Intermediaries in International Trade: Direct versus indirect modes of export

NBER WORKING PAPER SERIES

INTERMEDIARIES IN INTERNATIONAL TRADE:

DIRECT VERSUS INDIRECT MODES OF EXPORT

Andrew B. Bernard

Marco Grazzi

Chiara Tomasi

Working Paper 17711

http://www.nber.org/papers/w17711

NATIONAL BUREAU OF ECONOMIC RESEARCH

1050 Massachusetts Avenue

Cambridge, MA 02138

December 2011

The present work has been possible thanks to a research agreement between the Italian Statistical Office

(ISTAT) and the Scuola Superiore Sant'Anna. Grazzi and Tomasi gratefully acknowledge financial

support from the European Commission 6th FP (Contract CIT3-CT-2005-513396), Project: DIME

- Dynamics of Institutions and Markets in Europe. Tomasi gratefully acknowledges financial support

by the Marie Curie Program Grant COFUND Provincia Autonoma di Trento. Bernard acknowledges

support from the European University Institute and the Tuck School of Business. The views expressed

herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic

Research.

NBER working papers are circulated for discussion and comment purposes. They have not been peer-

reviewed or been subject to the review by the NBER Board of Directors that accompanies official

NBER publications.

of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full

credit, including © notice, is given to the source.

Intermediaries in International Trade: Direct versus indirect modes of export

Andrew B. Bernard, Marco Grazzi, and Chiara Tomasi

NBER Working Paper No. 17711

December 2011

JEL No. D22,F12,F14,L22,L23

ABSTRACT

This paper examines the factors that give rise to intermediaries in exporting and explores the implications

for trade volumes. Export intermediaries such as wholesalers serve different markets and export different

products than manufacturing exporters. In particular, high market-specific fixed costs of exporting,

the (lack of) quality of the general contracting environment and product-specific factors play important

roles in explaining the existence of export intermediaries. These underlying differences between direct

and intermediary exporters have important consequences for trade flows. The ability of export intermediaries

to overcome country and product fixed costs means that they can more easily respond along the extensive

margin to external shocks. Intermediaries and direct exporters respond differently to exchange rate

fluctuations both in terms of the total value of shipments and the number of products exported as well

as in terms of prices and quantities. Aggregate exports to destinations with high shares of indirect

exports are much less responsive to changes in the real exchange rate than are exports to countries

served primarily by direct exporters.

Andrew B. Bernard

Tuck School of Business at Dartmouth

100 Tuck Hall

Hanover, NH 03755

and NBER

Dndrew.E.E[email protected]

Marco Grazzi

LEM Scuola Superiore Sant'Anna

Piazza Martiri della Liberta', 33

56127 Pisa, Italy

[email protected]

Chiara Tomasi

University of Trento

Via Inama, 5

38122 Trento, Italy

and LEM Scuola Superiore Sant'Anna

[email protected]

Intermediaries in International Trade

1 Introduction

The growing availability of ﬁrm-level international trade data has contributed to th e blooming

of both theoretical and empirical literatures highlighting the importance of ﬁrm heterogeneity

in aggregate trade ﬂows. Since the initial empirical papers of Bernard and Jensen (1995, 1999),

Roberts and Tybout (1997) and the theoretical models of Melitz (2003) and Bernard et al. (2003),

a major f ocus in international trade has been on the relationship between the characteristics of pro-

ducing ﬁrms, most notably productivity, and their participation in international trade. An emerging

stream of research has examined diﬀerences among trading ﬁrms (Bernard, Jensen, Redding and Schott;

2010b; Ahn et al.; 2011; Antr`as and C ostinot ; 2011). These papers emphasize that exporters in-

clude both manufacturing ﬁrms that organize the production and distribution of their goods abroad

as well as intermediaries that specialize in distribution in foreign markets.

In this paper, we examine the underlying factors that give rise to exports by intermediaries

and the consequences in terms of trade volumes and the margins of adjustment to external shocks.

Existing theoretical and empirical work on intermediaries in exporting emphasizes the importance

of country-sp eciﬁc ﬁ x ed export costs and the variation of intermediary export shares across desti-

nations. We extend that focus to include both broader country-level ﬁ x ed costs as well as charac-

teristics of the produ cts themselves.

More importantly, this paper also considers additional implications of the lower ﬁxed costs for

wholesale exporters. If export intermediaries do indeed face lower ﬁxed costs of exporting then

they should also more easily enter and exit export markets in the face of changing proﬁtability.

Firm-level export volumes shou ld respond diﬀerently for wholesale exporters and manufacturing

exporters. Aggregating to the country-level, th is ﬁrm-level variation implies diﬀerential changes in

exports between destinations served primarily by direct and indirect exporters.

Using Italian ﬁrm-level trade data, we investigate the importance of intermediaries (wholesalers)

in exports across destinations and products and examine how they diﬀer from manufacturing ﬁrms

that export directly. More than one quarter of all exporters are intermediaries and they account

for over 10 percent of Italian exports. However, there is substantial variation in the importance of

intermediaries across countries and products. New Zealand and China have intermediary export

shares near 9 percent (25th percentile) while Paraguay and Malawi are at the 75th percentile with

shares above 23 percent.

Intermediary exporters diﬀer in a number of dimensions from manufacturing ﬁrms that export

directly. They are smaller in terms of exports, sales and especially emp loyment as would be

expected since they are only involved in the cross-border distribution of the products and not the

production. However, wholesale exporters display h igher sales per emp loyee and comparable exports

per employee. On average, intermediary exporters reach fewer countries and ship more products

Intermediaries in International Trade

than do direct exporters. One important diﬀerence between wholesalers and manufacturers lies in

their tendency to ad d and drop products. Intermediaries add and drop prod ucts at much higher

rates than direct exporters. These ﬁrms face lower sunk costs of exporting and thus are able to

adjust their extensive margin more easily.

The existence of intermediaries suggests that they overcome barriers to international trade at a

lower cost than manufacturers for some range of goods and for some countries. We examine the role

of both countr y and product characteristics in the choice of the mode of export and the magnitude

of country-product exports. The ﬁnding of previous studies that country-speciﬁc ﬁxed export costs

are corr elated with the u se of export intermediaries is conﬁrmed in the Italian data. In addition, the

quality of the general contracting environment is related to the choice of mode of export. Exports

through an intermediary are more likely when the quality of the general contracting environment

of the country is weak. Product characteristics also play a role in determining the choice of export

mode. Lower contract intensity, greater product homogeneity, and higher produ ct-level sunk costs

of exporting are associated with a greater reliance on intermediaries in exporting.

The diﬀerences in ﬁxed costs across destinations and products give rise to variation in response

to common external shocks to proﬁtability such as exchange rates. Total exports by wholesalers

are less responsive to exchange rate changes precisely because wholesalers are better able to adjust

along the extensive margin. Given the big diﬀerence in the share of intermediated exports across

countries and products, these ﬁrm-level results suggest that there are potentially large, predictable

diﬀerences in how aggregate exports will respond to changes in the value of th e domestic currency.

We indeed ﬁnd that the responsiveness of aggregate exports is much greater in destinations served

primarily by direct exporters.

Existing theoretical and empirical work on exporting intermediaries is reviewed in S ection 2.

Section 3 describes the ﬁrm and country level data. Section 4 documents diﬀerences between direct

exporters and wholesalers. The r ole of country and product ﬁxed costs on the choice of export

mode and export values are examined in Section 5. Section 6 explores the response of exports both

at the ﬁrm level and in the aggregate to exchange rates shocks. Section 7 concludes.

2 Theoretical frameworks

Recent models of international trade emphasize the role that heterogeneity in productivity plays in

explaining the s tructure of international commerce. According to these models an d a large q uan-

tity of associated empirical work, more productive ﬁrms are more likely to engage in exporting and

foreign direct investment. While these frameworks have been extended to examine multiple desti-

nations and multiple products, they generally assume that trade occurs directly between producers

in one country and ﬁnal consumers in another and do not account for the activity of intermediary

ﬁrms in trade.

Intermediaries in International Trade

Early theoretical work on the role of intermediaries in international trade, e.g Rauch and Watson

(2004) and more recently Petropoulou (2007), models international trade as an outcome of search

and networks. Several new papers in the theoretical literature on intermediaries in exporting have

taken a more technological pers pective based on models of heterogeneous ﬁrms (Ahn et al.; 2011;

Akerman; 2010; Felb er mayr and Jung; 2011).

New models of trade, in particular Akerman (2010) and Ahn et al. (2011), extend the heteroge-

neous ﬁrm tr ad e model of Melitz (2003) by introducing an intermediation technology which allows

wholesalers to exploit economies of scope in exporting. While all active ﬁrm s serve the domestic

market, manufacturers have a choice of how to potentially serve a foreign market. Domestic man-

ufacturing ﬁrms are allowed to choose between direct exports to a consumer in the foreign market

and the use of an intermediary ﬁrm who controls the goods as they cross the international border.

While the details of the models vary, the general framework is similar. Exporting directly incur s

a ﬁxed cost and a variable cost. Indirect exporting takes place thr ou gh an intermediary ﬁrm, or

using intermediary ‘techn ology’. The intermediary is assumed to be able to lower the ﬁxed costs of

exporting while possibly incurring additional variable costs. This choice means that a number of

manufacturing ﬁrms may export indirectly through a wholesaler, rather than managing their own

distribution networks. These ﬁrms pay an intermediary ﬁxed cost which is smaller than their own

ﬁxed cost of direct export. In this more realistic setting, ﬁrms choose to serve the foreign market

either directly or through domestically-based export intermediaries.

Firms sort according to produ ctivity into diﬀerent export channels. As in the standard model of

Melitz (2003), the least productive ﬁrms ser ve only the domestic market while the most p roductive

ﬁrms can export directly by incurring the ﬁxed cost of export and any variable trade costs. A

third category of ﬁrms chooses to export indirectly through wholesalers. This third group, which

looks like non-exporters in the data, includes some ﬁrms who would not have been exporters in

the absence of intermediaries an d some ﬁrms who would be marginal exporters in the absence of

intermediaries.

Analogous to Helpman et al. (2004), we can compare graphically the proﬁts generated by each

type of activity for ﬁrms with diﬀerent productivity.

The two solid lines in Figure 1 depict proﬁts

from the domestic market (π

) and additional proﬁts f or ﬁ rms that export d irectly (π

). The

proﬁt functions are increasing in productivity (α) as more productive ﬁrms are able to charge a

Blum et al. (2011) and Blum et al. (2010) look the role of intermediaries largely from the perspective of the

imp orting country while Rauch and Watson (2004) discuss when intermediary ﬁrms actually take possession of the

goods.

In this example we assume that the ﬁrm itself has access to the intermediation technology. Akerman (2010)

mod els intermediaries explicitly in a monopolistic competition setting. Intermediaries face ﬁxed costs of exporting

that are increasing in the number of varieties handled by the exporter and their variable costs per variety include

tariﬀs and the domestic price of the variety. Producing ﬁrms view intermediaries as identical to any other domestic

consumer and thus only face domestic ﬁxed costs of production. The resulting pictures an d cutoﬀs are similar

although his framework allows for a richer set of predictions on the size and scope of intermediaries.

Intermediaries in International Trade

αα

xdxi

Figure 1: Proﬁts from domestic sales, indirect and direct exports

lower price, capture a large market share and generate larger proﬁts. The intercept of the domestic

curve is smaller in absolute value than that of exports because the ﬁxed costs that are incurred for

selling on the domestic market (f

) are lower than what a ﬁr m must pay to export directly abroad

). Moreover, since there is a per unit variable cost of export, the slope of the proﬁt function

for direct exports is ﬂatter than the slope of the proﬁt function for domestic production. These

relationships introduce two productivity cut-oﬀs (α

and α

), that in turn indicate which ranges

of productivity determine exit, domestic sales only, or dir ect exports.

