Big Picture There are three big questions related to trade and environment 1) If we make trade freer, is our own environment going to deteriorate? 2) If.

Big PictureThere are three big questions related to trade and environment• 1) If we make trade freer, is our own environment going to deteriorate?• 2) If an open economy tightens its domestic regulations, are

(a) local firms going to lose their competitiveness?(b) local consumers going to be able to simply circumvent the regulations by

importing polluting goods from abroad instead?

Antweiler, Copeland and Taylor (2001) , Copeland and Taylor (2003) and Frankel and Rose (2005) attempt to answer question 1.

Ederington, Levinson and Minier (2005) are an example of an answer to question 2a.

Next: Levinson (2010) tries to answer 2b by assessing whether the pollution content of US trade has indeed been growing over time

Offshoring Pollution: Is theUnited States Increasingly Importing Polluting

Goods?

Arik Levinson Review of Environmental Economics

and Policy 4(1): 63–83 (2010)

Specific Question Posed

• “has the composition of U.S. imports shifted toward more polluting goods?” (p.64)

• i.e. “are [they] importing proportionately more or fewer polluting goods, regardless of the cause”? (p.64)

• Pp 65-67 gives highlights from US politics regarding trade and environment

• Mentions infamous Lawrence Summers memo– Suggests that rich countries should have their

polluting done by poor countries• Begs question as to whether rich countries are

indeed having someone else do their polluting for them

Two different ways to measure “pollution”

• Use pollution abatement and control costs as a proxy for pollution– ceteris paribus, we expect the most polluting

industries to have the highest control costs– US’ PACE: Pollution Abatement and Control

Expenditures (US Bureau of the Census)• Measure/impute emission intensities (per

dollar of output)– create a pollution “inventory”

Levinson’s (2010) Approach

• No regressions• A simple accounting exercise

• The US Bureau of Economic Analysis’ (BEA) input-output tables tell us aij = how much of good j is required to produce one unit of good i using US production techniques

• IPPS coefficients to find out how much pollution is generated directly when unit of output X is produced

• The direct+indirect emission coefficient (DIECi -- my terminology) for good i is

ei+Σj aijej[Yj-Xj]/[Yj-Xj+mj]

where Yj is domestic production of that input, Xj measures exports, and mj is imports of input j.

• Then– Multiply imports of good i by DIECi

– sum over all the goods i=1,..., M– Do this for every year to get a time series for pollution avoided via

imports• Levinson calls this the “pollution content of imports”

Several useful comparisons

Between 1972 and 2004, US imports grew by 574%. If imports grew proportionately, then we’d expect a 574% increase in pollution over same period: red

Green plots “pollution actually avoided via imports” over same period

Compare actual “pollution avoided via imports”---green--- with what we would think was avoided if only looked at the direct emission coefficients --- yellow:

Why is the gap between green and yellow growing?

• US is increasingly importing more finished products and few intermediate inputs– the adjustment matters more for imports of

finished products (because that’s when we are most likely to ignore intermediate pollution)

Tables

• Column 2 tells us how much lower the green line is than the red line: i.e. by how much embodied pollution has fallen relative to what we would have predicted back if used 1972 import patterns

Tables

• Column 4 compares total pollution embodied in US manufacturing to the amount of pollution US would have produced if its manufacturing mix had stayed same as in 1972.• Real value of US manufacturing grew by 60% over 1972-2001

period, but direct+indirect emissions only rose by 17%• i.e. US production reoriented toward less polluting sectors

What is causing the Green Shift in imports?

• Likely: US import patterns have changed:– “imports of highly polluting industries...grew faster

than domestic production, – “[but] they grew more slowly that imports in general.– “[In contrast,] clean imports grew faster than

domestic production, – “but clean imports also grew faster than overall

manufactured imports– “in many cases domestic manufacturing of relatively

clean industries actually declined. " (p.79)

US production of labor intensive goods actually fell

– “Apparel imports grew 9.8 percent annually, while domestic apparel production declined by 0.4 percent.

– “Shoe imports grew 6.2 percent annually, while domestic production declined 4.8 percent.

– “Imports of games and toys grew 12 percent annually, while domestic production declined 1.1 percent (p. 79)

Levinson’s Conclusions

• there has been a large “green” shift, whereby US imports have lower pollution content now than they did 40 years ago

• the shift in imports (54%) has been larger than the green shift in US manufacturing production (27%)

Criticisms (many of them self-levied)

• Only focuses on 4 air pollutants– but most other studies also restrict attention to air

pollutants

What’s the right measure?

