ELSNIT PARIS 2010, Comment MM backupx

Marcela Melendez

Informations

• Publié par: Nuzong
• Nombre de lectures: 42
• Langue: English

Extrait

“Institution-Driven Comparative

Advantage, Complex Goods and

Organizational Choice”

by Shon Ferguson and Sara Formai

Comment by Marcela Meléndez

VIII Conference of the Euro-Latin Study Network

on Integration and Trade (ELSNIT)

CEPII, Paris, France, October 2010

Building blocks

Nunn (2007) “Relationship-Specificity, Incomplete Contracts and the Pattern of

Trade”

•

Tests whether a country’s ability to enforce written contracts is an important

source of comparative advantage. Insight: when investments are relationship-

specific, underinvestment will occur if contracts cannot be enforced, so countries

with good contracting environments have a comparative advantage in goods that

require relationship-specific investments.

•

Measures for each commodity, the proportion of inputs that are relationship-

specific (US technology data). Inputs sold on organized exchange are not

relationship-specific. Inputs reference-priced in trade publications are also not

relationship-specific (alternative approach).

•

Measures comparative advantage in industry i as ratio of total exports from

country with better legal system to total exports from country with worse legal

system (each country relative to the US).

•

Finds that (1) countries with good contract enforcement specialize in industries

where relationship-specific investments are important, and (2) contract

enforcement explains more of the variation in trade-flows than factor

endowments.

•

Tests whether vertical integration reduces underinvestment when contracts are

imperfectly enforced, by restricting sample to sectors where vertical integration is

less feasible. Finds larger effects of judicial quality.

Building blocks

Acemoglu, Johnson and Mitton (2009) “Determinants of Vertical Integration:

Financial Development and Contracting Costs”

•

Investigates relationship between quality of contracting institutions, financial

development, and vertical integration across countries

•

Tests model predictions that (1) higher contracting costs together with greater

financial development lead to greater vertical integration, and (2) effect of

contracting costs on vertical integration decision is stronger in industries more

susceptible to holdup problems (i.e with more relationship-specific investments)

•

Computes vertical integration index for each country based on US technology

data and data on industry participation by firm, by country.

•

Finds little explanatory power of institutional characteristics once differences in

industrial composition are controlled for.

•

Computes country level “propensity to vertically integrate” to explore lack of

correlation, using share of industry sales over total sales, and average vertical

integration of the industry (in the US). Finds institutional characteristics effect on

this variable is swept away by development level control (GDP per capita)

•

But confirms model’s predictions (1) and (2).

•

Uses industry’s capital intensity as proxy for extent of holdup. Nunn’s relationship-

specificity measure is used in alternative models and yields non-significant, non-

robust results

The paper

In a nutshell

•

Investigates effect of legal and financial institutional quality on

comparative advantage across industries that vary in their complexity

and their propensity to vertically integrate (complexity understood as

contract intensity / extent of required relationship-specific investments).

•

Insight: ignoring ability to vertically integrate overstates importance of

contracts.

•

Uses Nunn’s (2007) relationship-specificity measure as proxy of industry

complexity.

•

Uses Acemoglu et al (2009) US industries’ average vertical integration

levels as industries’ propensities to vertically integrate.

•

Uses export levels by industry, by country, as measure of comparative

advantage.

•

Finds that (1) vertical integration lessens impact of a country’s ability to

enforce contracts on comparative advantage of complex goods, and (2)

countries with good financial institutions export more in sectors that

produce complex goods and have a high propensity to vertically

integrate.

The paper

Empirical strategy

•

Cross-section analysis

T

ci

=

β

0

+ β

1

(z

i

Q

c

) + β

2

(z

i

Q

c

vi

i

) + β

3

(z

i

CR

c

vi

i

) + β

4

(z

i

CR

c

)

(1)

+ β

5

(Q

c

vi

i

) + β

6

(CR

c

vi

i

) + X

ci

+ α

c

+ α

i

+ ε

ci

.

where

T

ci

is the log value of country cʼs exports to the rest of the world in

industry i; Q

c

is legal institutional quality; CR

c

is financial institutional quality; z

i

is Nunnʼs (2007) measure of contract intensity; vi

i

is Acemoglu et al.ʼs (2009)

measure of vertical integration propensity; X

ci

is a vector of country-industry

interaction controls; α

c

and α

i

are country and industry fixed effects

respectively.

