Trade data: Where to find it How to use it?siteresources.worldbank.org/INTRANETTRADE/... · Trade...
Transcript of Trade data: Where to find it How to use it?siteresources.worldbank.org/INTRANETTRADE/... · Trade...
Trade data:Where to find itHow to use it?
Olivier CadotPRMTR & University of Lausanne
Constructing the basic snapshot of a country’s trade
What should be in there:
• Sectoral composition • Geographical composition • Overall openness• Basic trends (growth in value and market share)
http://siteresources.worldbank.org/INTRANETTRADE/Resources/Pubs/Guide_To_Trade_Data_Analysis.pdf
Where to find it: Trade Flow Databases
Provider Name Content Location Access policy Comments
UNSD COMTRADEBilateral trade flows, by product, 1988 to current
year - 1http://wits.worldbank.org Free HS, SITC
CEPII BACIBilateral trade flows, by product, 1995 to current
year - 2http://www.cepii.fr Free
Reconciles exporter and
importer data for values and
unit values
CHELEM-ITBilateral trade flows between predefined country
and product groupings, 1967-current year - 2http://www.cepii.fr Free
Ad-hoc classification plus SIC
and GTAP
ITC Trade Map COMTRADE plus gadgets www.trademap.org
USITC Trade dataweb U.S. bilateral exports and imports, 1989 to current yearhttp://dataweb.usitc.gov FreeHS, SITC, NAICS (successor to
SIC)
DOTS Bilateral trade flows, aggregate, ddp
BOP
World
Bank
Trade,
production and
protection
Imports, exports, tariffs, production, wages etc. by
SIC 3 digits sector, 1976-2004google it Free Not updtated after 2004
IMF
The Harmonized system (HS)
• 4 harmonized levels: sections (HS1, 21), chapters (HS2, 99), Headings (HS4, 1’024), and subheadings (HS6, 5’000)
• More disaggregated levels (HS8, HS10) not harmonized and often not available to the public
• Frequent revisions• HS categories have unequal importance
in trade
How goods are classified
Other classifications
SITC: The main competitor to HS
• 5 levels: sections (1 digit, 10), divisions (2 digits, 67), groups (3 digits, 262); subgroups (4 digits, 1’023); basic headings (5 digits, 2’970 lines)
• Fewer revisions than HS• Concordance with HS relatively easy
Other useful classifications
• BEC (Broad Economic Categories):• Rauch: Product differentiation index• Input-output Categories
BEC and WTO classifications
• Both map HS codes into 3 categories: capital, intermediate or consumer goods for BEC, raw, semi-finished or finished for WTO
• Always check that the classifications make sense
Rauch’s classification
• Maps SITC4 into 3 categories: homogenous, référence-priced, or differentiated (all other)
• Two classifications: « conservative » or « liberal »--little difference between the two (conservative has 347 differentiated goods out of 642 vs. 361 in the liberal)
Number of records
Number of reporters
0
20000
40000
60000
80000
100000
120000
140000
160
165
170
175
180
185
190
195
200
2000 2001 2002 2003 2004 2005 2006 2007 2008
EBOPS Services Countries Records
1 Transportation (Passenger – Freight – Other) 191 32,497
2 Travel (Business – Personal – Other) 192 28,806
3 Communications services (Postal - telecom) 174 15,212
4 Construction services (Abroad – in the economy) 136 10,735
5 Insurance services (Life – freight – reinsurance) 176 12,930
6 Financial services 148 14,163
7 Computer and information services 145 12,407
8 Royalties and license fees 155 10,888
9 Other business services (Merchanting – Professional…_) 175 22,293
10 Personal, cultural, and recreational services 132 10,927
11 Government services 181 14,214
Mode 4
Compensation of employees 47 1,505
Workers' remittances 44 2,690
Migrants' transfers 24 314
Direct investment 43 1,391
Coverage of the UN Database
A portfolio approach: Using raw data to build a basic « trade performance diagnosis »
The golden rule: mirroring, but…
• There are large discrepancies between trade flows recorded at origin and destination
• Exports are typically badly monitored by customs (no revenue)• However under-billing/smuggling seem to be substantial• And there is a trade-off when looking at trade flows from industrial countries to
developing ones (export data from more reliable industrial-country customs vs. Import data from less reliable developing-country customs)
• So… call of judgement, but in general mirroring produces more stable/plausible results