With the possibility of exporting through intermediaries, ﬁrms now have also an additional

option of using the intermediation ‘technology’ to export. By assumption the ﬁx ed costs in the

intermediation technology are lower than the ﬁxed costs of direct exporting and are greater than or

equal to the ﬁxed costs of domestic sales; f

is between f

and f

in Figure 1. Th e degree to which

the intermediation ﬁxed costs are lower than those of direct exporting depends on the combination

of country, industr y and country-variety ﬁxed costs of selling in the foreign market as discussed

further below.

The dotted curve drawn in Figure 1 depicts proﬁts for ﬁrms that export indirectly (π

) through

an intermediary. If using an intermediary does not raise the variable costs of exporting then all

manufacturers would employ the intermediation technology and export indirectly, π

(α) > π

(α)

Intermediaries in International Trade

∀α. To allow for both direct and indirect exporting, the intermediary exporter faces additional

variable costs. In Akerman (2010), the intermediary sets the export price of each variety as a

standard mark-up over its own marginal cost, where its marginal cost includes both variable trade

costs an d the domestic p urchase price of the variety, which is itself a mark-up over the variable

cost of production. In Ahn et al. (2011), it is assumed that intermediaries face no ﬁxed costs of

exporting but charge a variable cost to transport the good s.

The combination of lower ﬁxed costs an d higher variable costs at intermediaries introduces a

third productivity cut-oﬀ, α

, which is the zero-proﬁt cutoﬀ for exporting through an intermediary.

If α

< α

then there will be an equilibrium with ‘pure’ domestic producers and both direct

and indirect exporting. Firms with productivity levels below α

earn negative proﬁts and exit the

industry. Firms with productivity levels between α

and α

, produce only for the d omestic market.

Firms with productivity between α

and α

now can proﬁtably access the foreign market thr ou gh

wholesalers. Finally, ﬁrms with productivity levels above α

produce for th e domestic market

and export directly. Note that the group of ﬁrms with indirect exports includes some ﬁrms with

productivity too low to ﬁnd it proﬁtable to export directly, α

≤ α < α

and some ﬁrm s of higher

productivity that prefer indir ect to direct exporting, α

≤ α < α

A ﬁrm’s decision regarding the mode of export is determined by variable and ﬁxed trade costs,

which in turn also depends on country and product characteristics. The degree to which ﬁxed costs

are reduced using intermediaries depends on the nature of the ﬁx ed cost, e.g. the combination of

country, industry and country-variety components. We can write the ﬁxed costs of direct exporting

of variety k in industry j to country c as

= f

+ f

where f

is a ﬁxed export cost common to all varieties exported to country c, f

is a ﬁxed export

cost common to all varieties in industry j regardless of the number of destinations, and f

is a ﬁxed

export cost s peciﬁc to the variety and country. The greater the share of idiosyncratic ﬁx ed costs,

, in total ﬁxed costs, f

, the lower the possibility for economies of scope and the lower the share

of exports handled by intermediaries. Both country and industry-speciﬁc ﬁxed costs allow for the

possibility of indirect exporting. Ex porting intermediaries may arise because they are able to share

the country-speciﬁc ﬁxed cost of exporting across many industr ies and varieties and/or they may

exist because they are able to spread industr y -speciﬁc ﬁxed costs across varieties and destinations.

Existing theoretical frameworks typically ignore the possibility of industry-speciﬁc ﬁxed costs but it

remains an empir ical question as to whether intermediaries are country- or industry-speciﬁc relative

to direct exporters.

It is possible that no produ cer will choose to export through an intermediary if the increase in variable cost is

suﬃciently large.

Intermediaries in International Trade

The simple framework provides some clear predictions for th e variation of direct and indirect

trade across countries. To the extent that intermediaries solve only the country-speciﬁc ﬁxed costs

of exporting, e.g. each variety exported faces indirect ﬁxed costs f

= f

/n + f

, where n is the

number of varieties handled by the intermediary, the diﬀerence between direct an d indirect ﬁxed

costs will be increasing as country ﬁxed costs rise.

The role of variable trade costs is less clear-cut in these models. A rise in variable trade costs that

aﬀects both direct and intermediary exporters, such as tariﬀs or tran sportation costs, can increase,

decrease or leave un changed the share of exports handled by intermediaries. In the empirical work

we examine the role of variable trade costs including distance and tariﬀs in determining the share

of exports handled by intermediaries.

The existing theoretical frameworks emphasize the interaction of producer ﬁrm heterogene-

ity and ﬁx ed export costs in th e decision to export directly or indirectly. While these models

are all static models of single-prod uct ﬁr ms, it is relatively easy to envision a dynamic exten-

sion where ﬁrms potentially make multiple p roducts and their proﬁtability evolves over time (see

Bernard, Redding and Schott; 2010; Bernard et al.; 2011). In a dynamic environment, variation in

the sunk cost of exporting across ﬁrm types would lead to predictable variation in product adding

and product dropping in the export market. Firms facing lower s unk costs would be more likely

to both add and drop products in steady state and in the face of exogenous shocks to proﬁtability.

As intermediary exporters have lower entry costs they should be more likely to churn their export

product mix.

2.1 Related empirical literature

Recent papers by Ahn et al. (2011), Akerman (2010) and Bernard, Jensen, Redding and Schott

(2010b) examine various aspects of intermediaries in exports for China, Sweden and the US, re-

spectively. None of the papers uses exactly the same deﬁ nition of an exporting intermediary so the

results are not directly comparable to each other or those presented below.

Bernard, Jensen, Redding and Schott (2010b) docu ment the role of intermediaries in US ex-

ports. They ﬁnd th at 35 percent of US exporters are wholesalers accounting for 10 percent of

US exports by value. Their work emphasizes the diﬀerences in the attributes between exporters

of diﬀerent types. Among exporting ﬁrms, pure wholesalers are much smaller than ‘producer-

consumer’ ﬁrms in terms of employment, but only slightly smaller in terms of exports per worker

Speciﬁcally, Ahn et al. (2011) deﬁne an intermediary as a ﬁrm with certain Chinese characters in its

name, Akerman (2010) uses the main activity of the ﬁrm and includes both wholesalers and retailers and

Bernard, Jensen, Redding and Schott (2010b) distinguish between pure wholesalers, pure retailers and two types

of ﬁrms that mix manufacturing with wholesaling and retailing. As discussed below we only consider ﬁrms with

wholesaling as their main activity as intermediaries.

Intermediaries in International Trade

and domestic sales per worker.

Other diﬀerences include the types of products exported and the

destinations served, wholesalers are more likely to export food-related sectors and export to lower

income countries.

Akerman (2010) reports slightly more exporting intermediaries than manufacturers and signif-

icant diﬀerences between the two types of exporters. Intermediaries are sm aller in terms of total

turnover, much smaller in terms of export value, but export more products and ship to more desti-

nations. Akerman (2010) regresses country-sector intermediary export sh ares on gravity variables

and proxies for country ﬁxed export costs. Intermediary export shares increase in distance and

measures of ﬁxed costs and fall with destination GDP.

In contrast with the other studies, Ahn et al. (2011) ﬁnd much higher exports per ﬁr m and

unit values for intermediaries than for direct exporters. Interm ediaries are also active in m any

more products than direct exporters. Regressions of p roduct-country intermediary export shares

on country characteristics s how positive relationships for distance, tariﬀs and a measure of ﬁxed

costs and a negative r elationship with destination GDP.

This paper builds on this growing empirical literature and extends it in a number of directions.

First, it documents the diﬀerences between produ cing exporters and intermediary exporters in

terms of their ﬁrm characteristics, destination and product m ixes, product churning and export

values and quantities. We then examine export participation and levels by direct and intermediary

exporters across countries and products and their relation to country and product characteristics.

Finally the underlying s ou rces of intermediated trade are shown to cause intermediaries to diﬀer

in terms of their responses to aggregate shocks, both in terms of export value and th e margins of

adjustment.

3 Data

3.1 Trade and Firm data

The analysis of dir ect versus indirect modes of export is based upon two ﬁrm-level datasets collected

by the Italian statistical oﬃce (ISTAT), namely S tatistiche del Commercio Estero (COE) and

Archivio Statistico Imprese Attive (ASIA).

The COE dataset consists of all cross-border trans actions performed by Italian ﬁrms and it

covers the period 2000-2007. COE includes the annual value and quantity of export tr an sactions

by the ﬁrm for product-country destination pairs.

A product is deﬁned as a six digit category in

‘Producer- consumer’ ﬁrms in Bernard, Jensen, Redding and Schott (2010b) include any ﬁrm with no reported

employment in wholesaling or retailing and thus include both manufacturers and other service ﬁrms.

The database has been made available for work after careful screening to avoid disclosure of individual information.

The data were accessed at the ISTAT facilities in Rome.

ISTAT, collects data on exports based on transactions. The European Union sets a common framework of rules

but leaves some ﬂexibility to member states. A detailed description of requirements for data collection on exports in

Italy is provided in the Appendix .

Intermediaries in International Trade

the Harmonized System (HS6).

Using the unique identiﬁcation code of th e ﬁrm, we link the ﬁrm-level export data to ISTAT’s

archive of active ﬁrms, ASIA. In ASIA, ﬁrms are classiﬁed according to their main activity, as iden-

tiﬁed by ISTAT’s standard codes for sectoral classiﬁcation of business (5-digit ATECO). This infor-

mation allows us to distinguish between four broad categories of ﬁrms: manufacturers, wholesalers,

retailers, and a residual group including the remaining sectors.

ASIA also contains information on

ﬁrms’ operations including the number of employees and total turnover.

The combined dataset

used for the analysis is not a sample but rather includes all active ﬁrms.

3.2 Country level data

Firm-level trade data are complemented by country characteristics including proxies for market size

and variable and ﬁxed trade costs.

For market size we use total GDP from the World Bank World

Development Indicators database. Variable trade costs may be either due to policy barriers, such as

tariﬀs and non-tariﬀ barriers, or to the cost of moving goods across borders, such as transportation

costs. Following the large gravity literature, transportation costs are proxied by geographic distance

calculated using the great circle formula (Mayer and Zignago; 2005).

As emphasized in the literature on ﬁrms and exporting (Roberts and Tybout; 1997; Melitz;

2003; Bernard and Jensen; 2004; Bernard et al.; 2007; Eaton et al.; 2011), ﬁrms incur ﬁxed entry

costs in order to enter foreign markets. These ﬁxed costs can be related to the establishment of a

foreign distribution network, diﬃculties in enforcing contractual agreements, or the uncertainty of

dealing with foreign bureaucracies. We create two measures of country-level ﬁxed costs. To generate

a proxy for the market-speciﬁc ﬁxed costs of exporting to a country, we u se information from three

measures from the World Bank Doing Business dataset: number of documents for importing, cost

of importing and time to import (Djankov et al.; 2011). Given the high level of correlation between

these variables, we use the primary factor (Market Costs) derived from pr incipal component analysis

as that factor accounts for most of the variance contained in the original indicators (see Table A1

in Appendix).

Data on the contracting environment are available from a variety of sources, e.g. World Bank,

Heritage Foundation, an d Transparency I nternational. To proxy for institutional quality we use

information from the six variables in the World Bank’s Governance dataset (Kaufman et al.; 2009):

Voice and Accountability, Political Stability and Absence of Violence/Terrorism, Government Ef-

fectiveness, Regulatory Quality, Rule of Law, and Control of Corruption. As these six measures are

highly correlated, we follow Bernard, Jensen, Redding and Schott (2010a) and use the primary fac-

In particular, we classify ﬁrms in sectors from 151 to 372 as manufacturers, and ﬁrms in sectors from 501 to 519

(with the exclusion of 502 which concerns the activity of repair of motor vehicles) as wholesalers. Retailers are ﬁrms

in sectors 521 to 527, and Others contains the remaining sectors.

Information on total turnover are available only for two years, 2000 & 2003.

