Exporter Dynamics DatabaseVersion 2.0

Ana Margarida FernandesMartha Denisse PierolaAldo Pazzini Bortoluzzi

Trade and International Integration UnitDevelopment Research Group

The World Bank

Washington DC October 20 2015

A project funded by KCP, MDTF, and SRP

What we knew about bilateral trade flows and what we know now

$35.2 millionTotal tradein 2013

78 exporters$451,836 average size

Herfindahl Index: 45%

Exporter entry rate: 36%

$35.8 million Total tradeIn 2013

32 exporters$1,119,836 average size

Herfindahl Index: 92%

Exporter entry rate: 41%

… and how about total exports of a given product?

Nicaragua:$ 105 million in

Thailand:$ 99 million in

41 exporters$2,581,599 average size

Top 5% exporters: 62%

Exporter Entry: 44%1st-y Entrant Survival: 39%

643 exporters$153,474 average size

Top 5% exporters: 84%

Exporter Entry: 48%1st-y Entrant Survival: 34%

• Firm-level data is key to understand the micro foundationsof export growth

– It provides information about the characteristics of the exporters

– The degree of concentration among these exporters

– And their entry, exit and survival dynamics

• The Exporter Dynamics Database (EDD) version 1.0 filledthis analytical gap by providing a comprehensive source ofinformation on firms that export and their dynamics

Exporter Dynamics Database

What measures are included in the EDD?

• Basic characteristics of export sector– Number of exporters, exporter size (in export value), exporter

growth

• Export concentration/diversification– Herfindahl indexes, share of top X% exporters, number of

products or destinations per exporter, number of exporters perproduct or destination

• Firm dynamics in the export sector– Exporter entry, exit and first-, second-, or third-year entrant

survival rates

• Firm-product dynamics– Product entry, exit and survival rates for incumbent exporters and

share of new products in exporter values

• Firm-destination dynamics– Destination entry, exit and survival rates for incumbent exporters

and share of new destinations in exporter values

• Unit prices

What is basis for constructing the measuresin the EDD?

• Customs data collected for each country covers the universeof export transactions

– Data includes in each country values and quantities exported by each firm of each product (HS 6-digit) to each destination country in each year

– We exclude only oil sector exports (HS chapter 27)

• For products we use a consolidated set of HS 6-digit codesthat are consistent across HS1996, HS2002, HS2007, andHS2012 revisions

Disaggregation levels of the EDD measures

• Exporting country-year level

• Exporting country-product-year level with product beingeither HS 2-digit or HS 4-digit or HS 6-digit

• Exporting country-destination-year level

• Exporting country-HS 2-digit product-destination-yearlevelNEW

• Exporting country-ISIC 3-digit-year levelNEW

Flavors of the EDD measures

• Measures at 7 different disaggregation levels areavailable in 4 different flavors:

– Covering all export transactions

– Covering export transactions for firms whose annualexports are all larger than 1,000 USDNEW

– Covering export transactions for manufacturing onlyNEW

– Covering export transactions for manufacturing onlyand for firms whose annual exports are all largerthan 1,000 USDNEW

EDD version 1.0 covered 45 countries

Data for 34 existing countries was updated

Data for 25 new countries was added

EDD version 2.0 covers 70 countries

Country 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 TotalAlbania 1 1 1 1 1 1 1 1 1 9

