Environmental Engineering Reference
In-Depth Information
highly developed infrastructures that support industrial manufacturing. And suppose
manufacturing is pollution intensive. Then we would expect such countries to have rela-
tively stringent environmental policy because of their high income. But we would also
expect such countries to be net exporters of manufacturing goods if the e
ff
ects of their
infrastructure and capital abundance are stronger than the e
ects of environmental policy
in determining production location. Consequently it would not be surprising to
ff
nd a
positive correlation between the stringency of pollution policy and net exports of pollu-
tion-intensive goods. However, it would be an error to conclude from this that tightening
up environmental policy would increase exports of manufacturing. In fact the reverse is
true in our example.
To deal with this problem, one could control for capital abundance and infrastructure.
But there are many other factors that a
fi
ows, and the researcher
cannot control for all of them. This is the problem of unobserved heterogeneity, which
can lead to omitted variable bias. Recent work has used panel data to deal with such prob-
lems. A good example is Keller and Levinson (2002), who looked at the e
ff
ect trade and investment
fl
ff
ects of the strin-
gency of environmental policy on foreign investment in
ows into US states. They had a
panel of data with both time series and cross-sectional variation. First they pooled all of
their data and replicated earlier work that found no e
fl
ff
ect of environmental regulation on
investment
ects to control for unobserved
heterogeneity, they found support for the competitiveness hypothesis: all else equal, more
stringent environmental policy reduced net foreign investment into a state.
Endogeneity bias has also a
fl
ows. But when they used state-level
fi
xed e
ff
ected results. As an example, suppose that governments
are less likely to tighten up environmental policy in industries that are subject to strong
pressure from imports. Then we would expect to see a negative correlation between the
stringency of environmental policy and net imports. This is opposite to what the com-
petitive hypothesis would predict, even though it is competitive pressure that is causing
governments to act in this way and induce the negative correlation. These types of prob-
lems have been dealt with by explicitly confronting the endogeneity problem using
techniques such as instrumental variables. Levinson and Taylor (forthcoming) and
Ederington and Minier (2003) studied the e
ff
ect of US environmental policy on US net
imports. Both studies found evidence of endogeneity bias and both found strong support
for the competitiveness hypothesis once they corrected for endogeneity using instrumen-
tal variables.
There is now a growing body of evidence coming from studies using panel data that are
consistent with the competitiveness hypothesis. Levinson (1999) found that (all else equal)
states with high taxes on the processing and disposal of hazardous waste were less likely
to attract shipments of such waste. Some of the most convincing evidence comes from a
series of studies that have found that the Clean Air Act in the USA has had a signi
ff
fi
cant
e
ect on the location of new plants (at the country level) in pollution-intensive industries.
Becker and Henderson (2000) found that new plant births in pollution-intensive indus-
tries were 26-45 percent lower in counties that were not in compliance with air quality
standards required by the Clean Air Act (plants in such counties are subject to more strin-
gent environmental policy).
Evidence on outward direct investment from the USA is more mixed. Hanna (2006)
found evidence that the Clean Air Act induced multinational
ff
rms to shift some produc-
tion out of counties with more stringent environmental policy. Eskeland and Harrison
fi
