Trade and the greenhouse gas emissions from international freight transport

Cristea, Hummels, Puzzello & Avetisyan JEEM 2013

Before this paper

• Mostly just case studies

Saunders, Barber & Taylor 2006 http://researcharchive.lincoln.ac.nz/dspace/bitstream/10182/125/1/aeru_rr_285.pdf

• prepared inventory of all energy and CO2 embodied in various agricultural products as produced in UK and New Zealand

• included emissions associated with transporting products to UK consumers– includes fertilizers, agri-chemicals, seed, grazing,

farm buildings, tractors, fences, irrigation, shipping and road transport

ResultsProduct NZ Kg

CO2/TonneUK Kg CO2/Tonne

lamb 688.0 2,849.1

onions 184.6 170.0

apples 185.0 271.8

dairy 1,422.5 2,920.7

Table created using data from Saunders, Barber and Taylor 2006

Question

• Trade and Environment literature looks almost exclusively at trade’s indirect effects– “how does trade liberalization affect a country’s

industrial emissions via scale, composition and technique effects?”

• Usually dismisses direct effects such as transport emissions

Question

• Why?– As a fraction of total global emissions, transport’s share is quiet

small– And international transport’s share of total transport emissions is

small as well• In 2004, international transport’s share of global CO2 emissions was only

3.5%

Question

• Begs the question as to whether indirect emissions associated with trade are any larger

• Moreover, aggregating the data may obscure important variation across sectors

This paper

• Calculates indirect emissions associated with trade– Emissions generated when producing goods that are traded

• Calculates direct emissions– Emissions generated when goods are transported across borders

• Compares direct to indirect emissions at sectoral level• Conducts thought experiments

– By how much would total emissions rise (or fall) if • all countries produced domestically all the goods they currently

consumed?• Doha round trade liberalizations implemented• Projected growth occurs

Data

• Production and Trade– Global Trade Analysis Project (GTAP) 7 • use 2004 as base year

– In principle, GTAP can accommodate • 57 “traded and non-traded sectors”• 113 countries

– Requires too much computational power– So aggregate the data into• 40 regions (of which 28 are individual countries)• 23 traded sectors, 6 non-traded service sectors

Calculating Weight• Three databases report trade by value and by weight

– US Imports and Exports of Merchandise– Eurostats Trade– Aladi trade database

• Data reported at 6 digit level of Harmonized system• Authors aggregate up to the 23 merchandise sectors used in this study• Calculate the weight of a particular sector’s products by “separately

sum[ming] the weight of trade and volume of trade and express[ing] them as a ratio” (p.158)– “this is equivalent to a share-weighted average of the weight/value ratio for

each HS6 product traded by that exporter” (p.158)– Then calculate weighted (by trade) average across all countries for which

requisite data exists

Modal Shares• Same three databases report modal shares by product category

– US Imports and Exports of Merchandise– Eurostats Trade– Aladi trade database

• Supplement with – Transborder Surface Freight Data

• But these databases only report bilateral trade data for a subset of countries

• For 35% of trade no modal information is available– Solution: estimate modal shares using aformentioned data and

independent variables such as geography and country and product characteristics

– Use fitted coefficients to predict model shares for remaining trade

• If measure trade by value– Road transport is very important (by value) for North America and Europe– For most regions (excluding North America and Europe) over 2/3 of shipments are by sea

• If measure trade by weight-distance– Sea-freight clearly dominates for all regions– Simple explanation: heavy stuff going long distances goes by sea

• And the high value shipments within North America and Europe that go by road only cover short distances, so won’t be very significant in a metric of trade by weight-distance

Emission Coefficients

• Output– Source: GTAP 7

• Computes emission coefficients by product and country– Multiplies use of coal, oil, gas, petroleum products, electricity, gas

distribution with CO2 coefficients for each– Also includes non-carbon CO2e based on IPCC methodology

» Particularly important for agricultural products

– GTAP assumes that emission coefficients • differ across countries

– especially according to mix and quantities of fuels used)

• Are static– no “technique” effects from trade or growth

• Transport– Ocean Freight

• Ship Emissions Study (National Technical University of Athens Laboratory for Maritime Transport)– Reports grams of CO2 per tonne-km for various vessel types

– Air Freight• Various sources• LOWest estimate is 552grams of CO2/tonne-km shipped (based on

Boeing 747)• HIGHest estimate is 950g/t-km based on fleet average as reported in

Aircraft Economics 1999

– Rail & Road• As reported by European Environmental Agency 2005

• State the obvious: – Ocean and Rail are considerably less polluting (per tonne-km) than Road or Air

• Air is about 80 times as polluting as Ocean

Use the data

ETodg ≡ GHG emissions

associated with transporting good g from origin o to destination d.

VALodg ≡ value of good g transported from origin o to destination d.

eTodg ≡ emission intensity

associated with moving good g from o to d.

WVog ≡ weight to value ratio for g when produced by o.

