1 / 15
EC 791 - International TradeTrade and the 2008-2009 Crisis
Stefania Garetto
Trade and the Current World Economic Crisis
Introduction
• Data
• References
LLT 2010
EKNR 2010
AKM 2010
2 / 15
[Source: Melitz]
Trade and the Current World Economic Crisis (contd.)
Introduction
• Data
• References
LLT 2010
EKNR 2010
AKM 2010
3 / 15
[Source: Melitz]
References
Introduction
• Data
• References
LLT 2010
EKNR 2010
AKM 2010
4 / 15
• Eaton, J., S. Kortum, B. Neiman and J. Romalis (2010).“Trade andthe Global Recession”. Mimeo.
• Alessandria, G., J. Kaboski and V. Midrigan (2010). “The GreatTrade Collapse of 2008-09: An Inventory Adjustment?”. IMFEconomic Review.
• Levchenko, A., L. Lewis and L. Tesar (2010). “The Collapse ofInternational Trade During the 2008-2009 Crisis: In Search of theSmoking Gun”. IMF Economic Review.
• Amiti, M., and D. Weinstein (2009). “Exports and Financial Shocks”.NBER Working Paper #15556.
• Bems, R., R.C. Johnson, and K. Yi (2010). “The Role of VerticalLinkages in the Propagation of the Global Downturn of 2008”.Mimeo.
• Chor, D., and K. Manova (2010). “Off the Cliff and Back? CreditConditions and International Trade during the Global FinancialCrisis”. NBER Working Paper #16174.
Levchenko, Lewis and Tesar (2010)
Introduction
LLT 2010
EKNR 2010
AKM 2010
5 / 15
Reduced-form empirical analysis of the trade collapse of 2008-2009:
• The drop in trade/GDP during the current crisis has been far largerthan in previous downturns and global in nature (interests the USand all its major trading partners).
• Larger drops of trade volumes in sectors where intermediate goodsare important (e.g. automobiles and industrial supplies).
• Empirical support for compositional effects (like in EKNR): largerdrops in trade in those sectors that experienced drops in domesticoutput.
• No empirical support for restrictions in trade credit: more tradecredit-intensive sectors did not experience larger drops in tradeflows.
Eaton, Kortum, Neiman and Romalis (2010)
Introduction
LLT 2010
EKNR 2010
• Summary
• Model
• Shocks
• Calibration
AKM 2010
6 / 15
The ratio global trade/GDP declined by about 30% during 2008-2009.
Eaton, Kortum, Neiman and Romalis (2010)
Introduction
LLT 2010
EKNR 2010
• Summary
• Model
• Shocks
• Calibration
AKM 2010
6 / 15
The ratio global trade/GDP declined by about 30% during 2008-2009.Why?
Two hypotheses:
1. The recession was more severe for more traded sectors (changingcomposition of global GDP).
2. Trade frictions increased during the recession (restrictions to tradecredit, “Buy America” provisions).
Eaton, Kortum, Neiman and Romalis (2010)
Introduction
LLT 2010
EKNR 2010
• Summary
• Model
• Shocks
• Calibration
AKM 2010
6 / 15
The ratio global trade/GDP declined by about 30% during 2008-2009.Why?
Two hypotheses:
1. The recession was more severe for more traded sectors (changingcomposition of global GDP).
2. Trade frictions increased during the recession (restrictions to tradecredit, “Buy America” provisions).
This paper:
• Develops a multi-country, multi-sector structural model withendogenous trade shares a la EK incorporating an input-outputstructure.
• Computes counterfactual scenarios of the model to evaluate theimpact of different channels on the decline of trade during the crisis.
Eaton, Kortum, Neiman and Romalis (2010) (contd.)
Introduction
LLT 2010
EKNR 2010
• Summary
• Model
• Shocks
• Calibration
AKM 2010
7 / 15
Preview of the results:1
1. The relative decline in demand for manufacturing goods (tradeables)was the most important driver of the decline of trade/GDP,accounting for about 80% of the global decline.