•

Panel regression

T

cit

=

β

0

+ β

1

(z

i

L&O

ct

) + β

2

(z

i

L&O

ct

vi

i

) + β

3

(z

i

Libdum

ct

vi

i

)

(2)

+ β

4

(z

i

Lib dum

ct

) + X

cit

+ α

ci

+ α

t

+ ε

cit

where

L&O

ct

is the proxy for legal institutional quality;

Libdum

ct

is the financial

liberalization dummy variable (= 1 since year of financial liberalization); X

cit

is

a vector of controls; and α

ci

and α

t

are country-industry and time fixed effects.

•

First-difference regression model (first-differencing of equation (2))

Comments and questions

General

•

Is question relevant?

•

We know:

•

Higher contracting costs together with greater financial

development lead to greater vertical integration, and effect of

contracting costs on vertical integration decision is stronger in

industries more susceptible to holdup problems, i.e with more

relationship-specific investments (from Acemoglu et al, 2009)

•

Countries with good contract enforcement specialize in

industries where relationship-specific investments are

important (from Nunn, 2007)

•

We presume ability to vertically integrate (partially) substitutes

contract enforcement, by allowing efficient production (and exports)

of goods that require relationship-specific investments. In fact, we

know vertical integration is positively correlated with contracting

costs in contexts of sufficient financial development.

•

But we don’t know by how much the importance of contracts as a

source of comparative advantage is overstated when ability to

vertically integrate is not taken into account. So question is

indeed interesting.

Comments and questions

•

Is empirical strategy appropriate?

•

Exports size as a measure of comparative advantage. Why not use

more direct measures of trade specialization patterns?

•

Revealed Comparative Advantage (Balassa, 1965)

•

Findings based on negative coefficients on triple interactions

•

If higher relationship-specificity in an industry comes together

with higher average vertical integration, this reduces the impact of

contract enforcement on exports of that particular industry in

countries with better contract enforcement.

•

Alternative reading (please correct me): In countries with better

contract enforcement, industries with higher potential holdup

problems and higher average vertical integration levels are

associated with lower exports than industries with higher

potential holdup problems that are not highly vertically

integrated. (?)

•

Comments and questions

•

Is empirical strategy appropriate?

•

Concerns:

•

If in fact vertical integration (partially) substitutes contract

enforcement as a solution to holdup problems, then the vertical

integration that matters is the actual vertical integration of

industry i in country c (not the industry average). But vertical

integration at that level is partly correlated with the country’s

institutional characteristics, so actual model wouldn’t work...

•

Also, we know that relationship-specificity is a reason for firms

to benefit from vertical integration (i.e. industries’ facing stronger

holdup problems should on average be more vertically

integrated), so even if the propensity to vertically integrate is

industry wide (and the correlation with country characteristics is

avoided ) we are still interacting the variable who’s effect we

want to explore with a variable that is (at least partially) an

outcome variable. Doesn’t this bias the coefficients?

•

Suggestion: define vertical integration index by industry, by

country as a weighted average of Acemoglu et al.’s firm-level

indices and try a version of Nunn (2007)’s model?

Questions and comments

•

Concern about endogenous controls (TFP, Inter-Industry trade,

Value-added).

•

Capital intensity: is it or is it not a good proxy for relationship-specificity?

If it is, it should not enter as control.

•

What is the question we intend to respond with the financial

development interactions? Whether more vertically integrated

relationship-specific industries have lower/higher exports in contexts of

higher financial development? (I think I am missing the point).

•

Panel regression adds little insight (and can be misleading in terms of

answering the main question the paper wants to explore).