GROSSMAN-HART (1986) GOES GLOBAL

Incomplete Contracts, Property Rights, and the International Organization of Production

Pol Antràs (Harvard University)

June, 25th 2011 Grossman and Hart at 25 Conference

I. Two features of the world economy particularly relevant to understand the influence of G-H 1986 in International Trade

1. Fragmentation of the production process across countries2. Limited enforceability of contracts in international transactions

II. Briefly review theoretical implementation of G-H(-M) in general-equilibrium, open-economy environments

1. See survey paper for more details

III. Review empirical implementation of these theories

IV. Broader impact of the GHM approach

Road Map

I. THE SLICING OF THE VALUECHAIN AND THE INTERNATIONAL

ENFORCEABILITY OF CONTRACTS

A Flashback to 1986

A cell phone in 1986 looked like this…

… a laptop like this…

…and Al Gore had not yet created the internet

What Has Happened Since Then?An ICT Revolution!

0

10

20

30

40

50

60

70

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

Cell phone subscriptions (per 100 people) Telephone lines (per 100 people) Broadband Internet subscribers (per 100 people) Internet users (per 100 people)

Source: World Development Indicators

Significant Effect on Society

Source: World Trade Report (2008) using Yeats’ (2001) methodology

Large Impact on International Trade FlowsIt’s not wine for cloth anymore…

• Assembled in China (and soon in Brazil) by Taiwan-based Foxconn

An Example: the iPad 2

LG Display, Samsung (Korea, China)

TPK, now Wintek(China, Taiwan, India)

Simplo Tech, Dynapack(Taiwan)

Infineon, Qualcomm (Germany, US,

Singapore, Malaysia…)

Catcher Tech. (case)(Taiwan, China)

STMicroelectronics,AKM, TAOS

(Italy-France, Japan, U.S)

We’re not done yet…

Firms with R&D centers in developed countries and manufacturing plants worldwide

0

50

100

150

200

250

300

350

400

450

500

550

600

650

700

750

800

850

900

950

1000

Nethe

rland

sUSA

Japan

Korea

South Africa

Malaysia

Ghana

Ugand

aUK

Turkey

India

Malaw

iVietnam

Kazakhstan

Belgium

Tanzania

Spain

Cote D'Ivoire

Germany

Russia

Philipp

ines

China

Brazil

France

Zambia

Morocco

Zimbabw

eCh

ileEgypt

Thailand

Guatemala

Ukraine

Indo

nesia

Romania

Nigeria

Kenya

Banglade

shCzech Re

p.Mexico

Argentina

Greece

Australia

Ecuado

rVe

nezuela

Pakistan

Hungary

Portugal

Canada

Sri Lanka

Peru

Colombia

Mozam

biqu

eIta

lyPo

land

Total Duration (Days) of Legal Process to Collect a Bounced Check

Source: Djankov et al. (QJE, 2003)

Heterogeneity in Contracting Environments

iPad 2 producing countries

• Contract disputes in international trade: which country’s laws apply?• Local courts may be unwilling to enforce a contract signed between

residents of two different countries

• Attempts to resolve this issue – United Nations Convention on Contracts for the International Sale of Goods (CISG)• But not universally ratified and many parties opt out via Article 6

• There exist other forms of arbitration (e.g., ICC in Paris), but not widely used in practice

• Implicit contracts may be harder to sustain too due to limited repeated interactions and lack of collective punishment

• Rodrik (2000): “Ultimately, [international] contracts are often neither explicit nor implicit; they simply remain incomplete”

International Contract Enforcement

• In “Poorly Made in China,” Paul Midler describes his (mis)adventures as an offshoring consultant

• Describes the last-minute pricing maneuvers and clever manipulation of quality of Chinese suppliers

“ʹPrice go up!ʹ was the resounding chorus heard across the manufacturing sector” (p. 184)

• Also describes the ineffectiveness of typical solutions:• Relational contracting (“Reverse Frequent Flyer” effect)• Foreign ownership (no “Joint Venture Panacea”)

• Chinese saying: “Signing a contract is simply a first step in the negotiations”

Example: “Price Go Up!”