More details on the country-level variables are available in the Appendix.

Intermediaries in International Trade

tor obtained from principal component analysis, Governance, as the proxy for country governance

quality.

If ﬁr ms must invest in ﬁxed resources to export to countries w ith weaker contracting

environments, one would expect better Governance to be associated with lower intermediary export

shares.

Finally, in order to account for the eﬀect of policy barriers on the presence of intermediaries and

manufacturers we also consider HS6 product-country im port tariﬀs, taken from World Integrated

Trade System (WITS).

3.3 Product level data

The paper also investigates a set of product and in dustry variables that aﬀect the probability that

a producer exports directly rather than through an intermediary. Goods with high destination-

speciﬁc costs of entering a foreign market, or ﬁnding a foreign customer, are more likely to be

exported directly. Similarly in dustries with higher entry costs are less likely to be served by direct

exports.

We consider both produ ct characteristics that are related to the speciﬁcity of the product

and those more generally related to market structure. A measure of industry contract intensity

develop ed by Nunn (2007) is used to measure the importance of relationship-speciﬁc investment in

intermediate inputs across industries. Nunn’s original data, corr esponding to US I-O in dustries,

is concorded to HS6 products.

Industries that require more relationship-speciﬁc investments are

expected to be less easily served by intermediaries as the product-market compon ent of ﬁxed costs

is relatively large.

In order to account for diﬀerentiation within a HS6 pr oduct class we employ the coeﬃcient

of price dispersion.

Lower price dispersion is assumed to be associated with m ore homogeneous

products. For homogeneous pr oducts, the product-market component of ﬁxed costs will be lower

and thus it is more likely that the export transactions will be carried out by an intermediary.

We adapt a measure of product-level sunk entry costs developed by Bernard an d Jensen (2007)

to the export market. In steady state, a produ ct with high sunk costs of entry into export markets

should have a low entry rate and a low (and equal) exit rate. During transitions between steady

states, either the entry rate (expanding product) or the exit rate (shrinking product) may be

Table A2 in Appendix reports the results of the principal component analysis for the governance measure.

WITS contains the TRAINS database on bilateral tariﬀs at the six-digit level of the Harmonized System (HS)

product classiﬁcation for about 5,000 products and 200 countries. TRAINS provides information on four diﬀerent

type of tariﬀs: Most-Favored National Tariﬀs (MFN), Preferential Tariﬀs (PRF), Bound Tariﬀs (BND) , and the

eﬀectively applied tariﬀs (AHS). We use the AHS tariﬀ in the empirical analysis. The AHS tariﬀ is the MFN Applied

tariﬀ, unless a preferential tariﬀ exists.

See the Data Appendix for a description of the concordance procedure.

The coeﬃcient of p rice variation is computed on COE data as the coeﬃcient of variation in the unit values of any

of the HS6 products across all ﬁrm-product-country transactions. In th e empirical analysis we use data from 2003,

but the product ranking in terms of price dispersion does not vary much over the years.

Intermediaries in International Trade

unusually high. However, the minimum of the two r ates should still correspond to steady-state

entry or exit. We calculate the m inimum of the ﬁrm-level export entry and exit rates for each

product, min(entry, exit).

A higher level of entry and exit indicates lower sunk costs of exporting

and a lower likelihood that the prod uct will be exported through an intermediary.

4 Manufacturers and Intermediaries

The focus of the present work is to investigate the role of intermediaries in exports. This section

documents the extent of intermediation in Italian exports, highlighting important stylized facts

about intermediaries and showing how they diﬀer from manufacturing ﬁrms. Table 1 reports the

total value of exports and the relative share of four broad categories of ﬁrms: manufacturers,

wholesalers, retailers, and a residual group including ﬁ rms in all the remaining sectors.

A prepon derance of exports, more than 84 percent of the value, is perform ed directly by man-

ufacturing ﬁrms. Manufacturing exporters also represent more than 50 percent of exportin g ﬁrms.

However, an increasing share of exp orts is conducted by the 27 percent of exporters that are whole-

salers, rising from 9.9 percent in 2000 to 11.3 percent of Italian exports in 2007. These ﬁgur es

are in line with those reported for the US in Bernard, Jensen, Reddin g and Schott (2010b) wh er e

wholesalers are 35 percent of exporting ﬁrms and control just over 10 percent of US exports. As

in other countries, retailers are relatively minor players in exporting, accounting for less than one

percent of exports by value. As a result the paper focuses on the role of wholesalers as export

intermediaries and uses the two terms interch an geably.

While intermediaries account for just 11 percent of Italian exports, there is s ubstantial variation

across both countries and products, see Table 2. At the country level, intermediary export shares

range fr om a low of zero to a high of 88 percent. At the bottom of the interquartile range are

countries such as Belgium, Norway, France, New Z ealand and China with intermediary export

shares close to 9 per cent; at the top of the interquartile range, we ﬁnd Paraguay, Moldova, Malawi

and Albania with wholesale export shares near 23 percent. While the overall share of intermediary

exports is just u nder 11 percent in 2003, across destinations, unweighted intermediary export shares

average 16.6 percent and are higher on average for non -EU countries. This indicates th at wholesalers

are relatively more important in smaller markets and in markets outside the EU.

The share of intermediaries across prod ucts also displays substantial variation, see the second

panel of Table 2. Wholesalers account for 21 percent of the exports for the average product, pointing

to the importance of intermediaries in p roducts with lower total export values. While there exist

The entry rate is the number of new exporters of the product between year t and t+s divided by the average

number of exporters in the two years. The exit rate is the number of ﬁrms that stop exporting the product between

t and t+s divided by the average numb er of exporters in the two years. The min(entry, exit) in a given product is

computed on COE data for years 2003 and 2007. Considering diﬀerent years for the computation of the rates does

not signiﬁcantly aﬀect the results.

Intermediaries in International Trade

both products which are sold ab road only through intermediaries, 1.8 percent of 5,125 products,

and others wh er e the share of wholesalers is zero, most products are exported both directly and

indirectly.

Specialization is more common at the product-country level. Of the 244,614 product-country

combinations with positive exports, 48.6 percent involve direct exports only and 10.4 percent are

served exclusively by intermediaries.

4.1 Firm characteristics

In their work on US traders, Bernard, Jensen, Redding and Schott (2010b) ﬁnd not only that

traders diﬀer from domestic ﬁrms, but also that substantial h eterogeneity exists between trad-

ing ﬁrms of diﬀerent ‘types’. The results here complement and extend that analysis by comparing

manufacturers and wholesalers along a number of dimensions including size, the number of desti-

nation countries and the number of products exported.

The top left panel of Figure 2 shows the distribution of employment for all wholesale and

manufacturing ﬁrms. Th e employment distribution for wholesalers lies far to the left of that for

manufacturers. Overall, intermediaries are much sm aller in terms of number of employees. However,

when we proxy size with total sales (top right panel) the diﬀerence between the two distributions

remains but is greatly reduced. The diﬀerences between the panels implies that the sales per

employee ratio of wholesalers is much h igher than that of manufacturers.

The bottom panels

of Figure 2 show the size distrib utions for wholesale and manufacturing exporters. The relative

ranking of the two distributions is similar to that seen above.

The ﬁgures are consistent with the idea that manufacturing ﬁrms are performing two activi-

ties, the physical p roduction of the goods and the intermediation of the goods to a downstream

customer, while wholesalers are only engaged in the latter activity. This distinction is important

when attempting to compare the exporting activities of wholesalers and manufacturers as the use

of employment as a proxy for ﬁrm size may yield misleading comparisons. A manufacturing ﬁrm

with 100 emp loyees will typically have lower sales and exports than a wholesale ﬁrm with the same

employment. As a consequence, we use both employment and total sales as proxies for size in the

analysis.

Figure 3 displays the binned relation between log exports and log employm ent, reporting the

(log) number of employees a ﬁrm needs, on average, for a certain level of exports.

The plot

For product-country pairs with a mix of direct and indirect exports the average indirect share is 25.3 percent.

We caution that sales per employee is not a good m easure of ﬁrm product ivity when comparing ﬁrms of d iﬀerent

types.

Binned plots allow for a succinct representation of the relation between two variables and avoid displaying clouds

of thousands of observations. Here data are placed in 20 equally-sized bins according to th eir (log of) export value,

and the x-coordinate d isplays the average of the bin. The y- coordinate is the average (log of) employment within

that bin.

Intermediaries in International Trade

.0001

.001

.01

Probability Density

2 4 6 8 10

Log(Number of Employees)

Manufacturers

Wholesalers

Size Distribution

.0001

.001

.01

Probability Density

5 10 15 20 25

Log(Sales)

Manufacturers

Wholesalers

Size Distribution

.0001

.001

.01

Probability Density

2 4 6 8 10

Log(Number of Employees)

Manufacturers

Wholesalers

Size Distribution for Exporting firms

.0001

.001

.01

Probability Density

5 10 15 20 25

Log(Sales)

Manufacturers

Wholesalers

Size Distribution for Exporting firms

Figure 2: Empir ical density of ﬁr m size in 2003 - All ﬁ rms (Top) and Exporters (Bottom). S ize is

proxied by (log of) employment (Left) an d (log of) sales (Right). Densities estimates are obtained

using the Epanenchnikov kernel with the b an dwidth set using the optimal routine described in

Silverman (1986).

conﬁrms that wholesalers require fewer employees to attain any given level of export value.

To quantify the diﬀerences between manufacturers and wholesalers, we estimate the following

cross-sectional OLS regression,

ln Y

= c + δD

+ βD

+ γ(D

· D

) + ε

(1)

where ln Y

denotes the logarithm of either total sales, number of employees, or sales per employee

ratio. D

is a ﬁrm-level dummy variable, one f or wholesaler and zero for manufacturer; D

a dummy indicating if a ﬁrm is an exporter; and (D

∗ D

) is the interaction between the two

dummies and takes value of one if a ﬁrm is a wholesaler exporter and zero otherwise. The results

are presented in Table 3.

As expected, manufacturers are on average larger than wholesalers, 0.111 log points (12 per-

cent) in terms of sales an d 0.533 log points (70 percent) in terms of emp loyment, δ is negative and

Intermediaries in International Trade

1.5

2.5

3.5

4.5

5.5

6 8 10 12 14 16 18

ln Employment

ln Exports

Employment per level of exports

Manufacturers

Wholesalers

Figure 3: Relation between (log of) employment and exports, 2003. Observations are placed in 20

equally-sized bin s according to the variable on x-axis. Coordinates of dots display the average of x

and y variables of the data in each b in (see text).

signiﬁcant in both speciﬁcations. In contrast, sales per employee are substantially higher at whole-

salers. We also conﬁrm the now-standard results that manufacturing exporters are dramatically

larger and have higher sales per employee than their domestic counterparts, β is large, positive and

signiﬁcant.

Perhaps unsurprisingly, we provide the ﬁr st evidence that the selection of ﬁrms into exporting

may be working for wholesalers as well. Exportin g wholesale ﬁ rms have total sales 14.8 times

larger than non-exporting wholesalers and employ 2.8 times as many workers, β + γ is positive and

signiﬁcant. Sales per employee at exportin g intermediaries are 5 times higher than at non-exporters.

Looking at exports in columns 4 and 5 of Table 3, we ﬁnd that the value of exports at wholesalers

is also much smaller than that of manufacturing exporters but that this diﬀerence largely disappears

when considering exports per employee.

The regression results of Table 3 conﬁrm the conclusions from the relative distributional plots in

Figure 2. In particular, the evidence on higher sales per employee, especially at exporters, supports

the idea that wholesalers focus on just the intermediation portion of the activities carried out by

manufacturers.

