A brief description of the evolution of the pollution haven hypothesis runs as follows.
• Between 1960 and 1995 the share of pollution intensive industries declined in OECD countries and increased in Latin America and Asia. This led to speculation that changes in income, regulation and energy prices were the cause (Mani and Wheeler, 1998). A consensus arose from empirical findings however that regulatory differences did not matter, except with respect to highly polluting industries.
• Forming an important context to the debate was the finding of an environmental Kuznetz curve.
While increases in GDP may be associated with worsening environmental conditions in very poor countries, economic growth appears to benefit air and water quality once some critical level of income has been reached (Grossman and Krueger, 1994).
• This inverted U-shaped relationship for environmental quality suggests that there is an induced policy response as citizens get richer. Another possibility is that structural shifts or other causes may be at work.
• Available econometric evidence did not in general support the contention that industry was attracted to low environmental standards. Grossman and Krueger (1993) using cross-section data on US imports from Mexico found, if anything, more rather than less imports in ‘low pollution’ industries.
These were modest but counter-intuitive results. Studies concluded in general that environmental regulations have little effect on location decisions (Jaffe HWDO1995) and that the pollution haven hypothesis could not be proven.
• Environmental costs were thus considered to be merely one of a broad number of factors determining location, including wage costs, skill formation, quality of infrastructure, access to inputs, income levels, openness and size and growth of potential markets in which to sell.
• The costs of adhering to environmental regulations are typically a small part, on average 2 to 3 per cent, of total production costs of most firms (Adams, 1997, OECD, 1998, UNEP, 2000).
• Rather than seeking lax environmental enforcement, multi-national enterprises would generally prefer consistent enforcement (OECD, 1997).
• Foreign investment, primarily from the US or Japan appears to be attracted to the services or the signals about the local investment environment that stringency provides. The inverse correlation between national environmental performance and perceived corruption (TI, 2000) may also mean that investment is attracted to countries with environmental regulatory stringency.
• Source-country shareholders may not want to be accused of allowing lower technological standards to be used abroad.
Despite voluminous research, until the mid-nineties econometric evidence to back up the pollution haven hypothesis eluded researchers. Alongside the above reasoning, the strong growth in FDI in developing countries was considered to reflect primarily the low cost of labour and the abundance of resources. It was thought that companies were unlikely to use a lower grade or dirtier technology than that which they employed in developed countries. They would tend to invest in state-of-the-art technologies regardless of
location, preferring to reap the scale benefits of standardisation in environmental, health and safety management systems, rather than simply exploit weaknesses in local legislation. They would also respond to pressure from the public at large and civil society. The dispersion of FDI and ensuing trade could therefore raise the environmental standard of production worldwide, it was argued.
The popular debate on the pollution haven hypothesis became more polarised with the demonstrations at the World Trade Organisation (WTO) meetings in Seattle and Genoa. At the same time the research consensus that regulatory differences do not matter altered somewhat as more thorough studies were embarked upon.
Recent research papers have used time-series panels rather than cross-section data. These studies can control for unobserved attributes that are correlated with regulatory stringency or economic strength. Another issue hinted in the above discussion of environmental regulations becoming stuck is that regulatory stringency may itself be endogenous. Location decisions and trade may respond to environmental regulations but these regulations may in turn react to location decisions and trade. Recent modelling work has attempted to separate the effect of regulatory stringency, without the confounding effect of trade on stringency. As a result of these changes in approach recent papers have in fact found a statistically significant effect of environmental regulations on trade and investment. Brunnermeier and Levinson, RS FLW cite studies supporting the pollution haven hypothesis, if only to a small degree, by Bartik (1989), Levinson (1996), van Beers and van den Bergh (1997), Osang and Nandy (2000), List and Co (2000), Ederington and Minier (2003) and Levinson and Taylor (2004). Compelling evidence is also found by Waldkirch and Gopinath (2004).
Empirical investigations of the pollution haven issue typically fit models of the form:
Yi = Fi + R i+ Ti+ i
where the dependent variable Y is some measure of economic activity. Economic activity can be represented by such items as inbound or outbound foreign direct investment, new plant openings, employment, pollution, net exports or net imports. These can variously represent location choice, output effects or input effects.
Regressor F represents factor endowments, such as human and physical capital and infrastructure, R is a measure of stringency of environmental regulations, T is trade barriers and LVDUDQGRPHUURUWHUP7KH pollution haven effect would be captured if ∂Y/∂R < 0, that is, if the economic activity is negatively affected by increased regulatory stringency.