GROSSMAN AND HART (1986) GOES GLOBAL

• Consider a situation in which the manager of a firm F has a access to a technology for converting a specialized intermediate input or component m into a final good

• The manager F is also in charge of providing “headquarter services” h, which raise the marginal product of m

• Given an amount m of components and h of headquarter services, sale revenue is given by

R(h,m) with Rh>0; Rm>0; Rhh<0; Rmm<0; Rhm>0

• F needs to contract with an operator of a manufacturing plant (denoted by M) for the provision of m.

• F can produce h at a constant marginal cost ch; M produces m at marginal cost cm

A Stylized Property-Rights Model

• Contractual structure: before investments in h and m are made, the only contractibles are the allocation of residual rights (i.e., the ownership structure) and a lump-sum transfer between the two parties

• Ex-post determination of price follows from symmetric Nash bargaining: ½ , ½

• Distribution of surplus is sensitive to the mode of organization because the outside option of F is naturally higher when it owns M than when it does not

• Ownership entails residual rights of control over input m:• under outsourcing, contractual breach gives 0 to both agents (relaxable)• under integration of M, F can selectively fire M and seize input m

(because it has property rights over it), but incurs productivity loss equal to a share 1- of revenue

A Stylized Property-Rights Model (cont.)

Discussion of Assumptions

• Simpler than G-H 1986 in that:• Ownership of asset = Ownership of input produced with asset• Effectively, only one form of integration is relevant (F is essential

for final good production)• More structure on how investments affect inside and outside

options (payoffs are proportional to an aggregator of h and mregardless of ownership structure)

• Richer than G-H 1986 in that:• More general production technology that allows for

complementary investments• Investments affect each other’s disagreement payoffs (no need

to stick to interpretation of human capital investments)

k O 1

2 if F outsources to M

V 12 1 − if F integrates M

• The (jointly) optimal ownership structure is the solution to the following program:

where

Formulation of the Problem

maxk∈V,O

k Rhk, mk − ch hk − cm mk

s. t. hk arg maxhkRh, mk − ch h

mk arg maxm1 − k Rhk, m − cm m

k∗ ∈ V, O

• Suppose that instead of choosing , F could choose the joint profit-maximizing

• Optimal satisfies:

(1)

where for j = h, m

• Optimal to allocate bargaining power to F whenever his investment has a (relatively) large impact on revenue or is (relatively) more responsive to changes in bargaining power

A Result

k ∈ V, O

R,j ≡jRjh, mRh, m

j, ≡djd

j

∗

1 − ∗ R,h h,

R,m −m,

• Most applications of the property-rights approach to international trade further assume

• This is often obtained by assuming:1. Cobb-Douglas production technology in h and m2. Constant-elasticity of demand system (Dixit-Stiglitz preferences in GE)

• In such case, (1) reduces to

• Proposition: There exists a unique threshold such that for all integration dominates outsourcing, while for outsourcing dominates integration.

A More Parametric Example

Rh, m Ahh mm

∗

1 − ∗ h/1 − h m/1 − m

h ∈ 0, 1h h

h h

• Suppose that we can write

and q(h,m) is linearly homogeneous. Then

• increases in the output elasticity of h for any elasticity of substitution between h and m

• Effect of and depends crucially on whether is higher or lower than 1/2 (negative if < 1/2

A New Result

Rh, m Aqh, m

∗

1 − ∗ /1 − − 11 −

1 − /1 − 1 − − 1

• Consider a multi-country version of the above model in which firms are allowed to locate different parts of the production process in different countries• denote by L the set of possible location decisions (a mapping from

production processes to locations) and by ℓ ∈ L a particular one.

