input linkages and the transmission of shocks:...
TRANSCRIPT
Input Linkages and the Transmission of Shocks:Firm-Level Evidence from the 2011 Tohoku
Earthquake
Christoph Boehm Aaron Flaaen Nitya Pandalai-Nayar
2016 AEA Annual Meeting
Discussion by Teresa Fort
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 1 / 12
Big Picture Questions
How important are global value chains in propagating shocksacross countries?
I Potential for amplification effects due to input linkagesI Focus on traded intermediate inputsI Role for multinationals and intra-firm trade
How substitutable are domestic and foreign inputs?I Key determinant of extent of propagation
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 2 / 12
Main Contributions
Exploit clearly exogenous shock
Construct multiple new datasetsI Identify MNEs in Census dataI New methodology to classify intermediate importsI Firm-specific measure of earthquake intensity
Reduced form evidence on propagation of shock across countriesI Monthly estimates of reduction in imports and outputI Large declines in imports and exports by Japanese subsidiaries
Structural estimates of elasticity of substitution between domesticand foreign inputs
I Suggest that production chains are rigidI Small shocks can be magnified
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 3 / 12
Main Contributions
Exploit clearly exogenous shockConstruct multiple new datasets
I Identify MNEs in Census dataI New methodology to classify intermediate importsI Firm-specific measure of earthquake intensity
Reduced form evidence on propagation of shock across countriesI Monthly estimates of reduction in imports and outputI Large declines in imports and exports by Japanese subsidiaries
Structural estimates of elasticity of substitution between domesticand foreign inputs
I Suggest that production chains are rigidI Small shocks can be magnified
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 3 / 12
Main Contributions
Exploit clearly exogenous shockConstruct multiple new datasets
I Identify MNEs in Census dataI New methodology to classify intermediate importsI Firm-specific measure of earthquake intensity
Reduced form evidence on propagation of shock across countriesI Monthly estimates of reduction in imports and outputI Large declines in imports and exports by Japanese subsidiaries
Structural estimates of elasticity of substitution between domesticand foreign inputs
I Suggest that production chains are rigidI Small shocks can be magnified
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 3 / 12
Main Contributions
Exploit clearly exogenous shockConstruct multiple new datasets
I Identify MNEs in Census dataI New methodology to classify intermediate importsI Firm-specific measure of earthquake intensity
Reduced form evidence on propagation of shock across countriesI Monthly estimates of reduction in imports and outputI Large declines in imports and exports by Japanese subsidiaries
Structural estimates of elasticity of substitution between domesticand foreign inputs
I Suggest that production chains are rigidI Small shocks can be magnified
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 3 / 12
Comment 1: How representative is the analysis?
Is supply chain disruption a common/important problem?
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 4 / 12
Supply chain risks
Source: Economist Intelligence Unit 2009 survey of 500 executivesBoehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 5 / 12
Comment 1: How representative is the analysis?
Is supply chain disruption a common/important problem?How do Japanese MNEs compare to other firms?
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 6 / 12
U.S. Intrafirm Import Shares by Country
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 7 / 12
U..S Intrafirm Import Shares from Japan and ROW byIndustry
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 8 / 12
U.S. Intrafirm Import Shares from Japan vs. Share ofU.S. Plants that Purchase CMS
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 9 / 12
Comment 2: Reduced form evidenceAuthors estimate:
V Mi,t = αi +
9∑p=−4
γpEp +9∑
p=−4
βpEpJPNi,p + ui,t
I V Mi,t is deviation from trend for imports or NA exports
Then divide βp estimates by average pre-shock levels
What happens if you estimate percentage drops directly?
Or if you estimate the effect on exports from a drop in imports,instrumenting with the shock?Additional support for this mechanism
I Can you exploit heterogeneity in impact across Japanese firms?I Does the impact increase in firm’s share of Japanese inputs?I Do affected firms have lower energy or other variable cost
expenditures?
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 10 / 12
Comment 2: Reduced form evidenceAuthors estimate:
V Mi,t = αi +
9∑p=−4
γpEp +9∑
p=−4
βpEpJPNi,p + ui,t
I V Mi,t is deviation from trend for imports or NA exports
Then divide βp estimates by average pre-shock levels
What happens if you estimate percentage drops directly?
Or if you estimate the effect on exports from a drop in imports,instrumenting with the shock?Additional support for this mechanism
I Can you exploit heterogeneity in impact across Japanese firms?I Does the impact increase in firm’s share of Japanese inputs?I Do affected firms have lower energy or other variable cost
expenditures?
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 10 / 12
Comment 2: Reduced form evidenceAuthors estimate:
V Mi,t = αi +
9∑p=−4
γpEp +9∑
p=−4
βpEpJPNi,p + ui,t
I V Mi,t is deviation from trend for imports or NA exports
Then divide βp estimates by average pre-shock levels
What happens if you estimate percentage drops directly?
Or if you estimate the effect on exports from a drop in imports,instrumenting with the shock?Additional support for this mechanism
I Can you exploit heterogeneity in impact across Japanese firms?I Does the impact increase in firm’s share of Japanese inputs?I Do affected firms have lower energy or other variable cost
expenditures?