There are several estimation issues that have come to dominate recent investigations, the main ones being:
7KH XVH RI SDQHO GDWD RU FURVVVHFWLRQ GDWD. The later studies use panel data, over a period of six to seventeen years. The counter-intuitive findings of the early studies using cross-section data could be explained by unobserved industry or country characteristics that are likely to be correlated with the propensity to impose strict regulations or the propensity to manufacture and export polluting goods. In a cross-section model omission of these variables risks giving misleading results. Panel data can control for characteristics and may overcome these problems.
(QGRJHQHLW\ RI HQYLURQPHQWDO SROLF\This can invalidate the model as it stands. For example, if greater economic activity leads to higher income, which in turn leads to demands for higher environmental quality, then environmental stringency could be a function of dependent variables trade or investment. The econometric solution is to employ instrumental variables and apply two-stage least squares. When endogeneity is accounted for, Levinson and Taylor (2004) among others found that the U.S. industries that experienced the largest increases in pollution abatement costs (a proxy for regulatory stringency) during the 1970s and 1980s also experienced the largest relative increase in net imports.
Measurement of environmental stringency: This is a source of difficulty that probably surpasses the problems associated with the data on foreign direct investment. Proxies are invented to measure stringency.
These include environmental indices, which can suffer from being objective. Ratios of emissions to output can be used, based on U.S. levels as developed in the World Bank’s industrial pollution projection system (IPPS). Their application to non-U.S. sectors is to assume common technologies, regulations and enforcement. Another measure of regulatory stringency is pollution abatement costs, which can comprise
pollution abatement capital expenditures or operating costs. Keller and Levinson (2002) find robust evidence that pollution abatement costs, when adjusted for state industrial composition, have a modest deterrent effect on the value and count of new foreign investment projects. A doubling of their abatement cost index is associated with a less than 10 per cent decrease in foreign direct investment. Another proxy measure can be constructed from actual pollution levels, such as ambient air quality, which have the benefit of objective measurement as do emissions data to a lesser extent.
$VWXG\RIWKHH[WHQWWRZKLFKFKDQJLQJHQYLURQPHQWDOVWDQGDUGVKDYHDOWHUHGSDWWHUQV RILQWHUQDWLRQDOLQYHVWPHQW
The study “Pollution abatement costs and foreign direct inflows to U.S. states” was undertaken by Keller and Levinson (2002) and published in the 5HYLHZ RI (FRQRPLFV DQG 6WDWLVWLFVRather than compare regulations in different countries, the authors look instead at inward foreign direct investment to the US. There are advantages in using US data and different levels of regulatory stringency apply between US states. The authors use a 17 year panel for their model of the broad form:
Ln (FDIst ) = β ln (S*st ) + γXst + δt + ds + εst
where FDIst is foreign direct investment in state s during year t, S* is a carefully constructed index of the relative stringency of environmental regulations in each state in each year, and Xst is a set of other state characteristics that may affect inward investment, such as market proximity, taxes, energy costs, land prices, wages, unionisation and so forth. Year dummies are given by δt, and unobservable state characteristics that would otherwise impart an omitted variable bias are represented by ds which are state fixed effects. εst is an error term.
The index of regulatory stringency S* measures state pollution abatement costs using the PACE data but, importantly, it is adjusted using each state’s industrial composition, at the level of 2-digit SIC codes. The stringency of environmental legislation differs much more across countries of the world than across US states, however the variation in other regressors is also greater across countries than across states. Thus, claim the authors, the analysis does not necessarily underestimate the sensitivity of FDI location to environmental stringency at international level.
From their model they obtain an elasticity ( β ) of FDI with respect to regulatory stringency of –0.079 for all manufacturing industry (statistically significant at 10%) and –0.198 for the chemical sector (statistically significant at 5%) which is used as an example of a pollution-intensive industry.
In sum, the authors find ‘robust evidence’ that pollution abatement costs have had a moderate deterrent effect on foreign direct investment, which is somewhat stronger in the case of the representative pollution-intensive sector.
Variants that test specifically whether competition for mobile capital induces countries to set suboptimal standards have not been applied. Several articles have tested for and found strategic behaviour in the US.