• Location decisions naturally affect marginal and fixed costs of production, and possibly revenue and ex-post division of surplus

• Fixed costs might also be affected by the ownership decision

Global Sourcing Decisions

maxk∈V,O,ℓ∈L

kℓ Rℓhk

ℓ , mkℓ − ch

ℓ hkℓ − cm

ℓ mkℓ − fk

ℓ gℓchℓ , cm

ℓ

s. t. hkℓ arg max

hk

ℓRh, mkℓ − ch

ℓ h

mkℓ arg max

m1 − k

ℓ Rhkℓ , m − cm

ℓ m

• In Antràs (2003), I interpreted F 's investment in h as being capital intensive relative to M ’s investment• can be boiled down to transferability of capital investments

• Proposition 1 then predicts a higher propensity to integrate suppliers in capital-intensive sectors

• I then embedded the model in a general equilibrium, factor proportions model of trade à la Helpman-Krugman• See also Grossman and Helpman (2002)

• Capital intensity not only affects the location of suppliers (or where M is produced), but also whether those sourcing decisions are integrated or not

• I showed how this had implications for how the share of intrafirm imports should correlate with capital intensity across industries and relative capital abundance across countries

An Application: The Role of Capital Intensity

Towards an Empirical Model of Global Sourcing Decisions

• In Antràs and Helpman (2004) we embed the framework in a Melitz (2003) style model with intraindustry heterogeneity

• Choice of organizational form faces two types of tensions:1. Location: South offers relatively low variable costs, but

relatively higher fixed costs (e.g., harder to find a supplier)2. Control: integration improves efficiency of variable production

when η is high (Proposition 1), but involves higher fixed costs

• We show that equilibrium can feature multiple organizational forms within an industry and study the determinants of the relative prevalence of these different organizational forms

Towards an Empirical Model of Global Sourcing Decisions

• Effect of headquarter intensity, but also of relative factor costs, trade frictions, or productivity dispersion

• Framework has been extended in several directions:• Partial contractibility (Antràs and Helpman, 2008)• Multiple suppliers (Acemoglu et al., 2006, Antràs and Chor, 2011)• Financial frictions (Carluccio and Fally, 2010, Basco, 2010, Conconi

et al., 2010)

• Model has served as springboard for extensive empirical literature

Brief Review of Other Related Theoretical Work

• Earlier work following the transaction-cost approach• Ethier (1986), Ethier and Markusen (1998), McLaren (2000),

Grossman and Helpman (2002, 03, 05)

• Work adopting organizational theories inspired by G-H’ 86• Authority: Marin and Verdier (2008, 09), Puga and Trefler (2002, 10)• Relational Contracting: Corcos (2006)

• Implications of incomplete contracting for international trade flows even in the absence of intermediate input trade• Acemoglu et al. (2006), Levchenko (2007), Nunn (2007), Costinot

(2009)

• Welfare implications of incomplete contracting in trade models• Antràs (2005), Levchenko (2007)

• Interaction with trade policy choices• Antràs-Staiger (2010), Díez (2010), Conconi et al. (2010, 2011)

CONFRONTING THE MODEL WITHINTERNATIONAL TRADE DATA

• Empirically validating the property-rights theory poses at least two important challenges1. Data on integration decisions is not readily available2. Predictions are associated with marginal returns to

investments that are generally unobservable in the data

• Admittedly, we have not made a lot of progress on point 2

• But data on international transactions is particularly accessible due to official records of goods crossing borders

• Fairly detailed data on U.S. intrafirm trade at the sector level (HS6; over 4000 of these) and origin/destination country level

• Also a few international firm-level datasets with detailed information on the sourcing strategies of firms• for example, the Spanish ESEE has data on domestic insourcing and

outsourcing and foreign insourcing and outsourcing at the firm level

Challenges and Opportunities

• Some pros:• Compiled from administrative records from official import and export

merchandise trade statistics• There is plenty of variation in the data (more on this later)• Easier to spot “fundamental” forces that appear to shape whether

international transactions are internalized or not• Potential to exploit ‘exogenous’ changes in sector characteristics or in

institutional features of importing/exporting countries

• Some cons:• Aggregates firm decisions; can’t control for firm-level determinants• Information only on the sector of the good being transacted

• Not always clear which sector is buying on the import or export side• Not always clear whether inputs or final goods are traded

• Not always clear who is integrating whom (backward vs. forward integration) and how large is the ownership stake

• U.S. firm level sourcing decisions might not be reflected in U.S. trade data (remember the iPad 2 example) – affiliates as intermediaries