Belgium 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17

Burkina Faso 1 1 1 1 1 1 1 1 8

Bangladesh 1 1 1 1 1 1 1 1 1 1 10

Bulgaria 1 1 1 1 1 1 6

Brazil 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 18

Botswana 1 1 1 1 1 1 1 1 1 1 1 11

Chile 1 1 1 1 1 1 1 1 1 1 10

Cameroon 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17

Colombia 1 1 1 1 1 1 1 7

Costa Rica 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 15

Dominican Republic 1 1 1 1 1 1 1 1 1 1 1 1 1 13

Ecuador 1 1 1 1 1 1 1 1 1 1 1 1 1 13

Egypt 1 1 1 1 1 1 1 7

Spain 1 1 1 1 1 1 1 1 1 1 10

Estonia 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 15

Guatemala 1 1 1 1 1 1 1 1 1 9

Iran 1 1 1 1 1 5

Jordan 1 1 1 1 1 1 1 1 1 1 10

Kenya 1 1 1 1 1 1 1 1 1 9

Cambodia 1 1 1 1 1 1 1 1 1 1 10

Kuwait 1 1 2

Laos 1 1 1 1 4

Lebanon 1 1 1 1 1 5

Morocco 1 1 1 1 1 1 1 1 1 1 1 1 12

Mexico 1 1 1 1 1 1 1 1 1 1 1 1 12

Macedonia 1 1 1 1 1 1 1 1 1 1 10

Mali 1 1 1 1 4

Mauritius 1 1 1 1 1 1 1 1 1 1 1 11

Malawi 1 1 1 1 1 1 1 7

Niger 1 1 1 3

Nicaragua 1 1 1 1 1 1 1 1 1 1 1 1 12

Norway 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 18

New Zealand 1 1 1 1 1 1 1 1 1 1 1 1 12

Pakistan 1 1 1 1 1 1 1 1 1 9

Peru 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 17

Portugal 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 16

Senegal 1 1 1 1 1 1 1 1 1 1 1 1 1 13

El Salvador 1 1 1 1 1 1 1 1 8

Sweden 1 1 1 1 1 1 1 1 1 1 10

Turkey 1 1 1 1 1 1 1 1 1 1 1 1 12

Tanzania 1 1 1 1 1 1 1 1 1 1 10

Uganda 1 1 1 1 1 1 1 1 1 1 10

Yemen 1 1 1 1 1 5

South Africa 1 1 1 1 1 1 1 1 1 1 1 1 12

Bolivia 1 1 1 1 1 1 1 7

Paraguay 1 1 1 1 1 1 6

Uruguay 1 1 1 1 1 1 1 1 1 1 1 1 12

Germany 1 1 1 1 4

Denmark 1 1 1 1 1 1 1 1 1 1 1 1 12

Croatia 1 1 1 1 1 1 6

Slovenia 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 15

Romania 1 1 1 1 1 1 1 7

Madagascar 1 1 1 1 1 1 6

Zambia 1 1 1 1 1 1 1 1 1 1 1 1 1 13

Rwanda 1 1 1 1 1 1 1 1 1 1 1 1 12

Gabon 1 1 1 1 1 1 1 7

Cote d'Ivoire 1 1 1 1 4

Guinea 1 1 1 1 4

Georgia 1 1 1 1 1 1 1 1 1 1 10

Kyrgyz Republic 1 1 1 1 1 1 1 7

Nepal 1 1 1 1 4

Ethiopia 1 1 1 1 1 5

Thailand 1 1 1 3

Myanmar 1 1 1 3

Timor-Leste 1 1 1 1 1 1 1 7

Swaziland 1 1

Sao Tome 1 1

Sri Lanka 1 1

Total 9 10 12 16 22 31 36 37 43 50 53 57 59 56 51 48 20 10 620

EDD version 2.0 covers mostly the 2000s and 2010s

Public and Free Access to the Exporter Dynamics Database:

http://econ.worldbank.org/exporter-dynamics-database

Facts and Research Findingsbased on the EDD

Across countries exports are highly concentrated among their largest firms

• The role of these export superstars for countries’ comparative advantage is studied in Freund and Pierola (2015, REStat)

Source: based on Exporter Dynamics Database version 2.0.

Entrant survival is higher in the presence of lower trade costs and larger networks

Survival rates of new exporters are higher in lower trade costs economies

Note: Average for 2006-2008 period

ALB

BEL

BFA

BGD

BOL

BRA

BWA

CHL

COL

CRI

DNKDOM

ECU

ESP

GAB

GEO GTMHRV

IRN

JOR

KEN

KGZ

KHM

LAO

MAR

MDGMKD

MLI

MUS

MWI

NIC

NOR NZL

PAKPERPRT

PRY

RWA

SEN

SLV

SVN

TUR

TZA

UGA

URY

ZAF

0

100

02

00

03

00

04

00

0

Cost to

exp

ort

(U

S$ p

er

con

tain

er)

.2 .3 .4 .5 .6 .7First-year Entrant Survival Rate

Source: based on Exporter Dynamics Database version 2.0.