In sum, empirical work investigating whether economic activity responds to regional regulatory differences has arrived at differing conclusions. These tended initially to point to a counter-intuitive, modest, attractive effect of regulatory stringency. However several recent studies that use panel data and control for endogeneity find statistically significant though small pollution haven effects. It does not appear to matter whether the studies look across countries, industries, states or counties, or whether they examine plant location, investment, or international trade patterns.
Such a finding is not inconsistent with studies of international capital flows and company tax in general. In their meta-analysis, which is a synthesis of research results from the literature on taxation and FDI, de Mooij
and Ederveen (2007) find a 2% to 3.9 % increase in FDI for a one point change in the tax rate. That is, a tax rate decline from, say 30 per cent to 29 per cent.
$VWXG\RIZKDWPDNHVDQH[LWWKUHDWFUHGLEOH
If a firm’s exit threat is credible, the more likely it is to be able to exert influence on legislative and fiscal provisions. Four relevant characteristics have been highlighted by Massey (1999). These are :
• 7KHVL]HRIWKHIL[HGLQYHVWPHQW - the larger the required fixed investment, the more costly relocation would be and thus the less credible the threat to exit.
• $EVHQFH RI SURGXFW GLIIHUHQWLDWLRQ - if the firm’s products are traded and indistinguishable from those of another firm, extra costs incurred by environmental laws or taxes cannot be passed on in the form of a premium that consumers can recognise as environmental. The firms are price-takers and their exit threat could be credible.
• +LJKDEDWHPHQWFRVWV - if regulated abatement or a tax constitutes a high share of expenses in general then the firm can credibly claim to be vulnerable unless, that is, other characteristics give them price-setting power in the manner outlined in the previous section, or technological options.
• 'HSHQGHQFHRQH[KDXVWLEOHUHVRXUFH±As the resources near depletion the exit threat becomes more credible but then the loss is smaller.
$QDO\VHVRISROOXWLRQLQWHQVLYHVHFWRUV
• It is possible that firms in the more pollution intensive industries, such as chemicals, metallurgy, and pulp and paper, and industries in the resource extraction sector gravitate to pollution havens and there could be a tendency for environmental policy to acquiesce. This appears to have been experienced in both developing and industrialised countries (OECD, 2002).
• The pattern of investment in China, however, in pulp and paper and other polluting industries suggests that foreign investment is attracted to provinces with more stringent regulations, the opposite of the pollution haven hypothesis (Dean HW DO, 2004). Chinese-sourced equity, by contrast, gravitates to states with low standards.
• In the steel industry, an OECD-wide carbon tax of $ 25 per tonne of CO2 without revenue recycling, would reduce OECD steel production significantly, by an estimated 9 per cent, the reduction being concentrated at the heavily polluting end of the sector. Relocation to non-OECD countries is estimated to raise their production by almost 5 percent, implying a world fall in steel production of 2 per cent and, despite higher emission intensities in non-OECD countries, global emissions from the sector would decline by 4.6 per cent due to cleaner processes in the OECD area. If revenues were recycled back to the steel industry as an output subsidy, OECD production would restructure towards cleaner processes and the reduction would be less than 1 per cent and global emissions reductions in the sector would reduce from 4.6 per cent to 3 per cent (Maestad, 2003).
The implications of the re-instatement of the pollution haven hypothesis (though its effects may be very minor overall) are not that regulations or ETR should be scaled down, as if their costs outweighed the benefits overall. Rather, efforts should be made to induce host countries, via trade or EU accession agreements, to strengthen their environmental policies. The results highlight the logic of co-ordinated global environmental rules (Waldkirch and Gopinath RSFLW). This would not necessarily mean uniform pollution standards because of differences in the capacity to assimilate certain pollutants. It may be efficient for polluting industries to move to regions that put less emphasis on environmental quality provided that the
regions do so for appropriate reasons, meaning that there is no market failure, political failure or redistributional concern involved and there are no global or trans-boundary externalities (Bhagwati, 1993).
Global warming is however a case of global externalities that suggests uniform world rates of tax/subsidy on emissions/sequestration, given that a unit of carbon dioxide tends to have a similar global warming effect regardless of location (Parry, 2003).
Another obvious possible response that gives ETR the advantage over regulations is to avert industrial relocation by using the revenues to implement mitigation measures. This is the subject of Work Package 5.