• Levinson calculates pollution avoided by imports– if the US had produced all these goods, instead of

importing them, by how much would its own pollution rise?

• That’s not the same as the pollution content of imports– How much pollution was generated in order to

produce the goods that the US imported?

Embodied Pollution

• To measure pollution actually embodied in pollution, you’d need to know the DIECs for foreign production, not US production

• Recall Weber & Matthews (2007)– found “ ‘CO2 embodied in U.S. imports doubled ...

between 1997 and 2004,’ while U.S. imports grew 67 percent.” (p.78)

– Suggests shift was brown, not green

What could explain this?

• Different dependent variables: C02 versus unweighted sum of SO2, NOX, CO, and VOCs

• US’ trade partners are shifting into production of goods that are clean (relative to other US production) but dirty (relative to other foreign production) abroad

• suggests a negative correlation between US and foreign DIECs– but Gamper-Rabindran 2006 finds positive

correlation between US and Mexican direct emission coefficients (of between .54 and .62)

Do we want to look at correlation coefficients for all goods being produced/traded?

• No.• We should look at the correlations between

the goods that have seen the largest changes in their export share

Thought experiment

• Let * denote Chinese production; no asterisks indicates US production

• Suppose could partition set of all possible products into two sets: {V} and {Z} whereby for any two goods i and j in set Z, ei*/ej* = ei/ej

– that is, there’s a perfect correlation between relative pollution intensities

• but, for goods in set Z, you see negative correlations between the relative intensities, i.e there exist goods k,l in set V such that ek*<el* and el>ek

Question

• if trade frictions fall, in which goods are you likely to see trade expand the most?

• It might well be the ones in which you’ve got a comparative advantage?

• what drives comparative advantage?– one source: differences in the e’s• should see the greatest expansion occur in V set,

because these are the ones where China has a bigger emission advantage than the US

Simultaneous Green shift and Brown Shift

• So trade grow fastest for goods in which US emission coefficient is lower than average but China’s is higher– see US pollution avoided via US imports fall, but

Chinese pollution embedded in US imports rise

But do we really think that differences in the e’s are what drive comparative advantage?

Other sources of comparative advantage

• relative factor endowments (Copeland and Taylor)

• labour costs• in constrast, PACE are small– US EPA estimates env'l control costs amount to 2

percent of US GNP.

Other factors that determine firm location and pollution intensity

Blueprint costs

• In a 1990 UN survey of MNCs, most MNCs claim they follow the health, environmental and safety standards as set in their home countries.

• Knogden (1979) finds 90% of the W. German firms operating in LDCs make this claim too, on the grounds that re-engineering is more costly than overcomplying.

Expectations re future regulations

• We expect environmental regulation to get stricter over time in developing countires– so firms would rather design clean overseas plants

now than pay to retrofit later

Raising Rival’s Costs

• Since rich-country MNCs have the green technology already, they have comparative advantage in setting up in LDCs that have stricter env'l policies.

• So while they might set up in LDCs rather than DCs because of variation in other costs, they'll choose the greener of the LDCs and let the locals do more of the investment in the brown LDCs.

Other issues that need to be considered when testing for PH Effects

Endogeneity of Pollution Taxes

• Suppose MD is increasing in production/pollution– should see stricter policies in regions with more

production because they have more pollution. • Empirical implication: – if you treat pollution taxes as exogenous, you

might well observe a positive correlation between pollution tax rates and total pollution

Levinson (1999) State Taxes and Interstate Hazardous Waste Shipments, AER 1999

• Looks at state-level taxes on hazardous waste disposal and total amount of waste being disposed

• Baseline approach: – treat tax rate as independent variable– conduct cross-sectional regression• finds positive correlation between a state's hazardous

waste taxes and its imports of hazardous waste

• Re-runs regressions putting in state-level fixed effects– the fixed effects should pick up time-invariant “suitability”

of certain states as sites for waste dumps (and hence to have higher waste disposal taxes)

– finds sign of correlation is reversed: • higher waste taxes have a negative impact on interstate waste

shipments for disposal– but is statistically insignificant

• also conducts 2SLS and uses a natural experiment– In each case, also see sign on tax rate flip (relative to

baseline model)

Next: Environmental Policy in an Open Economy with Market Failures