Where to find it: A few sources of ready-made indicators
Provider Name Content Location Access policy Comments
growth (real & nominal) Free
composition (goods vs. services) Free
Openness (X+M)/GDP Free
Balance Free
Share of world trade (level and growth) Free
Diversification Free
World BankWITS Extensive/intensive margins Free
Trade complementarity indices Free
Trade intensity indices Free
Revealed comparative advantage Free
Free
World Bank Import demand elasticities at HS6/country level DECRG Free For one year only (2004)
CEPII Distances Distances between countries (for gravity)
World
BankWTI
May be discontinued/merged
into World Bank's global data
platform
Export portfolio quality: growth orientation
23
45
67
Log
of W
orld
Gro
wth
of P
rod
ucts
(2
00
0-0
8)
-2 -1 0 1 2
Log of Share of Total Exports (%)
Pakistan's Growth Orientation of Products in 2008
44
.55
5.5
66
.5
Log
of Im
port
Gro
wth
of C
oun
trie
s (
20
00
-08
)
-2 -1 0 1 2 3
Log of Destination's Share in Exports (%)
Growth Orientation of Pakistan's Destinations, 2008
• Take every export product, calculate its share in country’s exports for a given (recent) year, call it x, and put it in logs
• Take the growth rate of world imports of that product over a 10-year window around the year of interest (or before that year if it is the last available), call it y
• Plot y against log x and draw the regression line; good if slopes upward (large export items are also the fastest growing in world trade), bad if it slopes downward
• Do the same thing with export destinations taking the growth of their aggregate imports
Putting concepts to work: Basic decomposition of export growth
Intensive margin: higher volumes of existing products & destinations
Export growthExtensive margin
New products
New destinations
Sustainability margin: Survival of new products/destinations
Largest contributors to export growth (across countries and time)• Intensive margin• New-destination margin
General pattern: diversification followed by reconcentration (at high levels of income)—this must be used to benchmark national export diversification
34
56
7
Theil
index
0
1000
2000
3000
4000
5000
num
ber
of export
ed p
roduct
s
0 20000 40000 60000GDP per capita PPP (constant 2005 international $)
Active lines - quadratic Active lines - non parametric
Theil index - non parametric Theil index - quadratic
# active export lines
Theil index
Along the path of income growth, countries
• First diversify (up to $25’000 PPP per capita GDP)• Then re-concentrate
IRL
ESP
GRC
GBR23
45
15000 20000 25000 30000 35000 40000GDP per capita, PPP (constant 2005 international $)
and this is true even for individual country trajectories :
Intensive and extensive margins
Hummels-Klenow’s original formulation (product-wise)
Let Ki be the set of products exported by country i, 𝑋𝑘𝑖 the dollar value of i’s exports of product k to the
world, and 𝑋𝑘𝑊 the dollar value of world exports of product k.
The (static) intensive margin is defined by HK as
𝐼𝑀𝑖 = 𝑋𝑘
𝑖𝐾𝑖
𝑋𝑘𝑊
𝐾𝑖
In words, the numerator is i’s exports and the denominator is world exports of products that are in i’s
export portfolio.
The extensive margin (also static) is
𝑋𝑀𝑖 = 𝑋𝑘
𝑊𝐾𝑖
𝑋𝑘𝑊
𝐾𝑊
Interpreting the intensive and extensive margins
Pakistan
India
Vietnam
Indonesia
Pakistan
India
Vietnam
Indonesia
.2.4
.6.8
11
.2
Inte
nsiv
e M
arg
in
85 90 95 100
Extensive Margin
1998 2008
Intensive and Extensive Margin in Products, 1998-08
Your market share in your export portfolio
Weight of your export portfolio in world trade
Big fish in a small pond
Small fish in a big pond
Interpreting the intensive and extensive margins: caveats
Costa Rica
Pakistan
Costa Rica
Pakistan
.1.1
2.1
4.1
6.1
8.2
Inte
nsiv
e M
arg
in
84 86 88 90 92 94
Extensive Margin
1993 2003
Intensive and Extensive Margin in Products, 1993-03
Your market share in your export portfolio
Weight of your export portfolio in world trade
Big fish in a small pond
Small fish in a big pond
Extension to geographical markets
Let Di be the set of destination markets where i exports (anything from one to 5’000 products—it
doesn’t matter), XiD the dollar value of i’s total exports to destination d, and XWd the dollar value of world
exports to destination d (i.e. d’s total imports). All these dollar values are aggregated over all goods.