2. Trade frictions increased for some countries (China and Japanamong others), but dropped or remained stable for others. Theaggregate impact of changes in trade frictions was negligible.
Eaton, Kortum, Neiman and Romalis (2010) (contd.)
Introduction
LLT 2010
EKNR 2010
• Summary
• Model
• Shocks
• Calibration
AKM 2010
7 / 15
Preview of the results:1
1. The relative decline in demand for manufacturing goods (tradeables)was the most important driver of the decline of trade/GDP,accounting for about 80% of the global decline.
2. Trade frictions increased for some countries (China and Japanamong others), but dropped or remained stable for others. Theaggregate impact of changes in trade frictions was negligible.
Why this is important?
- If only compositional effects matter, trade works as a “perfect” channel oftransmission of shocks across countries.- If there are changes in trade frictions specific to the crisis, trade not onlypropagates but also “amplifies” shocks.
1Notice that these results are calibration-specific, i.e., the model is not constructed to deliverthis answer. The same model calibrated to match the Great Depression of the 1930s generatesa larger quantitative importance of trade frictions versus compositional effects.
EKNR (2010) - Sketch of the Model
Introduction
LLT 2010
EKNR 2010
• Summary
• Model
• Shocks
• Calibration
AKM 2010
8 / 15
Like in EK (2002), the price that a producer from country i charges incountry n to sell good z in sector j is:
pjni(z) =cjid
jni
aji (z)
where:
- djni: iceberg cost of delivering one unit of a sector-j good from i to n
- aji (z): country i’s efficiency at producing sector-j good z
- cji : cost of an input bundle in country i and sector j.
The modeling of cji differs from EK (2002): it incorporates an “input-output”structure (see next).
EKNR (2010) - Sketch of the Model (contd.)
Introduction
LLT 2010
EKNR 2010
• Summary
• Model
• Shocks
• Calibration
AKM 2010
9 / 15
The cost of an input bundle in country i and sector j is:
cji =1
AjSi
wβj
i
i
∏
l∈ΩM
(
pli)γ
jl
i(1−β
j
i)
where:
- AjSi : effect of sector j productivity in non-tradables on sector j ’s costs
- wi: wage in country i- pli = cli/A
li
- ΩM : set of manufacturing sectors- βj
i : value-added share in sector j in country i
- γjli : share of manufacturing sector l- intermediates used in sector j in
country i.
EKNR (2010) - Sketch of the Model (contd.)
Introduction
LLT 2010
EKNR 2010
• Summary
• Model
• Shocks
• Calibration
AKM 2010
9 / 15
The cost of an input bundle in country i and sector j is:
cji =1
AjSi
wβj
i
i
∏
l∈ΩM
(
pli)γ
jl
i(1−β
j
i)
where:
- AjSi : effect of sector j productivity in non-tradables on sector j ’s costs
- wi: wage in country i- pli = cli/A
li
- ΩM : set of manufacturing sectors- βj
i : value-added share in sector j in country i
- γjli : share of manufacturing sector l- intermediates used in sector j in
country i.As in EK (2002), trade shares are given by:
πjni =
T ji
(
cjidjni
)−ϑj
∑Ik=1 T
jk
(
cjkdjnk
)−ϑj
where T ji and ϑj are sector-specific parameters describing absolute and
comparative advantage, respectively.
EKNR (2010) - Structure of Shocks
Introduction
LLT 2010
EKNR 2010
• Summary
• Model
• Shocks
• Calibration
AKM 2010
10 / 15
Let αji denote the share of sector j consumption in country i’s aggregate
final demand, and let Dji = Xj
i − Y ji a country’s deficit in sector j.
The computation of the model includes four types of shocks:
1. shocks to αji represent shocks to the composition of demand
(Hypothesis 1)
2. shocks to djni represents changes in trade frictions (Hypothesis 2)
3. shocks to productivity Aji (not linked to the hypotheses, feed them
from the data)
4. shocks to deficits Dji (not linked to the hypotheses, feed them from
the data).