4.2 Product and Geographic Diversity

The theoretical models discussed in Section 2 generally focus on the role of intermediaries in solv-

ing the ﬁxed cost problem for speciﬁc markets. This section prov ides evidence on the presence

of intermediaries in markets and sectors. Figure 4 displays th e relation between geographic and

product diversiﬁcation of the ﬁrm and its size, distinguishing between wholesalers and manufactur-

ers. Geographic diversiﬁcation is proxied by the number of destination countries (Countries

) and

product diversiﬁcation by the number of products exported (P roducts

); size is represented both

Intermediaries in International Trade

0 1 2 3 4 5 6

Number of countries

ln Employment

Number of Countries and Employment

Manufacturers

Wholesalers

6 8 10 12 14 16 18

Number of countries

ln Exports

Number of Countries and Exports

Manufacturers

Wholesalers

0 1 2 3 4 5 6

Number of Products

ln Employment

Number of Products and Employment

Manufacturers

Wholesalers

6 8 10 12 14 16 18

Number of products

ln Exports

Number of Products and Exports

Manufacturers

Wholesalers

Figure 4: Top Number of countries and (left) employment and (right) exports, in 2003. Bottom

Number of products and (left) employm ent and (right) exports, in 2003. Observations are placed

in 20 equally-sized bins according to the variable on x-axis. Coordinates of dots display the average

of x and y variables of the data in each bin (see text).

by employment and export value.

The evidence in Figure 4 suggests that the wholesalers’ technology does not convey them an

advantage in terms of geographic diversiﬁcation, wholesalers export to fewer countries than do

manufacturers at similar levels of employment and exports.

On the contrary, when considering

the relation between ﬁrm size and product diversiﬁcation (bottom panel), we ﬁnd that, at every

size class, wholesalers export more produ cts than manufacturers.

Table 4 reports the results of the regression of the number of products exported and the number

of destination markets (Products and Countries, respectively) on the ﬁrm wholesaler dummy, D

and a proxy f or ﬁrm size,

= c + δD

+ ln Size

+ ε

if D

= 1. (2)

Ahn et al. (2011) report that Chinese intermediaries export more products and export to more countries than

direct exporters. However, as noted previously, Chinese intermediary export ﬁrms are almost twice as big as direct

exporters in terms of total export value.

Intermediaries in International Trade

The ﬁrs t row of Column 1 shows that, unconditionally, wholesale exporters export fewer HS6

products. However, including a control for ﬁrm size, either log employment or log export value, the

coeﬃcient becomes positive and signiﬁcant; exporting intermediaries are active in a wider range of

products compared to similarly-sized manufacturers. In contrast, intermediaries serve fewer export

markets even when adjusting for ﬁrm size. These resu lts su ggest that intermediaries are indeed

able to spread country-speciﬁc ﬁxed costs over a wider range of products.

4.3 Within Product-Country

The availability of product level data allows the comparison of wholesalers and manufacturing

exporters within product-country destinations.

Using exports to Extra-EU d estinations for 2003

and considering product-country pairs where both wholesalers and manufacturers are active, we

estimate the following speciﬁcation,

ln Y

fcp

= c + δD

+ β ln Sales + d

+ ε

fcp

(3)

where ln Y

fcp

denotes the logarithm of, respectively, th e total value, quantity and unit value of

the ﬁrm’s exports in the country-product pair, D

is the ﬁrm wholesaler dummy and d

denotes

country-product ﬁxed eﬀects. The results in the ﬁrst two columns of Table 5 show that wholesalers

have a substantially lower total value of exports relative to direct exporters within product-country

pairs. The diﬀerence in exports across ﬁrm types remains even after controlling for ﬁrm size,

although the magnitude is reduced. Columns 3-6 report similar regressions for export qu antities

and unit values. The lower exports for wholesalers are driven entirely by lower export quantities;

unit values are not statistically diﬀerent for direct and intermediary exporters.

4.4 Product Adding and Dropping

The cross-sectional analysis reveals that exporting wholesalers are smaller than manufacturers and

that they export a larger number of products to a sm aller set of destinations. In addition to lower

ﬁrm-level trade ﬂows, intermediaries also ship less within a product-country p air. These results

are broadly supportive of a framework emph asizing country-speciﬁc ﬁxed costs of exporting. As

discussed earlier, the presence of s unk export costs that vary across ﬁrm types also has implications

for export dynamics. Lower sunk costs should result in higher probabilities of both entry into

exporting and exit from exporting. A dynamic extension of the framework presented earlier would

suggest that intermediaries should be more likely to add and drop products from their export

portfolio than direct exporters. This is conﬁrmed by the unconditional drop rates across ﬁrms

We focus all the remaining empirical work on exports to Extra-EU destinations for several reasons. Most im-

portantly, ﬁrm-level exports to the EU are not recorded for all exporters and these criteria have changed over time.

Also, real exchange rate changes within the eurozone countries are driven entirely by changes in relative price levels.

Intermediaries in International Trade

types: on average the fraction of exported ﬁrm-produ cts outside the EU that is dropped every year

is 50% among all ﬁrms, 48% for manufacturers and 53% for wholesalers.

Following Bernard, R edding and Schott (2010), we analyze export product switching between

t and t + 1 using those years for which we have information on ﬁrms’ total turnover, 2000-2001 and

2003-2004. We estimate a linear probability model of product dropp ing for ﬁrms that export in

year t and t + 1 of the form,

Drop

fpt

= c + δD

+ β

ln Sales

+ β

Deviation

fpt

+ β

ln P roducts

+ d

+ ε

fpt

(4)

where Drop

fpt

takes value 1 if the product is exported by the ﬁrm f in year t and not exported in

year t +1 an d equal zero if the produ ct is exported in both years. D

is the ﬁrm wholesale dummy

and is the variable of interest. To control for ﬁrm attributes associated with product switching, we

include ﬁrm size, ln Sales

, the relative importance of the ﬁrm in the exports of the product given

by the log diﬀerence between the ﬁrm’s exports in product p and average ﬁrm exports in product p ,

Deviation

fpt

, and the number of products exported by the ﬁrm in year t, ln P roducts

. Product

and year ﬁxed eﬀects, d

and d

respectively, are also included.

Table 6 reports the results of the estimation of the ﬁrm-pr oduct droppin g speciﬁcation equa-

tion 4 for Extra-EU countries. Wholesaler exporters are much more likely to drop a pr oduct than

manufacturer exporters, 6.9 percentage points or 14.4 percent. This diﬀerential persists even con-

trolling for ﬁrm size, the number of exported products and the relative importance of the ﬁrm in

the product, although the magnitude of the coeﬃcients is reduced.

If wholesalers have lower sunk costs per product, then they s hould also be more likely to add

products.

We examine the probability that a current exporter adds a product to its export

portfolio between years t and t + 1 in the speciﬁcation,

Add

= c + δD

+ β

ln Sales

+ β

ln P roducts

+ d

ind

+ d

+ ε

(5)

where Add

takes value 1 if the ﬁrm adds an export product and zero otherwise. D

, ln Sales

and ln P roducts

are deﬁned as above. Add itional controls include year ﬁxed eﬀects, d

, and

industry-mix ﬁxed eﬀects, d

ind

that controls f or ﬁrm s with the same mix of industries at the HS2

level.

Results of Table 7 show that intermediaries are more likely to add a product th an manufacturers.

This ﬁnding is robust to controlling for ﬁrm size and number of exported products, and the eﬀect

is more pronounced when comparing wholesalers and manufacturers among single product ﬁrms.

On average the fraction of ﬁrms that add at least one export product outside the EU every year is 79%: 79% for

manufacturers and 80% for wholesalers. Among single product (multiple products) ﬁrms the ratio is 65% (83%) for

all ﬁrms, 62% (83%) for manufacturers and 69% (83%) for wholesalers. On average the fraction of ﬁrms that drop

at least one export p roduct outside t he EU every year is 80%: 80% for manufacturers and 79% for wholesalers.

While the dropping regression was estimated at the ﬁrm-product level, the adding speciﬁcation is at th e ﬁrm-level.

Intermediaries in International Trade

b=−0.011 (0.003)

0 .1 .2 .3 .4 .5

Intermediary Export Share

15 20 25 30

Log (GDP)

Linear Fit Observed Value

Wholesale Export Share & Market Size

b=−0.0013 (0.008)

0 .1 .2 .3 .4 .5

Intermediary Export Share

6 7 8 9 10

Log (Distance)

Linear Fit Observed Value

Wholesale Export Share & Market Distance

Figure 5: Wholesale export share and gravity variables, 2003. Figures report the relationship be-

tween wholesale export share and gravity variables: (Left) Real GDP; (Right) Geographic distance.

Each panel reports the coeﬃcient, b, of a country-level univariate regression for intermediary export

share. Robust standard error is shown in parenthesis.

The results on both export product dropping and export produ ct adding suggest that interme-

diary exporters face lower sunk costs of participation in the export market. These ﬁndings suggest

that shocks such as changes in tariﬀs or exchange rates may have diﬀerential eﬀects on wholesalers

and manufacturers even within the same country-product pair. We return to examine the eﬀects

of exchange rate changes across ﬁrm types in the ﬁnal s ection of the paper.

5 Exports by Intermediaries

The previous sections have shown that exporting wholesalers diﬀer from manufacturing exporters

in terms of size, geographic coverage, product portfolio and entry and exit. This section focuses on

diﬀerences in the role of wholesalers across countries and products.

5.1 Intermediary E xport Share

We start by exploring the relationship between the intermediary export sh are by destination market

and a set of relevant country variables (Figures 5-6). The correlation of intermediary export shares

by country with market size and distance is displayed in the two panels of Figure 5. Wh olesale

export share is declining in log GDP, smaller markets have greater intermediary export shares,

consistent with the idea that in s maller destination markets, ﬁxed entry costs have to be spread

over fewer units. In contrast, there is no statistically signiﬁcant relationship between distance, a

common proxy for variable trade costs, and the intermediary export share.

Country-speciﬁc ﬁxed costs of trade are generally expected to be positively related to inter-

Intermediaries in International Trade

b=0.033 (0.007)

0 .1 .2 .3 .4 .5

Wholesale Export Share

−2 0 2 4

ln (Market Costs)

Linear Fit Observed Value

Wholesale Export Share & Market Costs

b=−0.035 (0.005)

0 .1 .2 .3 .4 .5

Intermediary Export Share

−2 −1 0 1 2

Governance Indicator

Linear Fit Observed Value

Wholesale Export Share & Governance Indicator

Figure 6: Wholesale export share and country-level ﬁxed costs, 2003. Figures report the relationship

between wholesale export share and the two proxies for ﬁxed market entry costs: (Left) Market Size;

(Right) Governance indicator. Each panel reports the coeﬃcient, b, of a country-level univariate

regression for intermediary export share. Robust standard error is shown in parenthesis.

mediary tr ad e shares.

The plot at the left of Figure 6 displays the relationship between the

percentage of export value that goes through intermediaries and the Market Costs variable. As

found by Ahn et al. (2011) and Akerman (2010 ), this measure of market access costs is positively

and signiﬁcantly related to intermediary trade shares.

The right panel of Figure 6 plots the intermediaries export share against country Governance.

As expected, the quality of country governance is negatively and signiﬁcantly related to intermedi-

aries export share. This evidence supports the idea that as country-level ﬁxed costs increase, more

ﬁrms use wholesalers for exporting.

Finally, we investigate the link between the HS6 p roduct characteristics and intermediary export

shares. While the theoretical models remain largely silent on th is aspect, product characteristics

would be expected to play a r ole in explaining the type of ﬁ rm handling the exports.

If goods w ith

higher relation-speciﬁcity have relatively larger product-country ﬁxed costs of exporting, the share

of direct exports is likely to be greater. Transactions involving complex goods, whose production

process is intensive in the use of highly specialized and customized inputs, may require speciﬁc

knowledge and tasks because of the eﬀort associated with the identiﬁcation of potential customers,

more detailed contracts, post-sale service, etc. For those goods, the product-market component of

ﬁxed costs is relatively large and such goods are more likely to be exported directly by manufacturing

Higher country-level ﬁxed costs of exporting and weaker governance are associated with smaller total levels of

exports (Lawless; 2010; Djankov et al.; 2011), here we consider their relationship to the comp osition of exports by

ﬁrm type.

