Environmental Engineering Reference
In-Depth Information
environmental policy. The idea here is that US states are required to have a plan to bring
all counties into compliance with the federally mandated air quality standards. Hence
those counties that have not met the standards should on average have experienced an
increase in the stringency of environmental regulations after the Act came into e
ect.
Many studies use abatement costs (usually from the USA) as a proxy for the stringency
of environmental policy. This is problematic for several reasons. First, the data are
obtained from
ff
rms to isolate abatement costs from
other production costs, especially when the design of the production facility has been
in
fi
rm-level surveys. It is di
cult for
fi
rms have incentives to be
strategic in reporting abatement costs if they think the information might in
fl
uenced by the environmental policy regime. Moreover,
fi
uence future
environmental regulation. Even if abatement costs are accurately reported, they are
endogenous and, as Keller and Levinson (2002) have emphasized, this can create biases.
To see this, suppose there is heterogeneity across
fl
rms within an industry in the costs of
responding to pollution regulations. If the competitiveness hypothesis is correct, then
stringent environmental policy will drive away the
fi
fi
rms with the highest abatement costs.
When
fi
rms are surveyed and asked about their abatement costs, only the low-abatement-
cost
rms will remain. So in cases where the competitiveness hypothesis holds, measured
abatement costs may be low even in regions with stringent environmental policy.
To test the competitiveness hypothesis, researchers estimate the e
fi
ects of the stringency
of environmental policy on the pattern of trade, direct foreign investment, or plant loca-
tion decisions. Early studies in this area used cross-sectional data, and almost universally
found no support for the competitiveness hypothesis (see Ja
ff
e et al., 1995 and Levinson,
1996 for surveys). The environmental policy variable was either insigni
ff
cant or had a sign
opposite to what was predicted (that is, it was sometimes found that more stringent envi-
ronmental policy was positively correlated with inward investment or export success in
pollution-intensive industries).
For example, Tobey (1990) used a sample of data on net exports of pollution-intensive
goods from 23 countries. He used a country-level measure of the stringency of environ-
mental policy and estimated its e
fi
ff
ect on trade
fl
ows. Data on factor endowments con-
trolled for other factors that a
ff
ect trade
fl
ows. The environmental policy variable was not
statistically signi
cant. Kalt (1988), Grossman and Krueger (1993) and others used US
data and asked whether industry-level variation in the stringency of environmental policy
(measured using industry-level abatement costs) a
fi
ff
ected US industry-level net trade.
Data on other industry characteristics (such as tari
ff
s, skilled labor intensity, etc.) con-
trolled for other factors that a
ff
ect trade
fl
ows. These studies found either that abatement
costs are not statistically signi
cant or (as in the case of Kalt, 1988) that higher pollution
abatement costs were associated with increased net exports. Levinson (1996) and others
used similar techniques to estimate the e
fi
ect of environmental policy on plant location
and again found little or no support for the competitiveness hypothesis.
Although some interpreted these results as support for the Porter hypothesis, recent
research has shown that omitted-variable bias and endogeneity problems are the most
likely factors responsible for the failure to
ff
fi
nd support for the competitiveness hypothe-
sis in early studies.
Omitted-variable bias arises when factors correlated with the stringency of environ-
mental policy, and which also a
ows, are not controlled for. As
an example, consider countries that are rich because they are capital abundant and have
ff
ect trade and investment
fl