EKNR (2010) - Calibration
Introduction
LLT 2010
EKNR 2010
• Summary
• Model
• Shocks
• Calibration
AKM 2010
11 / 15
• Calibrate the model to match trade flows and GDP levels of 22countries + aggregate ROW.
• “Extract” the shocks directly from data on manufacturing, tradedeficits, TFP residuals, trade shares and prices.
• With the calibrated model, quantify the importance of each shock forthe changes in trade/GDP, global and across countries:
The model with all the shocks fits the data perfectly byconstruction.
Shut down all the shocks but one, one shock at a time, andcompute the model to recover the effect of each individualshock on the magnitudes of interest.
• Support for Hypothesis 1: the major determinant of the drop in globaltrade/GDP flows appears to be a disproportionate change indemand for goods in traded sectors.
Alessandria, Kaboski and Midrigan (2010)
Introduction
LLT 2010
EKNR 2010
AKM 2010
• Summary
• Evidence
• Model
12 / 15
The recent crisis has been characterized by a large drop in trade, largerthan the drop in manufacturing production.
Alessandria, Kaboski and Midrigan (2010)
Introduction
LLT 2010
EKNR 2010
AKM 2010
• Summary
• Evidence
• Model
12 / 15
The recent crisis has been characterized by a large drop in trade, largerthan the drop in manufacturing production.
What is the role of inventories in the downturn? Could inventories explainthe size of the drop in trade?
Alessandria, Kaboski and Midrigan (2010)
Introduction
LLT 2010
EKNR 2010
AKM 2010
• Summary
• Evidence
• Model
12 / 15
The recent crisis has been characterized by a large drop in trade, largerthan the drop in manufacturing production.
What is the role of inventories in the downturn? Could inventories explainthe size of the drop in trade?
Main idea: Y = C + I , so production more volatile than sales ifinventories investment is procyclical. If inventories are relatively moreimportant for traded goods, in a downturn we should expect trade to fallmore (to be more volatile) than both total production and sales.
Alessandria, Kaboski and Midrigan (2010)
Introduction
LLT 2010
EKNR 2010
AKM 2010
• Summary
• Evidence
• Model
12 / 15
The recent crisis has been characterized by a large drop in trade, largerthan the drop in manufacturing production.
What is the role of inventories in the downturn? Could inventories explainthe size of the drop in trade?
Main idea: Y = C + I , so production more volatile than sales ifinventories investment is procyclical. If inventories are relatively moreimportant for traded goods, in a downturn we should expect trade to fallmore (to be more volatile) than both total production and sales.
This paper:
• Documents the large size of the drop in trade flows, and the role ofinventories during the crisis.
• Develops a model with endogenous inventories and trade frictionsthat can account quantitatively for the experience of the currentcrisis.
AKM (2010): Empirical Evidence
Introduction
LLT 2010
EKNR 2010
AKM 2010
• Summary
• Evidence
• Model
13 / 15
• Trade is more volatile than production or expenditure in tradedgoods.
In downturns, we should expect trade to fall more than production andexpenditure, but to recover faster.
The current recession shows exactly these features, hence hasnothing unusual in terms of qualitative behavior.
What is unusual is the size of the trade collapse during the current
recession.
AKM (2010): Empirical Evidence
Introduction
LLT 2010
EKNR 2010
AKM 2010
• Summary
• Evidence
• Model
13 / 15
• Trade is more volatile than production or expenditure in tradedgoods.
In downturns, we should expect trade to fall more than production andexpenditure, but to recover faster.
The current recession shows exactly these features, hence hasnothing unusual in terms of qualitative behavior.
What is unusual is the size of the trade collapse during the current
recession.
• Inventories-to-sales ratio has been countercyclical in the current andin all past recessions.
Inventories increases appear to lead the drop in trade: suggestiveevidence that the drop in trade (see imports of intermediate goods forexample) could reflect an inventory adjustment, i.e. reduceexcessive inventories.