Pros and Cons of Using Intrafirm Trade Data

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%Ba

nglade

shPeru

Ecuado

rCo

lombia

Chile

Vietnam

Indo

nesia

Hong

 Kon

gTurkey

China

India

Taiwan

Russia

Nigeria

Angola

Argentina

Spain

Algeria

Brazil

South Africa

Italy

Australia

Vene

zuela

Trinidad

Norway

Phillipines

Thailand

Canada

France

Israel

Belgium

Nethe

rland

sUK

Mexico

Hond

uras

Malaysia

Korea

Switzerland

Kuwait

Finland

Austria

Germany

Swed

enDe

nmark

Iraq

Singapore

Saud

i Arabia

Japan

Ireland

Costa Rica

Share of U.S. Intrafirm Imports for Top 50 Exporters in 2010

Source: U.S. Census

Variation in the Share of Intrafirm Trade across Countries

Aggregate Share of Intrafirm Trade: 48.57%

Variation in the Share of Intrafirm Trade across Industries

Source: U.S. Census

WOMEN

 S/G

IRLS OTH

ER OUTERW

EAR

WOMEN

 S/G

IRLS BLO

USES & SHIRTS

INSTITUTIONAL

 FURN

ITURE

OTH

ER NONFERR

OUS METAL

S

ALL OTH

ER PLA

STICS PR

ODUCT

S

JEWELER

S'  M

ATER

IAL & LAP

IDAR

Y WORK

CRUDE PE

TROLEUM & NAT

URA

L GAS

MOTO

R VE

HICLE PAR

TS, N

ESOI

GAM

ES, TOYS

 & CHILDRE

N'S VEH

ICLES

IRON & STEEL

AIRC

RAFT ENGINES & ENGINE PA

RTS

PETR

OLEUM REFINER

Y PR

ODUCT

S

RADIO/TV  & W

IRELESS CO

MM. EQUIP

ALL OTH

ER BAS

IC ORG

ANIC CHEM

ICAL

S

AUDIO & VIDEO

 EQUIPMEN

T

OTH

ER COMPU

TER EQ

UIPMEN

T

ELEC

TRONIC COMPU

TERS

SEMICONDUCT

ORS

LIQUID NAT

URA

L GAS

PRINTED CIRCU

IT ASSEM

BLIES

MED

ICINAL

 DRU

GS & VIT.

PHAR

MA. PRE

PARA

TIONS

PETR

OCH

EMICAL

S

AUTO

S & LIGHT DUTY

 VEH

ICLES

HEA

VY DUTY

 TRU

CKS & CHAS

SIS

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Share of U.S. Intrafirm Imports for Top 25 Importing Industries (NAICS6) in 2010

Large Variation in the Share of Intrafirm Trade within Sectors

Source: Nunn and Trefler’s dataset

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

8705

2087

0490

8716

3187

1610

8702

9087

1495

8705

4087

0530

8711

9087

1680

8703

9087

1491

8714

9387

1500

8716

9087

0331

8701

1087

0333

8713

9087

1411

8709

9087

1496

8712

0087

1499

8705

9087

1492

8716

4087

1000

8714

2087

1620

8706

0087

0870

8709

1187

0432

8708

1087

0860

8714

9487

0839

8707

9087

0899

8702

1087

1310

8708

9287

0893

8701

9087

1110

8709

1987

1419

8708

9187

0880

8708

9487

1639

8707

1087

0821

8704

2187

0829

8708

5087

0510

8708

3187

0321

8704

2387

0310

8703

2287

0840

8711

2087

0422

8704

3187

0410

8711

3087

0324

8701

3087

1150

8701

2087

1140

8703

2387

0332

Variation in the Share of U.S. Intrafirm Imports within HS2 sector 87(Vehicles, except Railway or Tramway, and Parts)

… And Also Within More Narrowly Defined Sectors

Source: Nunn and Trefler’s dataset

0%

10%

20%

30%

40%

50%

60%

70%

80%

Variation in the Share of U.S. Intrafirm Imports within HS4 Sector 8708 (Auto Parts)