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 10 / 12
Comment 2: Reduced form evidenceAuthors estimate:
V Mi,t = αi +
9∑p=−4
γpEp +9∑
p=−4
βpEpJPNi,p + ui,t
I V Mi,t is deviation from trend for imports or NA exports
Then divide βp estimates by average pre-shock levels
What happens if you estimate percentage drops directly?
Or if you estimate the effect on exports from a drop in imports,instrumenting with the shock?
Additional support for this mechanismI Can you exploit heterogeneity in impact across Japanese firms?I Does the impact increase in firm’s share of Japanese inputs?I Do affected firms have lower energy or other variable cost
expenditures?
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 10 / 12
Comment 2: Reduced form evidenceAuthors estimate:
V Mi,t = αi +
9∑p=−4
γpEp +9∑
p=−4
βpEpJPNi,p + ui,t
I V Mi,t is deviation from trend for imports or NA exports
Then divide βp estimates by average pre-shock levels
What happens if you estimate percentage drops directly?
Or if you estimate the effect on exports from a drop in imports,instrumenting with the shock?Additional support for this mechanism
I Can you exploit heterogeneity in impact across Japanese firms?I Does the impact increase in firm’s share of Japanese inputs?I Do affected firms have lower energy or other variable cost
expenditures?
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 10 / 12
Comment 3: Structural estimates of the elasticity ofsubstitution between domestic and foreign inputs
New micro approach to estimating an important elasticityI Firm-level production function estimationI N.A. exports rather than output
Small sample dominated by a few industriesI 105 Japanese MNEsI 35 in Motor Vehicle Manufacturing
How to aggregate micro elasticities?I Explanations for differences between Japanese and non-Japanese
elasticities
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 11 / 12
Comment 3: Structural estimates of the elasticity ofsubstitution between domestic and foreign inputs
New micro approach to estimating an important elasticityI Firm-level production function estimationI N.A. exports rather than output
Small sample dominated by a few industriesI 105 Japanese MNEsI 35 in Motor Vehicle Manufacturing
How to aggregate micro elasticities?I Explanations for differences between Japanese and non-Japanese
elasticities
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 11 / 12
Comment 3: Structural estimates of the elasticity ofsubstitution between domestic and foreign inputs
New micro approach to estimating an important elasticityI Firm-level production function estimationI N.A. exports rather than output
Small sample dominated by a few industriesI 105 Japanese MNEsI 35 in Motor Vehicle Manufacturing
How to aggregate micro elasticities?I Explanations for differences between Japanese and non-Japanese
elasticities
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 11 / 12
Comment 4: Opens the door to many new questions
Other channels for global value chains in propagating shocksacross countries
I Role for outsourcing?I Role for trade in final goods?I Mechanism for persistence?
How were U.S. upstream and downstream suppliers affected?I Employment effects at wholesale and retail firms?I Decreased or increased demand for U.S. input suppliers?
More nuanced employment effects?I Use ASM to assess effect to production workersI Heterogeneous effects within group of Japanese MNEs?I Plant entry or exit?
Optimal organization of supply chainsI Single vs. multiple sourcing
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 12 / 12
Comment 4: Opens the door to many new questions
Other channels for global value chains in propagating shocksacross countries
I Role for outsourcing?I Role for trade in final goods?I Mechanism for persistence?
How were U.S. upstream and downstream suppliers affected?I Employment effects at wholesale and retail firms?I Decreased or increased demand for U.S. input suppliers?
More nuanced employment effects?I Use ASM to assess effect to production workersI Heterogeneous effects within group of Japanese MNEs?I Plant entry or exit?
Optimal organization of supply chainsI Single vs. multiple sourcing
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 12 / 12
Comment 4: Opens the door to many new questions
Other channels for global value chains in propagating shocksacross countries
I Role for outsourcing?I Role for trade in final goods?I Mechanism for persistence?
How were U.S. upstream and downstream suppliers affected?I Employment effects at wholesale and retail firms?I Decreased or increased demand for U.S. input suppliers?
More nuanced employment effects?I Use ASM to assess effect to production workersI Heterogeneous effects within group of Japanese MNEs?I Plant entry or exit?
Optimal organization of supply chainsI Single vs. multiple sourcing
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 12 / 12
Comment 4: Opens the door to many new questions
Other channels for global value chains in propagating shocksacross countries
I Role for outsourcing?I Role for trade in final goods?I Mechanism for persistence?
How were U.S. upstream and downstream suppliers affected?I Employment effects at wholesale and retail firms?I Decreased or increased demand for U.S. input suppliers?
More nuanced employment effects?I Use ASM to assess effect to production workersI Heterogeneous effects within group of Japanese MNEs?I Plant entry or exit?
Optimal organization of supply chainsI Single vs. multiple sourcing
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 12 / 12
Back-Up Slides
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 1 / 10
Single vs. multiple sourcing
Number of countries per HS10 products imported by a firm
Firm LevelMean Median Max
Mean 1.11 1.00 1.61Median 1.03 1.00 1.0095%tile 1.78 1.00 4.00
Source: Antràs, Fort, and Tintelnot (2014)
Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 2 / 10
Same product from multiple countries
Source: Blaum, Lelarge, and Peters 2013Boehm, Flaaen, Pandalai-Nayar Fort Discussion 1/3/2016 3 / 10