ, ξ ∼ N 0, σλ2 Idiosyncratic shocks. The worker shock ε evolves according to ε0 = ρε ε + η, η ∼ N 0, σε2 and the searcher draws a value of ε with ε = υ, υ ∼ N 0, συ2 Finally unemployment compensation is set to µ = ηy ∗ Calibration parameters Income dynamics Heaton and Lucas(1996) ln (yit /yit−1 ) = υ0 + υ1 ln (yit−1 /yit−2 ) + υ2 ln (yt /yt−1 ) + µit Hubbard et al’s (1995) ln (yit ) = υ1 ln (yit−1 ) + µit Model υ1 = 0.5 and σµi =0.19 vs Data υ1 = 0.95 and σµi =0.14 Comparative Statics Impulse response, positive shock Impulse response, positive shock Impulse response, positive shock Welfare Cost of business cycle fluctuations " Eb ∞ X t=0 # β t U (ct ) = En " ∞ X t=0 # β t U ($ct ) Welfare Cost of business cycle fluctuations " Eb ∞ X # β t U (ct ) = En t=0 ∞ X # β t U ($ct ) t=0 ( $= " P∞ t 1−σ )1/(1−σ) Eb t=0 β ct P∞ t 1−σ En t=0 β ct Welfare Cost of business cycle fluctuations " Eb ∞ X # β t U (ct ) = En t=0 ∞ X # β t U ($ct ) t=0 ( $= " P∞ t 1−σ )1/(1−σ) Eb t=0 β ct P∞ t 1−σ En t=0 β ct For the model economy $ − 1 = 0.0056. This implies that the agent prefers to live in an economy with aggregate shocks. Some Remarks I The model does not distinguish between unemployment due to quits and layoffs Some Remarks I I The model does not distinguish between unemployment due to quits and layoffs Flows between employment and nonparticipation are as large as flows between employment and unemployment. Some Remarks I I I The model does not distinguish between unemployment due to quits and layoffs Flows between employment and nonparticipation are as large as flows between employment and unemployment. What about the long term unemployed? Some Remarks I I I I The model does not distinguish between unemployment due to quits and layoffs Flows between employment and nonparticipation are as large as flows between employment and unemployment. What about the long term unemployed? The model abstracts from vacancies, so that the number of new job openings always equals the number of unemployed workers