... And Across Countries Within HS6 Sectors

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%Ecuado

rPo

rtugal

Sri Lanka

Vene

zuela

Lebano

nIndo

nesia

Argentina

Singapore

New

 Zealand

Peru

Hong

 Kon

gDe

nmark

China

Taiwan

Vietnam

Canada

Spain

Brazil

Philipp

ines

UAE

Mexico

Israel

Malaysia

India

France

Nethe

rland

sThailand

Czech Re

p.Sw

itzerland

Italy

South Ko

rea

Germany

Australia

South Africa

Norway UK

Japan

Belgium

Austria

Swed

enSlovakia

Hungary

Variation in the Share of U.S. Intrafirm Imports within HS6 Sector 870810 (Bumpers)

Source: Nunn and Trefler’s dataset

• Several studies have made use of U.S. intrafirm trade data• Antràs (2003), Yeaple (2006), Nunn and Trefler (2008a,b), Bernard

et al. (2010), Antràs and Chor (2011)

• There also exists work using detailed Chinese trade data• Feenstra and Hanson (2005), Fernandes and Tang (2010)

• Results are generally supportive of the predictions of the model though, admittedly, one might worry about the power of some of these tests

• I will next illustrate some of the results for U.S. imports through correlations (which also hold conditional on wide set of covariates)

Using Intrafirm Trade Data

aud

bev

checlecom

dru

ele

fme foo

ima

ins

lum

och

oel

oma

pap

pla

pri

rub

sto

tex

tra

veh

-4-3

-2-1

0S

hare

of I

ntra

firm

Tra

de

3 3.5 4 4.5 5 5.5Log of capital intensity

• Positively correlation with alternative measures of “headquarter intensity”

Intrafirm and Headquarter Intensity

aud

bev

checle com

dru

ele

fmefoo

ima

ins

lum

och

oel

oma

pap

pla

pri

rub

sto

tex

tra

veh

-4-3

-2-1

0Sh

are

of In

trafir

m T

rade

-7 -6 -5 -4 -3 -2Log of R&D intensity

aud

bev

checle com

dru

ele

fmefoo

ima

ins

lum

och

oel

oma

pap

pla

pri

rub

sto

tex

tra

veh

-4-3

-2-1

0S

hare

of I

ntra

firm

Tra

de

-2 -1.5 -1 -.5 0 .5Log of skill intensity

0.2

.4.6

.81

Sha

re o

f U.S

. Int

rafir

m Im

ports

-2.5 -2 -1.5 -1 -.5Log of skill intensity

0.2

.4.6

.81

Sha

re o

f U.S

. Int

rafir

m Im

ports

-7 -6 -5 -4 -3 -2Log of R&D intensity + 0.001

Intrafirm and Headquarter Intensity

0.2

.4.6

.81

Sha

re o

f U.S

. Int

rafir

m Im

ports

0 1 2 3 4 5Log of capital intensity

• Positively correlation with alternative measures of “headquarter intensity”

• Robust to various controls and to country fixed effects in country/industry regressions

What is Behind the Effect of Capital Intensity?

-.4-.2

0.2

.4.6

e( im

p_in

trafir

m |

X )

-2 -1 0 1 2e( k_l_struct | X )

coef = -.03425467, se = .01541797, t = -2.22

-.50

.5e(

imp_

intra

firm

| X

)

-2 -1 0 1 2e( k_l_equip | X )

coef = .08361554, se = .01623926, t = 5.15

• As pointed out by Nunn and Trefler (2009), effect of capital intensity is driven by capital equipment intensity, not structures

• Graphs plots partial effect of the log capital intensity of a particular type of capital

• Regressions controls for R&D intensity and skill intensity

• And within capital equipment, the correlation is not driven by spending on autos or computers (see Nunn and Trefler, 2009)

-.4-.2

0.2

.4e(

imp_

intra

firm

| X

)

-4 -2 0 2 4e( k_l_autos | X )

coef = -.02385051, se = .01169016, t = -2.04

-.4-.2

0.2

.4e(

imp_

intra

firm

| X

)

-1 0 1 2e( k_l_comput | X )

coef = .0126022, se = .02068955, t = .61

-.50

.5e(

imp_

intra

firm

| X

)

-2 -1 0 1 2e( k_l_other | X )

coef = .06709692, se = .01533816, t = 4.37

What is Behind the Effect of Capital Intensity?