Hard to disentangle if the hypothesis above is correct from aggregate
data: most industries do not report sales and inventory data
separately for domestic and foreign goods. ⇒ Focus on the auto
industry, for which disaggregated data are available.
AKM (2010): Empirical Evidence: the Auto Industry
Introduction
LLT 2010
EKNR 2010
AKM 2010
• Summary
• Evidence
• Model
14 / 15
Why the auto industry?
1. Relevance:
• Autos are one of the most important traded sectors (accountfor about 11% of exports and 18% of imports of non-petroleumproducts for the US, 1999-2008);
• The auto industry experienced the largest and most immediatedecline in the current recession.
2. Data availability:
• Direct measures of domestic sales of imported autos and ofimports of autos;
• Measures of foreign and domestic inventories.
AKM (2010): Empirical Evidence: the Auto Industry
Introduction
LLT 2010
EKNR 2010
AKM 2010
• Summary
• Evidence
• Model
14 / 15
Why the auto industry?
1. Relevance:
• Autos are one of the most important traded sectors (accountfor about 11% of exports and 18% of imports of non-petroleumproducts for the US, 1999-2008);
• The auto industry experienced the largest and most immediatedecline in the current recession.
2. Data availability:
• Direct measures of domestic sales of imported autos and ofimports of autos;
• Measures of foreign and domestic inventories.
Evidence on inventories:
• Before the collapse in auto imports, inventories of foreign autos rosewhile sales were falling: the drop in imports brought the inventorylevels back in line with sales.
AKM (2010): The Idea Behind the Model
Introduction
LLT 2010
EKNR 2010
AKM 2010
• Summary
• Evidence
• Model
15 / 15
Punchline: goods (firms) with high inventory holdings experience relativelyhigh drops in imports in response to a drop in sales.
Intuition: when sales drop, inventories must also drop to keep theinventories/sales ratio constant. For inventories/sales ratios > 1, the dropin imports must be larger than the drop in sales, and will be larger thelarger the inventories/sales ratio itself.
AKM (2010): The Idea Behind the Model
Introduction
LLT 2010
EKNR 2010
AKM 2010
• Summary
• Evidence
• Model
15 / 15
Punchline: goods (firms) with high inventory holdings experience relativelyhigh drops in imports in response to a drop in sales.
Intuition: when sales drop, inventories must also drop to keep theinventories/sales ratio constant. For inventories/sales ratios > 1, the dropin imports must be larger than the drop in sales, and will be larger thelarger the inventories/sales ratio itself.
Example: consider a firm with inventories/sales =3. If it sells 10, must haveinventories of 30. Every time it sells 10, must import 10 to keep the ratio =3.Now sales drop to 9. The firm has inventories of 21, and should have9x3=27, so it imports 6. Hence sales dropped of 10% and imports droppedof 40%.
AKM (2010): The Idea Behind the Model
Introduction
LLT 2010
EKNR 2010
AKM 2010
• Summary
• Evidence
• Model
15 / 15
Punchline: goods (firms) with high inventory holdings experience relativelyhigh drops in imports in response to a drop in sales.
Intuition: when sales drop, inventories must also drop to keep theinventories/sales ratio constant. For inventories/sales ratios > 1, the dropin imports must be larger than the drop in sales, and will be larger thelarger the inventories/sales ratio itself.
Example: consider a firm with inventories/sales =3. If it sells 10, must haveinventories of 30. Every time it sells 10, must import 10 to keep the ratio =3.Now sales drop to 9. The firm has inventories of 21, and should have9x3=27, so it imports 6. Hence sales dropped of 10% and imports droppedof 40%.
The model formalizes this intuition to a two-country GE framework withendogenous inventory adjustment. Larger frictions to international tradecompared to domestic trade induce the inventories of imported products tobe higher than for domestic ones, and generate the result sought.Calibration and simulation exercises show that the mechanism of the modelis able to reproduce the drop and rebound in trade during the current crisis.