While not discussed explicitly in his paper, Akerman (2010) models the p rice of exports by intermediaries as

a double mark-up over tariﬀ-adjusted marginal cost. Increases in the d emand elasticity reduce the mark-ups and

narrow the diﬀerence between the export p rices of intermediaries and th ose of direct exporters and increase the share

of exports by intermediaries.

Intermediaries in International Trade

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Wholesale Export Share

Relation Specificity

Wholesale Export Share & Relation Specificity

b=-0.236 (0.023)

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

0 1 2 3 4 5 6 7 8 9

Wholesale Export Share

Price dispersion

Wholesale Export Share & Coeff. of Variation

b=-0.020 (0.001)

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Wholesale Export Share

Entry/Exit rate

Wholesale Export Share & min(entry,exit)

b=-0.116 (0.017)

0.15

0.16

0.17

0.18

0.19

0.2

0.21

0.22

0.23

0.24

0.25

0.26

0 5 10 15 20 25 30 35 40 45 50

Wholesale Export Share

Tariff

Wholesale Export Share & (all product-country) tariffs

b=0.0003 (2.959e-05)

Figure 7: Wholesale export share an d Product/Country-Product characteristics, 2003. Figures

display the relationship between wholesale export share and the following characteristics: (Top

Left) Relation Speciﬁcity; (Top Right) Coeﬃcient of Variation of the unit values for each prod uct;

(Bottom Right) min(entry, exit) in the export market for a given product; (Bottom Left) Country-

Product export tariﬀs. Observations are placed in 20 equally-sized bins. Coordinates of dots d isplay

the average of tariﬀs and intermediary export share of the data in each bin. Each panel reports

the coeﬃcient, b, of a product-level univariate regression for intermediary export share. Robust

standard error is shown in parenthesis.

Intermediaries in International Trade

ﬁrms. On the contrary, the indirect mode of export would prevail if the traded good does n ot

require a relation-speciﬁc investment, as for commoditized products. T his prediction is in line

with the hypothesis put f orward by Peng and Ilinitch (2001) “the higher the commodity content of

the product, th e more likely that export intermediaries will be selected by manufacturers”. This

is conﬁrmed by Figure 7 (top left) which shows a negative and s igniﬁcant relationship between

intermediary export share an d the measure of relation speciﬁcity. Note that, given the very large

number of ob servations, data are binned in all plots of Figure 7, although the regression coeﬃcients

are based on all the data.

The relation between p roduct price dispersion, as pr oxied by the coeﬃcient of variation of

export unit values, and intermediary share is negative and signiﬁcant, Figure 7 top right. This is

the expected relationship if lower price dispersion is associated with more homogeneous products

which are more likely to be hand led by intermediaries.

The plot at th e bottom left of Figure 7 displays the r elation between min(entry, exit) rate in

a product and intermediary export sh are. The negative and s igniﬁcant slope suggests that easier

export entry and exit is associated with a lower export share f or wholesalers. Products that have

higher sunk costs of entry (low rates of entry/exit) are more likely to be handled by intermediaries.

Finally we consider the incidence of tariﬀs on the presence of intermediaries in product-country

pairs. The bottom right of Figure 7 shows the relation between product-country tariﬀ and in-

termediary export share. There is a small, positive relation between product-country tariﬀs and

intermediary share.

The overall message of these ﬁgures is consistent with the idea that there is a systematic

relationship between the share of exports m an aged by wholesalers and both country and product

characteristics.

5.2 Selection into diﬀerent modes of export

The previous section has investigated the univariate correlations of country and product charac-

teristics with the intermediary export s hare. This s ection and the next explore the relationship

between country and product characteristics and export participation and levels by wholesalers and

manufacturers.

Recent work on the importance of distance and GDP on the level of exports has emphasized

the role of ﬁrm selection, e.g. Helpman et al. (2008). The empirical literature on intermediation

has not considered selection eﬀects. The iss ue is m ore diﬃcult than the standard export problem in

that there are potentially diﬀerent selection criteria for intermediary and d irect exporters. Ideally,

we would s ee the characteristics of the producing ﬁrm and th e choice of whether or not to export

and by what method. However, the trade data limits us to observing whether or not a product is

exported by wholesalers alone, manufacturers alone or by both methods. We pursue an admittedly

Intermediaries in International Trade

imperfect solution by ﬁrst examining the relationship between pr oduct and country attributes and

the type of exporter and then examinin g how the value of exports varies between wholesalers and

manufacturers.

The theoretical framework outlined in Section 2 gives some guidance to the selection problem.

Four mutually exclusive situations might arise as ﬁxed costs of exporting increase. If such costs are

zero, or very low, m anufacturers will choose to export directly. For higher ﬁxed costs, some ﬁrms

will ﬁnd it proﬁtable to export in directly while some, more productive ﬁrms still export directly.

With suﬃ ciently high ﬁx ed costs, a third outcome is possible where all ﬁrms go through wholesalers.

At the extreme, prohibitive ﬁxed costs will result in no exports in th e product-country pair.

We rank the categories in order of increasing diﬃculty in entering the export market (direct

only, both dir ect and indirect, indirect only) and estimate an ordered probit model to investigate

the eﬀects of country and product variables in determining the probability of being in one of the

three states,

Category

= c + β

+ β

+ ε

(6)

where Category

accounts for the possible three ou tcomes described above, C

and P

are country

and prod uct variables, and τ

is tariﬀ to a particular country-product destination. Results are

reported in the ﬁrst column of Table 8. While most of the coeﬃcients have the predicted sign and

are signiﬁcant, the proportional odds approach is easily rejected as indicated by the χ

test.

Given the rejection of the ordering of outcomes, column s 2 and 3 of Table 8 report the results

for a multinomial logit speciﬁcation, where the baseline category is exports by manufacturers only.

Country GDP is positively associated with both type of ﬁrms serving the market but is negatively

associated with the likelihood of the market being served by wholesalers alone. Geographical

distance lowers the prob ab ility of both groups exporting relative to only manufacturing; comparing

only wholesalers to only manufacturing we ﬁnd the same negative sign with a lower magnitude.

The measures of ﬁxed costs of exports, Market Costs and Governance, behave as predicted by

the theory. Market Costs show the expected positive sign; higher market costs increase the probabil-

ity of both categories of ﬁrms exporting with r espect to the baseline category, only manufacturing,

and also increase the probability that wholesalers are the only exporters in the pr oduct-country

pair. Better country governance reduces the likelihood that the market will be served by both

wholesalers and manufacturers. Similarly, the Governance measure is negatively associated with

only wholesaling.

Looking at the product characteristics, we ﬁnd the expected sign on the min(entry, exit) p roxy

for product-level sunk costs. Lower sunk costs reduce the probability of both categories exporting

relative to the baseline category and the same relation holds also for wholesalers versus manufac-

turers only. The eﬀects of product price disper s ion and relation speciﬁcity are less clear and depend

Intermediaries in International Trade

on the groups being compared.

5.3 Product-country exports

This section investigates the eﬀects of country and product characteristics on the level of exports

of manufacturers and wholesalers for product-country pairs. Columns 1 to 3 of Table 9 report

regression results with the log of country-product exports by exporter type, ln X

, as dependent

variable and country and product characteristics, C

and P

, as explanatory variables, together

with a full set of interactions w ith the wholesaler dummy, D

ln X

= c + δD

+ β

+ γ

∗ D

+ β

+ γ

∗ D

+ β

+ γ

∗ D

+ d

+ ε

. (7)

Columns 1 and 2 add country and product ﬁxed eﬀects, respectively, while column 3 includes all

the available product and country characteristics.

Results on the country characteristics in column s 1 and 3 suggest that the level of exports of

both manufacturers and wholesalers exports is positively correlated with GDP, however the eﬀects

are signiﬁcantly lower for wholesalers. In contrast to Ahn et al. (2011), geographical distance aﬀects

negatively the value of trade equally for both types of ﬁrms.

The results on Market Costs and Governance are in line with the theoretical predictions. Inter-

mediaries’ exports increase with market costs, suggesting that wholesalers are better able to spread

ﬁxed costs across produ cts. The country governance indicator yields a similar pattern of results:

better governance is associated with higher exports from manufacturers but that eﬀect is greatly

reduced or disappears entirely for wholesalers.

Columns 2 an d 3 of Table 9 report the r esults on the p roduct characteristics. We focus on

the sign and signiﬁcance of the interaction term s with the wholesaler dum my. Wholesalers export

relatively less in products w ith lower sun k entry costs, i.e. greater min(entry, exit), higher price

dispers ion, and higher relationship speciﬁcity. All th ese coeﬃcients have the expected signs and

point to product characteristics playing an important part in the endogenous choice of ﬁrms to

export directly or through an intermediary.

Finally, we include tariﬀs in column 3. As expected, the coeﬃcient of tariﬀs exhibits a negative

sign; ﬁrms export less in country-product pairs with higher tariﬀs. The wholesaler interaction is

positive but not signiﬁcant.

This section has examined th e role of country and product characteristics in the choice of

the mode of export and the magnitude of country-product exports. Results conﬁr m the ﬁndings

of p revious studies th at country-speciﬁc ﬁxed export costs are correlated with the use of export

In Column 3 we cluster both on countries and products using cluster2 packages for Stata (Petersen; 2009).

The speciﬁcation of Ahn et al. (2011) is slightly diﬀerent as they include a smaller set of covariates and do not

include the interacted wholesale dummy.

Intermediaries in International Trade

intermediaries. We further show that the quality of the more general contracting environment is

related to the choice of mo de of export. Exports through an intermediary are more likely wh en the

quality of the general contracting environment of the country is weak.

We also show that the characteristics of the product play a role in determining the choice of

export mode. Lower contract intensity, greater product homogeneity, and higher produ ct-level sunk

costs of exporting are associated with a greater reliance on intermediaries in exporting.

6 Intermediaries and exogenous shocks

We have documented the variation in the share of indirect exports across countries and products.

The results support the id ea that export intermediaries arise in large p art because of the presence

of s igniﬁcant ﬁxed export costs at the country and product level. In addition, product adding an d

product dropping in the export market are greater for wholesale exporters than for manufacturers.

Taken together this evidence suggests that a common sh ock to proﬁts across destinations, e.g.

a common tariﬀ cut, may have diﬀerent eﬀects both across types of exporting ﬁrms and in the

aggregate across countries due to variation in the composition of exporters.

This section examines whether intermediaries and manufacturers respon d diﬀerently to exoge-

nous cur rency shocks. Using annual ﬂuctuations in bilateral real exchange rates as measures of

exogenous changes in export proﬁtability, we investigate th e eﬀects on ﬁrms’ export behavior. We

consider the impact of exchange rate changes on ﬁr m exports to country c, as well as on the number

of exported products and the average value of exports to that destination. Following Bernard et al.

(2007) a ﬁrm’s total exports to a destination can be decomposed into extensive and intensive

margins,

ln X

= ln P rod

+ ln avgX

(8)

where ln X

is th e log of total exports by ﬁrm f to country c, P rod

is th e number of distinct

HS6 products exported by ﬁrm f to country c, and avgX

is th e average exports per exporting

ﬁrm f to country c. We regress the annual log change from 2000 to 2007 of ﬁrm total exports to

country c and the annual changes of the two components on a dummy for wholesaler (D

), the

change in the log of the real bilateral exchange rate of the Italian currency (∆ln RER

) and their

interaction

∆ln X

fct

= c

+ δ

+ β

∆ln RER

+ γ

∆ln RER

∗ D

+ d

+ ε

fct

(9)

∆ln P rod

fct

= c

+ δ

+ β

∆ln RER

+ γ

∆ln RER

∗ D

+ d

+ ε

fct

(10)

∆ln avgX

fct

= c

+ δ

+ β

∆ln RER

+ γ

∆ln RER

∗ D

+ d

+ ε

fct

(11)

Intermediaries in International Trade

where d

indicates a set of ﬁ x ed eﬀects. Using data from the International Financial Statistics

database (IMF, 2010), we deﬁ ne the RER

index for each year:

RER

= ER

CP I

where ER

is the nominal Italian exchange rate expressed as the number of foreign currency

units per home currency unit and

CP I

is the ratio of the domestic consum er price level and

the consumer price index abroad.