Intensive margin
𝐼𝑀𝑖 = 𝑋𝑖𝑑𝐷𝑖
𝑋𝑊𝑑𝐷𝑖
Extensive margin
𝑋𝑀𝑖 = 𝑋𝑊𝑑𝐷𝑖
𝑋𝑊𝑑𝐷𝑊
Interpreting the intensive and extensive margins: destination markets
Pakistan
India
Vietnam
Indonesia
Pakistan
India
Vietnam
Indonesia
.2.4
.6.8
11
.2
Inte
nsiv
e M
arg
in
97.5 98 98.5 99 99.5 100
Extensive Margin
1998 2008
Intensive and Extensive Margin in Markets, 1998-08
Your market share in your destinationportfolio
Weight of your destination portfolio in world trade
Big fish in a small pond
Small fish in a
big pond
-2-1
01
2
lp1
5 10 15 20ldiffimport
Agri-food product: A product-risk index and a portfolio risk-return snapshot-2
-10
12
lp1
5 10 15 20ldiffimport
Indonesia India
Step 1: construct a product-level risk index using the EU’s alert database
Log SPS risk coefficient
Log Demand growth
Better portfolio Better
portfolio
Log SPS risk coefficient
Log Demand growth
Hidden protectionism
(also control for initial export volume)
Count of alerts 2000-2008
Step 2: construct a risk-return profile for national agri-food export portfolios
Key ingredients
• Agglomeration effects/spillovers/scale effects• Latent comparative advantage in the targeted sector
Good 1
Good 2
E1
PPF
World price line
Good 1
Good 2
E1
E2PPF
World price line
Zone of increasingreturns/agglomeration in the production of good 2
Targeted by IP
When does it make sense to « do industrial policy »? (Harrison Rodriguez-Clare 2009)
Identifying latent comparative advantage: matching national factor endowments withthe factor intensities of traded goods
Endowments and factor intensities
Let ki = Ki/Li be country i’s stock of capital per worker, and let Hi be a proxy for its stock of human capital,
namely the average level of education of its workforce in years. These are national factor endowments.
Let Ji be the set of goods that country i exports. Let
𝜔𝑗𝑖 =
𝑋𝑗𝑖
𝑋𝑗𝑖
𝐽 𝑖
be Balassa’s Revealed Comparative Advantage (RCA) for country i and product j. Let also Ij be the set of
countries that export good j. Good j’s revealed intensity in capital is
𝜅𝑗 = 𝜔𝑗𝑖𝐾𝑖
𝐼𝑗
Similarly,
ℎ𝑗 = 𝜔𝑗𝑖𝐻𝑖
𝐼𝑗
is product j’s revealed intensity in human capital.
A glance at the endowment data…
… and the « revealed factor intensities »
Industrial countries (past $25’000 at PPP) reconcentrate their exports…
…by closing export lines that are out of tune with their endowments.
24
68
10
12
0 50000 100000 150000 200000
Product intensities Country endowments
Capital
Human capital
Intensive margin
New-productsmargin
Deathmargin
• Large volume of churning: This year’s new-product margin is next year’s death margin• Low survival of export spells at product (HS6) level: median 2 years• This is why the extensive margin contributes little to growth (low survival)
Basic decomposition of export change
Baseline exports
𝑋0 = 𝑋𝑘0𝐾0
Terminal exports
𝑋1 = 𝑋𝑘1𝐾1
Decomposition of the change in total volume
∆𝑋 = ∆𝑋
𝐾0∩𝐾1
+ 𝑋𝑘𝐾1\𝐾0
− 𝑋𝑘𝐾0\𝐾1
Costa Rica: intensive, new-products, and death margins
New products 2003-5Deaths 2003-5
Baseline export portfolio: 1991-3 Export portfolio 2003-5
02
46
810
12
Re
vea
led
Hu
man
Cap
ital I
nte
nsity
Inde
x
0 50000 100000 150000 200000Revealed Physical Capital Intensity Index
02
46
810
12
Re
vea
led
Hu
man
Cap
ital I
nte
nsity
Inde
x
0 50000 100000 150000 200000Revealed Physical Capital Intensity Index
02
46
810
12
Re
vea
led
Hu
man
Cap
ital I
nte
nsity
Inde
x
0 50000 100000 150000 200000Revealed Physical Capital Intensity Index
02
46
810
12
Re
vea
led
Hu
man
Cap
ital I
nte
nsity
Inde
x
0 50000 100000 150000 200000Revealed Physical Capital Intensity Index