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

Lagged Ln (Number of Exporters) 0.026*** 0.029*** 0.026*** 0.030***

(0.000) (0.000) (0.000) (0.000)

Year Fixed Effects Yes Yes Yes Yes

HS 2-Digit Fixed Effects Yes

Destination Fixed Effects Yes

HS 2-Digit-Destination Fixed Effects Yes

Observations 719,409 719,409 719,409 719,409

R-squared 0.015 0.047 0.027 0.098

First-Year Entrant Survival Rate

Regressions based on measures

at exporting country-HS 2-digit-

destination-year level Presence of other exporters of same product to same market helps survival of new exporters

Sources of the great trade collapse

• The trade collapse resulting from the global financial crisiswas entirely driven by the intensive margin (mean exportersize)• After accounting for countries’ different degrees of distress from

the crisis and for sectoral demand and supply shocks

Source: based on Exporter Dynamics Database version 2.0.

Growth in

Total Exports

Growth in

Number of

Exporters

Growth in

Mean Exports

per Exporter

(1) (2) (3)

Dummy for Global Financial Crisis Year -0.254*** -0.024 -0.230***

(0.030) (0.014) (0.023)

Exporting Country Fixed Effects Yes Yes Yes

HS 2-Digit Fixed Effects Yes Yes Yes

Observations 31,242 31,242 31,242

R-squared 0.018 0.026 0.015

Regressions based on measures

at exporting country-HS 2-digit-year level

Number of Exporters, 9%

Average size, 91%

Export growth decomposition during the collapse

How do export-sector characteristics vary with country size and stage of development?

Positive correlations between number,

sizeconcentration of

exporters and

either GDP per capita or GDP

Source: based on Exporter Dynamics Database version 2.0. following Fernandes, Freund and Pierola (2015, JDE)

Establishing the robustness of those correlations

• Larger economies and more developed economies have:more exporters, larger exporters, and more concentration in thetop 5% of firms• Even after controlling for sectoral composition of exports, differences

across destination markets, and cyclical effects

Ln Number

of Exporters

Ln Mean

Exports per

Exporter

Share of Top

5%

Exporters

Ln Number

of Exporters

Ln Mean

Exports per

Exporter

Share of Top

5%

Exporters

Ln Number

of Exporters

Ln Mean

Exports per

Exporter

Share of Top

5%

Exporters

(1) (2) (3) (4) (5) (6) (7) (8) (9)

Ln GDPpc 0.479*** 0.463*** 0.045*** 0.441*** 0.173** 0.060*** 0.271** 0.233*** 0.028***

(0.098) (0.093) (0.012) (0.092) (0.067) (0.009) (0.102) (0.064) (0.009)

Ln GDP 0.660*** 0.408*** 0.040*** 0.783*** 0.280*** 0.038*** 0.416*** 0.341*** 0.034***

(0.067) (0.065) (0.007) (0.071) (0.058) (0.006) (0.072) (0.049) (0.006)

HS 2-Digit Fixed

EffectsYes Yes Yes

Destination Fixed

EffectsYes Yes Yes

Bilateral Gravity

CovariatesYes Yes Yes Yes Yes Yes

HS 2-Digit-

Destination Fixed

Effects

Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes

Observations 33,488 33,459 22,670 46,162 45,716 24,751 599,299 599,299 131,613

R-squared 0.716 0.441 0.316 0.736 0.318 0.307 0.466 0.394 0.419

Regressions for measures

at exporting country-HS 2-digit-year

level

Regressions for measures

at exporting country-destination-year

level

Regressions for measures

at exporting country-HS 2-digit-

destination-year level

Source: based on Exporter Dynamics Database version 2.0. following Fernandes, Freund and Pierola (2015, JDE)

What is the magnitude of the effects?

• Relative to a country at the 25th percentile of the GDPper capita distribution (Cameroon), a country at the 75th

percentile of the GDP per capita distribution (Mexico)has:

• 87 percent larger number of exporters• 84 percent larger mean exports per exporter• 8.2 percentage points larger share of exports

accounted for by the top 5% of exporters

How do exporter dynamics vary with country size and stage of development?