An upward (downward) movement therefore represents an

appreciation (depreciation) of the domestic currency.

Table 10 reports results from estimating equations 9-11. Since real exchange rate variations

inside the Eurozone are related only to price levels changes and given the relevant role of wholesalers

in Extra-EU destinations, we include in the regressions only countries ou tside the E U.

The ﬁrst two columns of Tab le 10 present the r esults for export value, including country and year

ﬁxed eﬀects (column 1) and country and ﬁrm ﬁxed eﬀects (column 2). Exchange rate movements

have the expected eﬀects on ﬁrm exports to country c: an appreciation of the euro currency is

associated with a decrease in ﬁrm exports. However, the interaction of wholesaler type and the real

exchange rate is positive and signiﬁ cant in both columns; ﬁrm exports fall less (3.7-8.4 percent) for

intermediaries than for manufacturers when the Italian currency appreciates.

Looking at columns 3-6 we observe that, for both manufacturing and wholesale ﬁrms, the fall in

exports in respon se to an appreciation of the domestic currency is driven both by a decrease in the

number of products exported and by a decline in the ﬁr m’s average exports per country. However,

for w holesalers, the adju stment on the extensive margin of the number of products is greater,

while the response of aver age exports is more muted. These results would appear to conﬁrm that

wholesale exporters face lower ﬁxed costs and are thus able to adjust more easily along the extensive

margin than direct exporters.

We next explore the sensitivity of the ﬁrm’s response within a country-product pair to annual

exchange rate movements by considering export value, qu ality and unit value. The estimation

equation is

∆ln Y

fpct

= c

+ δD

+ β

∆ln RER

+ γ∆ln RER

∗ D

+ d

+ ε

fpct

(12)

where ln Y

fpct

is the log of ﬁrm-level product-country export value, quantity or unit value. Columns

1, 3 and 5 of Table 11 report results with country, product and year ﬁxed eﬀects, while in columns

2, 4 and 6 replace product dummies with ﬁrm-product ﬁxed eﬀects. As before, exports fall as

the Italian currency appreciates but the eﬀect for wholesalers is 15-30 percent sm aller. For direct

exporters the ad justment to a stronger home cur rency is primarily due to reductions in export

Details on the exchange rate variables are available in the Appendix.

Intermediaries in International Trade

quantities (90 percent) rather than in unit value (10 percent). For wholesalers, the overall adjust-

ment is smaller due to a much smaller quantity response. Wholesalers drop their unit values more

as the currency rises, pass-through is lower, and quantities fall less.

The literature on intermediaries in trade largely has focused on the underlying choice of ﬁrms

to export directly or indirectly and in particular on the role of ﬁxed export costs. This section has

shown that these choices give rise to diﬀerent respon ses to common external shock s to proﬁtability.

Wholesalers are less responsive to exch an ge rate changes precisely because they are better able to

adjust along the extensive margin.

6.1 Aggregate Exports

The ﬁrm-level results presented above suggest that the en dogenous choice of direct or in direct

exporting by producing ﬁrms should matter for the response of aggregate, country-level export

volumes to exogenous shocks s uch as changes in the exchange rate. Destinations with high wholesale

export shares should show smaller responses of exports in response to exchange rate changes than

countries with low wholesale export shares.

In Table 12 we consider a simple speciﬁcation of the form

∆ln Y

= c

+ δD

+ β

∆ln RER

+ γ∆ln RER

∗ D

+ d

+ ε

(13)

where ln Y

is the log of country exports and D

is a dummy that equals one if the country-level

share of wholesale exports is greater than the median (mean), and the exchange rate is deﬁned as

before. Columns 1 and 3 report r esults with year ﬁxed eﬀects, while columns 2 and 4 include both

year and country ﬁxed eﬀects.

In every case, the results strongly conﬁrm the importance of the mode of export in sh ap ing the

aggregate responses to changes in the real exchange rate. The exchange rate export elasticity for

countries with low wholesale shares is negative and signiﬁcant, ranging from -0.232 to -0.499 across

the speciﬁcations. In contrast, countries with wholesale export s hares above the mean or median

have elasticities that are insigniﬁcantly diﬀerent from zero.

7 Conclusions

The present paper examines the role of intermediaries in exporting, the factors that lead ﬁrms to

export indirectly, and the consequences of intermediary exporters on tr ad e value and the margins

of adjustment to external shocks. Using Italian ﬁrm-level trade data, we investigate the impor-

tance of wholesalers in exports across destinations and products an d examine how they diﬀer from

manufacturing ﬁ rms that export dir ectly. Intermediary exporters are smaller, ship more products

and reach fewer countries than direct exporters.

Intermediaries in International Trade

We conﬁrm the ﬁndings of previous research that w holesalers are more likely to export to

countries with high ﬁxed export costs and to smaller m arkets. However, exporting by wholesalers

is also more common in destinations with weak contracting environments and in products that

are more homogeneous, have higher sun k entry costs and have lower relationship speciﬁcity. The

ability of intermediaries to eﬀectively lower destination and product ﬁxed costs means that they

are to churn their product mix more often and adjust along the extensive margin.

The diﬀerences in ﬁxed costs across destinations and products h ave important implications

for ﬁrm-level and aggregate responses to exogenous changes in proﬁtability such as exchange rates.

Wholesalers are more likely to adju st their product mix in r esponse to an exchange rate change and

their total exports adjust less. Given the big diﬀerence in the share of intermediated exports across

countries and products, these ﬁrm-level results suggest that there are potentially large diﬀerences

in how aggregate exports will respond to changes in the value of the domestic cur rency that are

linked to the type of the exporting ﬁrm. We ﬁnd signiﬁcantly lower r esponses of aggregate exports

to changes in the exchan ge rate for destinations served primarily by wholesale exporters.

Intermediaries in International Trade

Table 1: E x ports and Number of exporting ﬁrms: share by type of ﬁrms, 2000-2007

Year Total Exports Manuf Whol Retail Others

(billion) Share (%)

2000 246.79 85.09 9.85 0.74 4.32

2001 258.99 86.49 9.88 0.86 2.76

2002

260.75 84.75 10.93 0.83 3.49

2003 254.91 85.52 10.71 0.86 2.91

2004

274.38 85.65 10.5 0.82 3.91

2005 286.56 85.5 10.75 0.85 4.91

2006 319.01 84.95 11.32 0.85 5.91

2007

350.57 85 11.27 0.84 6.91

Year Exporters Manuf Whol Retail Others

(N. of ﬁrms) Share (%)

2000 137347 57.3 26.43 7.67 8.6

2001 141520 56.46 27.01 7.95 8.58

2002 145473 55.64 27.06 8.14 9.16

2003

143421 55.57 27.41 7.72 9.3

2004 139598 55.34 27.61 7.46 10.3

2005 133473 54.96 27.48 7.3 11.3

2006

139360 53.7 28.07 7.31 12.3

2007 128472 54.77 27.91 6.88 13.3

Note: Table reports the share of exports and the share of exporters by type of ﬁrms (Manufacturers, Wholesalers,

Retailers and Others).

Intermediaries in International Trade

Table 2: Descriptive statistics of Wholesale export share at Country, P roduct and Country-product

level, 2003

Obs Zeros Ones Mean Median

All sample 228 8 0 .166 .133

Country Intra-EU 14 0 0 .118 .109

Extra-EU 214 8 0 .170 .137

All Sample 5125 226 95 .211 .098

Product Intra-EU 5009 579 156 .204 .056

Extra-EU 5011 332 129 .220 .116

All Sample 244614 118891 25506 .208 .001

Country-Prod I ntra-EU 51274 17717 3559 .187 .014

Extra-EU 193340 101174 21907 .213 0

Table 3: Export premia, 2003

ln Sales

ln Employment

ln Sales/Empl.

ln Exports

ln Exports/Empl.

(1) (2) (3) (4) (5)

-0.111*** -0.533*** 0.433*** -1.047*** -0.025

(0.004) (0.002) (0.003) (0.016) (0.015)

2.775*** 1.533*** 1.229***

(0.007) (0.005) (0.004)

· D

-0.081*** -0.489*** 0.388***

(0.012) (0.008) (0.008)

R-squared 0.22 0.29 0.14 0.03 0.001

Observations 985719 1022424 985710 118994 118994

Notes: Table reports OLS regression of noted characteristic on dummy for wholesaler (D

), dummy for exporter

), and their interaction (D

· D

). Robust standard errors are reported below coeﬃcients. Asterisks denote

signiﬁcance levels (***: p <1%; **: p<% 5%; *: p<10%). Data are for 2003.

Intermediaries in International Trade

Table 4: E x port premia: Number of Countries and Number of Products, 2003

Products

Countries

-1.269*** 3.005*** 1.668*** -4.562*** -0.158*** -1.630***

(0.093) (0.118) (0.088) (0.058) (0.053) (0.043)

ln Employment

4.180*** 4.307***

(0.070) (0.036)

ln Exports

2.805*** 2.801***

(0.027) (0.015)

R-squared 0.001 0.13 0.25 0.03 0.27 0.45

Observations 118994 118994 118994 118994 118994 118994

Notes: Table reports OLS estimates of the number of HS6 products exported (Products

) and the number of desti-

nation countries (Countries

) on a dummy for wholesaler (D

). The regression sample is exporting ﬁ rms. Robust

standard errors are reported below coeﬃcients. Asterisks denote signiﬁcance levels (***: p < 1%;**: p <5%; *: p<%

10%). Data are for 2003.

Table 5: Firm’s exports, quantity and unit value by product an d country by diﬀerent type of ﬁrms,

2003 - Extra-EU

ln Exports

f cp

ln Exports

f cp

ln Quantity

f cp

ln Quantity

f cp

ln UnitValue

f cp

ln UnitValue

f cp

(1) (2) (3) (4) (5) (6)

-0.307*** -0.113*** -0.314*** -0.115*** 0.007 0.002

(0.011) (0.010) (0.015) (0.015) (0.010) (0.010)

ln Sales

0.196*** 0.201*** -0.005

(0.003) (0.005) (0.004)

Country-Product FE Yes Yes Yes Yes Yes Yes

Clustering Firm Firm Firm Firm Firm Firm

Adj R-squared 0.15 0.19 0.42 0.44 0.63 0.63

Observations 1190313 1190313 1190313 1190313 1190313 1190313

Countries 184 184 184 184 184 184

HS6 Products 4042 4042 4042 4042 4042 4042

Firms 105649 105649 105649 105649 105649 105649

Note: Table reports results of regressions at the ﬁrm product country level, using data on exports, quantity and u nit

value for 2003 and Extra-EU destinations only. D

is a dummy for wholesaler; Sales is ﬁrm’s total sales. Only

product-country pair in which both wholesalers and manufacturers are both active are included. Robust standard

errors clustered at ﬁrm level are reported in parenthesis below t he coeﬃcients. Asterisks denote signiﬁcance levels

(***: p <1%; **: p<5%; *: p<10%).

Intermediaries in International Trade

Table 6: Product dropping each year (2000&2003) by diﬀerent type of ﬁr ms, Extra-EU.