QSmodg ≡

quantity share of good g shipped via mode m

DISTmod ≡

distance from o to d via mode m

em ≡ GHG emissions associated with mode m (when providing one kg-km of transport services)

Assumed to be linear in distance (which fails in case of air transport)

Assumes homogeneity within a mode; fails if old, polluting vehicles used for some shipments and newer equipment for others; example: drayage fleet that crosses US-Mexico border

Another consideration

• Domestic production also requires transport• Authors treat this as part of output emissions

(on next slide)• “If production for external transport uses

domestic transport to the same extent as production for domestic use” then there is no problem

• Fails if production for export occurs near seaport/border/airport

Production emissions EY

Emission intensity of o’s g production (grams of CO2e per dollar of output)

o‘s output of g (measured in dollars)

Country o’s total output emissions

Share of good g in o’s production

Analysis

Dirtiest Industries

• Aggregate transport emissions for each industry by summing over all country pairs.

• Divide by value of trade• Yields “weighted average transport emission

intensity for that industry”

“fraction a given sector contributes to the world-wide CO2 emissions from international transport” (p.163)

“Fraction of a sector’s transport emission intensity in that sector’s total trade-related CO2 emission intensity” i.e. = eT

g/(eTg+eY

g)

Sectors for which transport emissions are more than half of total trade-related emissions

Only refers to traded output

North America is ranked fourth in terms of output emissions but first in terms of export-transport emissions

Reason: US exports are disproportionately by air

Problem: this graph obscures intensity because doesn’t control for size (US is very large)

=grams of CO2e per dollar of trade

Notable:eT

o > eYo for all these countries

Canadian production is fairly CO2e intensiveSome countries with low eY

o have quite high eTo

-i.e USAOther comparisons: ‘India’s production has 143% more emissions (per dollar of trade) than US, but after incorporating transportation, its exports are less emission intensive in total’ (p.165)

Thought Experiment #1

• What if there wasn’t any trade?• Assume everyone continues to consume what they do

now• But have to produce it themselves using current

technology– Ignore input shortages and/or possible scale economies

• Would total emissions rise or fall?– On the one hand, there won’t be any transport emissions– On the other hand, some goods currently produced by low

eYog countries will be produced by high eY

og countries

• Horizontal Axis: change in embodied emission intensity of a country’s consumption• A negative value means that trade reduces this country’s “embodied” emissions

– 26.5% of trade reduces emissions (31% of world trade by value)• For the mean and median country in the sample, a move to trade raises (embodied)

emission intensity of current consumption

Special Cases

Not much dispersion

By value, “80 percent of trade in wearing apparel raises emissions’ (p.165)

Highly variable• 41.6% of trade reduces emissions (34.4% by value)• Variability reflects both differences in production

techniques: hot houses versus field grown • and freight modes: air for cut flowers versus ocean

for commodity grains

Thought experiment #2

• Suppose Doha round is successful• Use CGE simulation to predict how emissions

will change– By total output– By exports– By imports– By mode

• “LOW” assumes aviation emission intensity is that associated with most efficient long range planes

• “HIGH” corresponds to emission intensity of current US air cargo fleet)

• Doha scenarios 4 & 5: – Increased access to

agricultural markets– Tariff cuts of

• 40% to 60% (developed countries)

• 20% (developing countries)

• See – global output emissions rise– Global transport emissions

fall• except under “LOW”

scenario

• Doha scenario 9: – Increased access to

agricultural markets– Plus progressive

cuts in non-agricultural tariffs• “peak tariffs are cut

more than lower tariffs” (p.167)

• See – global output and

transport emissions rise

• All tariffs eliminated: – global output

falls– Exports rise– Output and

transport emissions rise

– Rail and road transport falls

Why? • Current system of preferential tariffs (via FTAs and Customs

Unions) favours trade between proximate regions.• Much of trade between proximate regions is by land• The tariffs that fall the most will be between distant nations,

which will need to use sea or air

• Growth occurs– Expected at 3.57%/annum

• More rapid in China and India than in N.Am & Europe

– Predict: • Exports to rise faster than

output• Sea and Air transport to

increase more than rail and road

Why? • The countries that are expected to grow the most are India and China • These countries are not adjacent to European and North American markets• Will need to use ocean and air freight

Final thought experiment

• Suppose all countries adopted $50/ton carbon tax

• Levied it on transport emissions• What would be the tariff equivalent?• (Expressed as a percent of the value of traded

goods)

• Highest (transport) tariff would only be 4%

• For reference: Average current rate ~ 3%

• (Current average tariff on minerals = 1.56%)

Conclusions & Other Take Away Points

• “Two-thirds of trade-related emissions in U.S. exports are due to international transportation” (p.170)

• “Worldwide over 75 percent of the trade-related emissions of transport equipment, electronic equipment, machinery, and manufactures NEC come from transportation” (P.170)

• Trade & Environment economists should not be ignoring transport emissions!

Unanswered questions• How do they calculate DISTm

od? E.g. use Chicago or LA or New Jersey?• For countries that are bundled into one of the aggregated regions, are

they including exports from one country within that region to another country in that region?

• What fraction of global emissions comes from output of traded goods? is it similarly tiny (i.e. on the order of ½%?– If so, then why T&E economists worrying so much about it?– Is it because those trade-exposed industries seem to care, and if you consider

the tariffs that are implicit in a 50$ carbon tax, you’d see that these tariffs are non-negligible.• But, really, they are still small. The largest implicit tariff that Cristea et al find is still

only 4%, and typical tariffs are around 1%. That seems inconsequential.

• Any sense of how direct and indirect emissions compare for other pollutants, e.g. Sulphur emissions?

Caveats

• Local distribution– French wine consumed in NYC has fewer transport

emissions than Californian wine consumed in NYC