0.2

.4.6

.81

Sha

re o

f U.S

. Int

rafir

m Im

ports

0 .2 .4 .6 .8 1Nunn's measure of contractibility

0.2

.4.6

.81

Shar

e of

U.S

. Int

rafir

m Im

ports

.2 .4 .6 .8Bernard et al.'s measure of contractibility

Contractibility

• Some evidence of a negative correlation between intrafirmtrade and contractibility

• What are these measures capturing and are these patterns consistent with theory?

• Antràs and Helpman (2008): key is whether contractibility is low in headquarter services or in the provision of the input

0.2

.4.6

.81

Sha

re o

f U.S

. Int

rafir

m Im

ports

1 1.5 2 2.5 3Productivity Dispersion

Implications of Productivity Heterogeneity

• Some evidence of a positive correlation with productivity dispersion

• Consistent with Antràs and Helpman (2004)

• Díez (2010) finds a positive correlation with U.S. tariffs

• Again consistent with A-H’ 04

25%

30%

35%

40%

45%

0.00 0.66 1.68 2.50 3.35 5.35

Sha

re o

f U.S

. Int

rafir

mIm

ports

U.S. Tariff (%)

AGO

ARG AUS

AUTBDI

BEL

BEN

BFA BGD

BOL

BRA

BRBBWACAF

CAN

CHE

CHLCHN

CIV

CMRCOG

COL

COM

CPV

CRI

CYP

DNK

DOM

DZAECU

EGY ESP

ETH

FIN

FJI

FRA

GAB

GBR

GHA

GIN

GMB

GNB

GNQ

GRCGTM

GUY

HKG

HND

HTI

HUN

IDN

IND

IRL

IRN

ISLISR

ITA

JAM

JOR

JPN

KEN

KOR

LKA

LSO

LUX

MAR

MDG

MEX

MLI

MOZ

MRT

MUS

MWI

MYS

NAMNER

NGA NIC

NLD

NOR

NPL

NZL

PAKPANPER

PHL

PNGPOL

PRT

PRY

ROM

RWASEN

SGP

SLE

SLV

SWE

SYR

TCD

TGO

THA

TTOTUN

TURTZAUGA URY

VEN

ZAF

ZAR

ZMB

ZWE

0.2

.4.6

.81

Shar

e of

U.S

. Int

rafir

m Im

ports

6 8 10 12Log Capital/Labor Ratio (1996-2005)

Cross-Country Variation

• Positive correlation between physical capital abundance and the share of intrafirm trade (see Antràs, 2003)

• Positive correlation between rule and law and the share of intrafirm trade

• Holds for various institutional variables

ARG AUS

AUTBEL

BRACAN

CHE

CHLCOL

DEU

DNK

ECUEGY ESP

FIN

FRA

GBR

GRC

HKG

IDN

IND

IRL

ISR

ITA

JOR

JPN

KEN

KOR

LKA

MEXMYS

NGA

NLD

NORNZL

PAKPER

PHL PRT

SGP

SWE

THA

TUR

TWN

URY

VEN

ZAF

ZWE

0.2

.4.6

.81

Sha

re o

f U.S

. Int

rafir

m Im

ports

2 4 6 8 10Rule of Law

• Tomiura (2007) uses a rich sample of 118,300 Japanese manufacturing firms in 1998• Foreign outsourcing versus foreign integration

• Corcos et al. (2009) and Defever and Toubal (2009) use French data from a survey conducted in 1999 by SESSI (Service des Études Statistiques Industrielles)• Again foreign outsourcing versus foreign integration

• Kohler and Smolka (2009) use data from the Spanish Survey on Business Strategies (ESEE) • Data on both domestic and foreign outsourcing and integration

• Availability of these datasets opens the door for more structural tests of the model and might allow to circumvent some of the obvious “observability” issues in the literature

Firm-Level Studies

• Institutions and comparative advantage• Contracting institutions, financial institutions, labor institutions

• Understanding the financial structure of multinational firms• Implications for aggregate capital flows across countries

• Effects from trade when power matters in market (and nonmarket) transactions• Intermediaries, government pressures

• Trade Policy when power matters in market and (nonmarket) transactions• Implications for effectiveness of WTO rules

Broader Impact of GHM in International Trade