Drop

fpt

Drop

fpt

Drop

fpt

Drop

fpt

(1) (2) (3) (4)

0.069*** 0.043*** 0.017*** 0.021***

(0.001) (0.001) (0.001) (0.001)

ln Sales

-0.034*** -0.010*** -0.004***

(0.000) (0.000) (0.000)

Deviation

fpt

-0.099*** -0.099***

(0.000) (0.000)

ln Products

-0.013***

(0.001)

Year FE Yes Yes Yes Yes

Product FE Yes Yes Yes Yes

Clustering Firm-Product Firm-Product Firm-Product Firm-Product

Adj R-squared 0.06 0.07 0.19 0.19

Observations 1221737 1221737 1221737 1221737

HS6 Products 5259 5259 5259 5259

Firms 110452 110452 110452 110452

Note: Table reports OLS regression results of a dummy variable indicating a ﬁrm-p roduct drop between t an d t + 1.

f t

is a dummy for wholesaler; Sales

f t

is ﬁrm’s total sales; Deviation

f pt

is (log of) ﬁrm’s exports in product p minus

(log of) average exports in product p; and Products

f t

is the number of product s exported by each ﬁrm. All variables

are computed at time t. The regression sample is surviving exporting ﬁrms. Robust standard errors in parentheses

are adjusted for clustering by ﬁrm-product. Asterisks denote signiﬁcance levels (***: p <1%; **: p <5%; *: p<10%).

Intermediaries in International Trade

Table 7: Adding regression (2000&2003) by diﬀerent type of ﬁrms, Extra-EU

All ﬁrms SPF MPF All ﬁ rms SPF MPF All ﬁrms MPF

Add

f t

Add

f t

Add

f t

Add

f t

Add

f t

Add

f t

Add

f t

Add

f t

(1) (2) (3) (4) ( 5) (6) (7) (8)

f t

0.026*** 0.072*** 0.010** 0.031*** 0.071*** 0.017*** 0.036*** 0.022***

(0.005) (0.008) (0.004) (0.006) (0.009) (0.005) (0.006) (0.004)

ln Sales

f t

0.023*** 0.009*** 0.026*** 0.013*** 0.012***

(0.002) (0.003) (0.002) (0.002) (0.002)

ln Products

f t

0.057*** 0.085***

(0.006) (0.005)

Year FE Yes Yes Yes Yes Yes Yes Yes Yes

Industry-Mix FE Yes Yes Yes Yes Yes Yes Yes Yes

Clustering Industry-Mix Yes Yes Yes Yes Yes Yes Yes Yes

Adj R-squared 0.006 0.110 0.002 0.013 0.111 0.003 0.021 0.003

Observations 167081 31175 135906 167081 31175 135906 167081 135906

Firms 110452 28304 90041 110452 28304 90041 110452 90041

Industry-mix 32383 88 32382 32383 88 32382 32383 32382

Note: Table reports O L S regression results of a dummy variable indicating a ﬁrm adding a product between t and

t + 1. D

f t

is a dummy for wholesaler; Sales

f t

is ﬁrm’s total sales; and Products

f t

is the number of products exported

by each ﬁrm. SPF and MPF are, respectively, single and multi product ﬁrms. All variables are computed at time t.

The regression sample is surviving exporting ﬁrms. Industry-mix FE allows to control for ﬁrms with the same mix

of industries at the HS2 level. Robust standard errors in parentheses are adjusted for clustering by industry-mix.

Asterisks denote signiﬁcance levels (***: p<1%; **: p<5%; *: p<10%).

Intermediaries in International Trade

Table 8: L ogistic Regression, 2003 - Extra-EU

Ordered Probit Multinomial Logit

Category (1) Category (1)

vs vs

Category (2) Category (3)

ln GDP

0.061*** 0.162*** -0.169***

(0.003) (0.004) (0.006)

ln Distance

-0.332*** -0.474*** -0.152***

(0.005) (0.006) (0.011)

Market Costs

0.147*** 0.069*** 0.212***

(0.011) (0.012) (0.019)

Governance Indicator

-0.032*** -0.100*** -0.250***

(0.008) (0.009) (0.017)

Tariﬀ

0.047** 0.027 0.093***

(0.019) (0.024) (0.031)

min(entry,ex it)

-0.465*** -0.420*** -0.652***

(0.033) (0.036) (0.070)

Coeﬃcient of Variation

0.04*** 0.077*** -0.035***

(0.002) (0.002) (0.006)

Relation Speciﬁcity

0.349*** 0.876*** -0.649***

(0.045) (0.051) (0.074)

Observations 134322 134322

Observations Category 1 65775 65775

Observations Category 2 58556 58556

Observations Category 3 9991 9991

Test for prop. odd model

(Prob > χ

) 0.000

Note: Table reports ordered probit and multinomial logit regression of diﬀerent categories of product-country combinations.

Category (1) product-country in which only manufacture exports; Category (2) product-country in which both manufacture

and wholesale export; Category (3) product-country in which only wholesale exports. Categories (1), (2) and (3) contains

resp ectively, 4957, 4467 and 4624 products. Standard errors are reported in parenthesis below the coeﬃcients. Asterisks denote

signiﬁcance levels (***: p<1%; **: p<5%; *: p<10%). Data are for 2003. Results of the ordered probit suggests that the

propor tional odds approach is not appropriate since the χ

test is statistically signiﬁcant.

Intermediaries in International Trade

Table 9: Total exports by country-product, 2003 -Extra-EU

ln X

(1) (2) (3)

3.208*** -0.869*** 4.432***

(0.847) (0.141) (0.900)

ln GDP

0.487*** 0.370***

(0.102) (0.073)

∗D

-0.189*** -0.194***

(0.039) (0.039)

ln Distance

-0.503*** -0.276***

(0.120) (0.086)

∗D

-0.012 0.003

(0.060) (0.060)

Market Costs

-0.117 -0.100

(0.105) (0.085)

∗D

0.111* 0.103*

(0.072) (0.060)

Governance Indicator

0.264*** 0.134**

(0.099) (0.070)

∗D

-0.181*** -0.189***

(0.063) (0.063)

Tariﬀ

-0.165**

(0.068)

∗D

0.058

(0.043)

min(entry,exit)

-0.710*** -0.660 ***

(0.155) (0.171)

∗D

-0.305** -0.309**

(0.119) (0.128)

Coeﬃcient of Variation

0.101*** 0.103***

(0.013) (0.014)

∗D

-0.028*** -0.040 ***

(0.008) (0.009)

Relation Speciﬁcity

1.212*** 1.223***

(0.226) (0.275)

∗D

-0.798*** -0.929 ***

(0.140) (0.186)

Country FE No Yes No

Product FE Yes No No

Clustering Country HS6 Product Country-Product

Adj R-squared 0.44 0.25 0.24

Observations 117112 117112 1 17112

Countries 142 142 142

HS6 Products 3623 3623 3623

Note: Table reports OLS regression of logarithm of aggregate exports by type for Extra-EU. D

is a dummy for wholesale

and ∗D

is the interacted dummy. Robust standard errors clustered at diﬀerent levels are reported in parenthesis below the

coeﬃcients. Asterisks denote signiﬁcance levels (***: p<1%; **: p<5%; *: p<10%). Data are for 2003.

Intermediaries in International Trade

Table 10: E x change rates and ﬁrm-country exports (1 and 2), number of p roducts (3 and 4), average

exports (5 and 6) over time, by diﬀerent type of ﬁrms, Extra-EU

Annual Diﬀerences

ln X

f ct

ln X

f ct

ln Prod

f ct

ln Prod

f ct

ln Avg X

f ct

ln Avg X

f ct

(1) (2) (3) (4) (5) (6)

f t

-0.015*** -0.001 -0.014***

(0.004) (0.002) (0.003)

ln Real Ex Rate

-0.519*** -0.461*** -0.186*** -0.086** -0.333*** -0.375***

(0.150) (0.121) (0.047) (0.037) (0.107) (0.089)

∗D

f t

0.042* 0.017* -0.046** -0.046* 0.087** 0.064*

(0.026) (0.011) (0.023) (0.028) (0.039) (0.038)

Country FE Yes Yes Yes Yes Yes Yes

Year FE Yes No Yes No Yes No

Firm FE No Yes No Yes No Yes

Clustering Country-Year Country-Year Country-Year Country-Year Country-Year Country-Year

Adj R-squared 0.004 0.005 0.004 0.001 0.003 0.001

Observations 2483679 2483679 2483679 2483679 2483679 2483679

Countries 149 149 149 149 149 149

Firms 137262 137262 137262 137262 137262 137262

Note: Table reports results of regressions at the ﬁrm country level, using data on exports, number of products and

average exports b etween 2000 and 2007. The dependent and independent variables are deﬁned as annual diﬀerences.

f t

is a dummy for wholesaler and ∗D

f t

is the interacted dummy. Robust standard errors clustered at country-year

level are reported in parenthesis below the coeﬃcients. Asterisks denote signiﬁcance levels (***: p<1%; **: p<5%;

*: p<10%).

Intermediaries in International Trade

Table 11: Exchange rates and ﬁrm’s exports, qu antity and unit value by produ ct and country over

time, by diﬀerent type of ﬁrms, E x tra-EU

Annual Diﬀerences

ln X

f cpt

ln X

f cpt

ln Quantity

f cpt

ln Quantity

f cpt

ln UnitValue

f cpt

ln UnitValue

f cpt

(1) (2) (3) (4) (5) (6)

f t

-0.020*** -0.018*** -0.002***

(0.003) (0.004) (0.001)

ln Real Ex Rate

-0.321*** -0.385*** - 0.287*** -0.353*** -0.035*** -0.032***

(0.095) (0.113) (0.100) (0.117) (0.011) (0.011)

∗D

f t

0.072* 0.065* 0.092** 0.090** -0.020* -0.025*

(0.039) (0.041) (0.041) (0.042) (0.012) (0.014)

Country FE Yes Yes Yes Yes Yes Yes

Product FE Yes No Yes No Yes NO

Year FE Yes Yes Yes Yes Yes Yes

Firm-Pr oduct FE No Yes No Yes No Yes

Clustering Country-Year Country-Year Country-Year Country-Year Country-Year Country-Year

Adj R-squared 0.004 0.004 0.004 0.001 0.004 0.001

Observations 4008339 4008339 4008339 4008339 4008339 4008339

Countries 150 150 150 150 150 150

Firms 119201 119201 119201 119201 119201 119201

HS6 Products 5201 5201 5201 5201 5201 5201

Note: Table reports results of regressions at the ﬁrm product country level, using data on exports, quantity and u nit

value between 2000 and 2007. The depend ent and independent variables are deﬁned as annual diﬀerences. D

f t

a dummy for wholesaler and ∗D

f t

is the interacted dummy. Robust standard errors clustered at country-year level

are reported in parenthesis below the coeﬃcients. Asterisks denote signiﬁcance levels (***: p<1%; **: p<5%; *:

p<10%).

Intermediaries in International Trade

Table 12: Exchange rates and country exports, Extr a-EU

Annual Diﬀerences

ln X

(Above) Median Median Mean Mean

(1) (2) (3) (4)

0.0215 -0.004

(0.029) (0.032)

ln Real Exchange Rate

-0.269** -0.499*** -0.232** -0.460***

(0.145) (0.162) (0.115) (0.145)

∗D

0.253* 0.511*** 0.224** 0.497***

(0.150) (0.164) (0.117) (0.147)

Year FE Yes Yes Yes Yes

Country FE No Yes No Yes

Observations 1489 1489 1489 1489

Adj R-squared 0.021 -0.057 0.020 -0.06

R-squared 0.028 0.062 0.028 0.063

Countries 160 160 160 160

Note: Table reports results of regressions at the country-year level, using data on exports between 2000 and 2010.

The dependent and independent variables are deﬁned as annual diﬀerences. D

is a dummy that takes value 1 if

the intermediary export share to country c is above the med ian (mean) value of intermediary export share across

countries ∗D

is the interacted dummy. Robust standard errors are reported in parenthesis below the coeﬃcients.

Asterisks denote signiﬁcance levels (***: p<1%; **: p<5%; *: p< 10%).