Negative correlations between

exporter turnover and positive

correlations between entrant survival

and either GDP per capita

or GDP

Source: based on Exporter Dynamics Database version 2.0. following Fernandes, Freund and Pierola (2015, JDE)

Establishing the robustness of these correlations

• In more developed economies, exporter entry and exit ratesare lower while first-year survival of entrants is higher• Even after controlling for sectoral composition of exports,

differences across destination markets, and cyclical effects

Entry

Rate

Exit

Rate

Entrant

Survival

Rate

Net Entry

Rate

Turnover

Rate

Entry

Rate

Exit

Rate

Entrant

Survival

Rate

Net Entry

Rate

Turnover

Rate

Entry

Rate

Exit

Rate

Entrant

Survival

Rate

Net Entry

Rate

Turnover

Rate

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15)

Ln GDPpc -0.063*** -0.065*** 0.031*** 0.003 -0.120*** -0.043*** -0.042*** 0.018* -0.001 -0.071*** -0.050*** -0.054*** 0.030** 0.005* -0.074***

(0.011) (0.011) (0.009) (0.002) (0.022) (0.010) (0.010) (0.009) (0.003) (0.019) (0.009) (0.011) (0.013) (0.003) (0.019)

Ln GDP -0.005 -0.003 0.002 -0.002 -0.001 -0.034*** -0.033*** 0.025*** 0.001 -0.035*** -0.017** -0.015** 0.013** -0.001 0.043***

(0.007) (0.007) (0.005) (0.002) (0.016) (0.007) (0.007) (0.005) (0.002) (0.012) (0.006) (0.007) (0.005) (0.002) (0.013)

HS 2-Digit Fixed

EffectsYes Yes Yes Yes Yes

Destination Fixed

EffectsYes Yes Yes Yes Yes

Bilateral Gravity

CovariatesYes Yes Yes Yes Yes Yes Yes Yes Yes Yes

HS 2-Digit-

Destination Fixed

Effects

Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Observations 31,465 31,412 26,212 30,805 30,805 48,658 48,141 39,777 44,279 44,279 883,355 868,703 643,804 695,860 695,860

R-squared 0.234 0.242 0.131 0.008 0.249 0.202 0.194 0.105 0.011 0.143 0.147 0.149 0.096 0.018 0.181

Regressions for measures

at exporting country-HS 2-digit-year level

Regressions for measures

at exporting country-destination-year level

Regressions for measures

at exporting country-HS 2-digit-destination-year level

Source: based on Exporter Dynamics Database version 2.0. following Fernandes, Freund and Pierola (2015, JDE)

What is the magnitude of the effects?

• Relative to a country at the 25th percentile of the GDPper capita distribution (Cameroon), a country at the 75th

percentile of the GDP per capita distribution (Mexico)has:

• 11 percentage points lower exporter entry and exitrates

• 6 percentage points higher first-year survival rate ofentrants into export markets

Lessons from these facts

• Exporter dynamics change as countries get richer• In developing countries, there is high turnover with many

firms entering export markets and exiting almost immediately• As countries develop, entrant survival is higher as fewer but

more resilient exporters enter the market

• Larger exporters and more concentration at the top of theexporter-size distribution in richer countries is consistentwith allocative efficiency in export markets improving ascountries develop• In developing countries, distortions to resource allocation

inhibit the emergence of very large firms resulting in a“truncated top” (rather than a “missing middle”)

What other uses for the EDD?

• The EDD has supported external research– 1,112 downloads of the EDD in FY2014– EDD used for research on the impact of FDI promotion policies, banking crises,

etc.

• The EDD has been used in WB analytical work and policy advice ontrade and competitiveness– Reports on Bangladesh, Brazil, Dominican Republic, Ethiopia, Georgia, Malawi,

Peru, South Africa, Uruguay– Regional reports in SSA, MENA, LAC

• The EDD can help answer understand the relationship betweenexporter dynamics and policy factors such as:– Trade costs (logistics costs, customs delays)– Trade barriers (tariffs, non-tariff measures, contingent protection measures)– Business environment or economic shocks (exchange rates)

• The EDD can be used as a tool for the impact evaluation of tradefacilitation reforms and export promotion programs => Nepal

Thank You!