Intermediaries in International Trade

Appendix

Firm trade data

ISTAT collects data on export transactions, which are the basic unit of observation for trade ﬂows.

It is then possible to link tran sactions to ﬁrms u sing the value added tax identiﬁcation code (partita

IVA) of the ﬁrm which is also recorded in th e transaction.

There are diﬀerent requirements in

order for a transaction to be recorded. These requirements depend on the destination, Intra or

Extra-EU, and on the value of the tran sactions. The European Un ion sets a common framework

but leaves some ﬂexibility to member states.

As far as Extra-EU transactions are concerned there is a good deal of homogeneity among

member states as well as over time. Since the adoption of the euro as a common currency, Italy set

the threshold at 620 euro (or 1,000 Kg), so th at all transactions bigger than 620 euro (or 1,000 Kg)

are recorded. All these records of Extra-EU transactions report complete inf ormation, that is, also

information about the product. From 2007 onward the threshold is at 1,000 euro (or 1,000 Kg).

Most of the existing diﬀeren ces are due to varying Intra-EU requirements. In 2003 there were

two reporting thresh olds : 200,000 euros

and 40,000 euros. Firms with more than 200,000 euros

of exports (based on the previous year) have to ﬁll in the Intrastat document monthly. They report

complete information includ ing details about products. Firms with exports between 40,000 and

200,000 euros have to ﬁll in the Intrastat form on a quarterly basis. The value of exports is recorded

but not information on products. Below 40,000 euros per year the transactions are not recorded.

Country and product data

In Section 3.2 we complement the ﬁrm-level trade data with country characteristics including proxies

for market size and variable and ﬁxed trade costs.

To proxy transportation costs we use data on geographic distance taken from CEPII. Distances

are calculated following the great circle formula, which uses latitudes and longitudes of the most

important city (in terms of population) or of the oﬃcial capital, if diﬀerent.

As a pr oxy f or policy barriers we use a measure of country-level import tariﬀs. Tariﬀ data are

taken from World Integrated Trade System (WITS), a project jointly developed by the World Bank

and UNCTAD. WITS contains the TRAINS database on bilateral tariﬀs at the six-digit level of

the Harmonized System (HS) product classiﬁcation for about 5,000 products and 200 countries.

To generate a proxy for the market-speciﬁc ﬁxed costs of exporting to a country, we use in -

formation from the World Bank Doing Business database (DB). The World Bank compiles pro-

cedural r equirements for importing a standardized container of goods by ocean transport. All

The value-added tax identiﬁcation number also allows the linking of export data to various Censuses conducted

by ISTAT.

In 2007 this threshold was raised to 250,000.

Intermediaries in International Trade

documents needed by the exporters and importers in each country to trade goods across the border

are recorded, along with the time and cost necessary for completing the transaction (for details,

see Djankov et al.; 2011). For the purpose of the analysis three variables are used: number of

documents for importing in cludes all documents required per shipment to import th e goods from a

given destination; cost of importing measures the fees levied on a 20-foot container in US dollars;

time to import reﬂects the number of days needed to import a standard container of goods from a

factory in the largest business city to a ship in the most accessible port. Data are available from

2004 to 2010. Given the low variability of these indicators, we take the average value over the

available years.

To compute the real exchange rate used in Section 6 we use data from the International Financial

Statistics dataset (IMF, 2010). The oﬃcial exchange rate available at IMF refers to the exchange

rate determined by national authorities or to the rate determined in the legally sanctioned exchange

market. It is calculated as an annual average based on monthly averages. Consumer price index

reﬂects changes in the cost of a basket of goods and services that may be ﬁxed or changed at

speciﬁed intervals. The L aspeyres formula is generally used to compute such price index.

In addition to country characteristics, product and industry characteristics are also taken into

account. The ﬁrs t variable that we consider is a measure of ind ustry contract intensity developed by

Nunn (2007) to measure the importance of relationship-speciﬁc investment in intermediate inputs

across industr ies. Nunn’s data are classiﬁed according to the industry classiﬁcation of the US I-

O table compiled by the Bureau of Economic Activity. To match each I-O industry to an HS6

product, ﬁrst we use information from Lawson et al. (2002) to construct a concordance between

I-O industry classiﬁcation and NAICS1997 cod e. Then the data are converted from NAICS1997

to NAICS2002. Finally, we exploit the concordance between Harmonize System Codes and NAICS

Industries developed by Pierce and Schott (2009) to obtain the information on contract intensity

at the level of HS6 product.

Principal Components

Tables A1 and A2 report the principal component analysis (PCA) on standardized variables for

Market Costs and Governance, respectively.

Because principal component is intended to study correlation patterns, it is ﬁrst necessary to

standardize the variables of interest so that they all have the same variance. Without standardizing,

indeed, the principal component method would favor the variables with larger variances at the

expenses of those with smaller values.

The panel in the middle of Tab les A1 and A2 shows the total variance accounted by each

factor. The Kaiser criterion suggests to retain those factors with variance equal or higher than

1. In both cases there is only one factor that satisﬁes this criterion and this factor explain th e

Intermediaries in International Trade

77 percent and the 86 percent of the sum of all observed variances. The lower panel of the two

tables reports the factor loadings which are the parameters of the linear function that relates the

observed variables and the factor. T he higher the load the more relevant the variable in deﬁning

the factor’s dimensionality. According to Table A1, the loadings on Factor1 are relatively large for

all the variables. The same holds when looking at Table A2. Finally, uniqueness is the variance

that is “uniqu e” to the variable an d not shared with others. Again all variables, in both tables,

have a low percentage of variance not accounted by other variables.

Table A1: PCA for Market Costs

Number of Obs. 180

Retained Factors 1

Number of Parameters 3

Variance Proportion

Factor1 2.30 0.77

Factor2 0.51 0.17

Factor3 0.18 0.06

Standardized Variables Factor1 Loadings Uniqueness

Number of documents for importing 0.81 0.34

Cost of importing 0.87 0.23

Time to import 0.93 0.12

Table A2: PCA for Governance Indicator

Number of Obs. 193

Retained Factors 1

Number of Parameters 6

Variance Proportion

Factor1 5.16 0.86

Factor2 0.4 0.07

Factor3 0.28 0.05

Factor4 0.09 0.01

Factor5 0.05 0.01

Factor6 0.03 0.01

Standardized Variables Factor1 Loadings Uniqueness

Voice & Accountability 0.86 0.25

Political Stability 0.85 0.27

Government Eﬀectiveness 0.96 0.09

Regulatory Quality 0.95 0.1

Rule of low 0.98 0.05

Control of Corruption 0.96 0.09

Intermediaries in International Trade

References

Ahn, J., Khandelwal, A. K. and Wei, S.-J. (2011). The role of intermediaries in facilitating trade,

Journal of International Economics 84(1): 73–85.

Akerman, A. (2010). A theory on the role of wholesalers in international trade based on economies

of s cope, Research papers in economics, Stockholm University, Department of Economics.

Antr`as, P. and Costinot, A. (2011). I ntermediated trade, The Quarterly Journal of Economics

126(3): 1319–1374.

Bernard, A. B., Eaton, J., Jen sen, J. B. and Kortum, S. S. (2003). Plants and productivity in

international trade, American Economic Review 93(4): 1268–1290.

Bernard, A. B. and Jensen, B. J. (1995). Exporters, jobs, an d wages in U.S. manufacturing:

1976-87, Brookings Papers on Economic Activity: Microeconomics 1995: 67–112.

Bernard, A. B. and J ensen, B. J. (1999). Exceptional exporter performance: cause, eﬀect, or both?,

Journal of International Economics 47(1): 1–25.

Bernard, A. B. and Jensen, J. B. (2004). Why some ﬁrms export, The Review of Economics and

Statistics 86(2): 561–569.

Bernard, A. B., Jensen, J. B., Redding, S. J. and Schott, P. K. (2007). Firms in international trade,

Journal of Economic Perspectives 21(3): 105–130.

Bernard, A. B., Jensen, J. B., Redding, S. J. and Schott, P. K. (2010a). Intraﬁrm trade an d product

contractibility, American Economic Review 100(2): 444–48.

Bernard, A. B., Jensen, J. B., Redding, S. J. an d Schott, P. K. (2010b). Wholesalers and retailers

in US trade, American Economic Review 100(2): 408–13.

Bernard, A. B., Redding, S. J. and Schott, P. K . (2010). Multiple-product ﬁr ms and product

switching, American Economic Review 100(1): 70–97.

Bernard, A. B., Redding, S. J. and Schott, P. K. (2011). Multi-product ﬁrms an d trade liberaliza-

tion, Quarterly Journal of Economics, forthcoming .

Blum, B., Claro, S. and Horstmann (2011). Intermediation and the nature of trade costs: theory

and evidence, Unpublished, Rotman School of Management.

Blum, B. S., Claro, S. and Horstmann, I. (2010). Facts and ﬁgures on intermediated trad e, American

Economic Review 100(2): 419–23.

Intermediaries in International Trade

Djankov, S., Freund, C. and Pham, C. S. (2011). Trading on time, Review of Economics and

Statistics 92(1): 166–173.

Eaton, J., Kortum, S. S. and Kramarz, F. (2011). An anatomy of international trade: Evidence

from French ﬁrms, Econometrica 79(5): 1453–1498.

Felbermayr, G. and Jung, B. (2011). Trade intermediaries and the organization of exporters, Review

of International Economics 19(4): 634–648.

Helpman, E., Melitz, M. J. and Yeaple, S. R. (2004). Export versus FDI with heterogeneous ﬁrms,

American Economic Review 94(1): 300–316.

Helpman, E., Melitz, M. and R ubinstein, Y. (2008). Estimating trade ﬂows: Trading partners and

trading volumes, The Quarterly Journal of Economics 123(2): 441–487.

Kaufman, D., Kr aay, A. and Mastruzzi, M. (2009). Governance matter VIII: Aggregate and indi-

vidual governance indicators 1996-2008, Policy Research Paper 4978, World Bank.

Lawless, M. (2010). Deconstructing gravity: trade costs and extensive and intensive margins,

Canadian Journal of Economics 43(4): 1149–1172.

Lawson, A. M., Bersani, K. S., Fahim-Nader, M. and Gu o, J. (2002). Benchmark input-output

accounts of the united states, Survey of Current Business 82: 19–108.

Mayer, T. and Zignago, S. (2005). Market access in global and regional trade, Working Papers

2005-02, CEPII research center.

Melitz, M. J. (2003). The impact of trade on intra-industr y reallocations and aggregate industry

productivity, Econometrica 71(6): 1695–1725.

Nunn, N. (2007). Relationship-speciﬁcity, incomplete contracts, and the pattern of trade, The

Quarterly Journal of Economics 122(2): 569–600.

Peng, M. W. and Ilinitch, A. Y. (2001). Export intermediary ﬁrms: A note on export development

research, Journal of International Business Studies 29(3): 609–620.

Petersen, M. A. (2009). Estimating standard errors in ﬁnance panel data sets: Comparing ap-

proaches, Review of Financial Studies 22(1): 435–480.

Petropoulou, D. (2007). Information costs, networks and intermediation in international tr ad e,

Economics series working papers, University of O x ford, Department of Economics.

Pierce, J. and Schott, P. (2009). A concordance between ten-digit U.S. Harmonized System Codes

and SIC/NAICS product classes and in dustries, Working Paper 155486, NBER.

Intermediaries in International Trade

Rauch, J. E. and Watson, J. (2004). Network intermediaries in international trade, Journal of

Economics & Management Strategy 13(1): 69–93.

Roberts, M. J. and Tybout, J. R. (1997). The decision to export in Colombia: An empirical model

of entry with sunk costs, American Economic Review 87(4): 545–64.

Silverman, B. W. (1986). Density Estimation for Statistics and Data Analysis, London: Chapman

& Hall/CRC.