9LHZVRIGHYHORSPHQWDQGLQGXVWU\DJHQFLHV
The views of development agencies and business representatives were sought, in order to obtain first hand observations of how ETR influenced the location decisions of foreign direct investment and whether relocation had been prompted. The organisations contacted were not representative, but their responses provide some insights. They were asked two main questions:
(1) Has the introduction of carbon/energy taxes or eco-taxes influenced Foreign Direct Investment in your country?
(2) Are there instances that can be cited?
Their responses were as follows:
A response for East Germany indicated that there is a certain effect. Companies know that Germany is a high-cost and high-tax location, but not only because of eco-taxes. For some companies, Germany is no longer on their short-list and they search for ‘low-cost’ countries.
On the other hand there are a lot of initiatives and incentives offered by the German government to reduce the cost and tax burden for companies, like the “Energie-Einspeise Gesetz” which offers many incentives to companies engaged in new energy, production technologies, photo-voltaics HWFHWHUD, in order to reduce the capital burden.
For some companies investigating conditions in the UK, the Climate Change Levy is considered to be a discouraging factor but then all states have some sort of climate change imposition due to the Kyoto Protocol. Although there may be a push factor for some firms to leave, there is little pull factor in other states to encourage them to relocate. Meanwhile the UK’s targets in respect of the renewable energy strategy are themselves an attraction to overseas investors with this sort of technology.
There are apparently no instances to date of relocation due to the Climate Change Levy. Some firms in the energy-intensive sectors did relocate to Eastern Europe but this was not due to the Climate Change Levy, as other factors such as labour costs come into play.
With respect to the Netherlands, it is instructive that their Foreign Investment Agency highlights the advantages of the location for the chemical industry. Features that it draws attention to (as cited in a survey of U.S. chemical companies) are:
“Transparent and consistent environmental policies and a quick permitting process”, stating that the government’s emphasis is on reducing and simplifying environmental rules and regulations.
Other indications on the issue are to be found in some recent debates. For example, in the run-up to the introduction of ETR in the form of the Climate Change Levy in the UK, virulent national debates occurred between the authorities and the sectors most at risk (though not apparently from those who are gaining).
Sectors citing vulnerability to competition included the steel industry, aluminium, fertilisers, salt, horticulture, and cement, the last reporting that it was acquiring cement plants in Greece and Poland.
Recent discussions on the proposals in REACH (Regulation on the registration, evaluation and authorisation of chemicals) are pertinent to the issue. There is concern that the chemical industry could transfer some operations elsewhere. A recent report by consultancy KPMG examined the four sectors, automotives, flexible packaging, inorganics and electronics. The report however does not indicate that industries will
relocate because of REACH. In the case of the automotive industry it considered that proximity to customers and heavy investment in the EU were the main reasons for companies staying where they are.
In the literature review by Brunnermeier and Levinson (RSFLW), there is a summary of surveys undertaken over the last two decades that try to understand industry location decisions. Two surveys are worth noting.
UNCTAD (1993) sampled 169 corporations with sales exceeding $1 billion (1990) most of which claimed that corporations’ environmental, health, and safety practices overseas were determined by home country regulations. The second example is by the U.S. General Accounting Office which surveyed 2,675 wood furniture manufacturers in Los Angeles in 1988-1990. In the region of 1 per cent relocated at least some part of their operations to Mexico, citing labour costs and pollution control costs as significant factors affecting their decisions. Brunnemeier and Levinson are of the view that the results of these surveys are inconclusive.
&RQFOXVLRQVRQ)',
The debate continues but the main points that have emerged to date provide helpful context.
• No studies have been identified that specifically investigated the issue of concern here, that is, the effects of ETR, in Europe or elsewhere, on location decisions.
• Studies of the relocation effects of environmental regulatory stringency have been undertaken and these were examined for want of more pertinent analyses.
• There is a large literature investigating the hypothesis that firms relocate to areas with less stringent regulations, that is, to so-called ‘pollution havens’.
• Discussions with development agencies and industry associations give mixed indications but flag important issues, in particular the roles of labour costs and taxation in general, in the location decision. Regulations regarding health, safety and the environment, environmental policies that are transparent and consistent and a quick permitting process also feature.
• Early empirical investigations of the pollution haven hypothesis based on cross-section data found scant evidence for the hypothesis. Some counter-intuitive indications emerged that the reverse could be the case: that compliance costs or abatement costs could have a positive effect on net exports.
This tended to be explained by the fact that environmental compliance costs are for the most part quite small and by the greater importance attached to labour costs, local markets, and natural resources where relevant.
• Recent studies that use more extensive records and panel data are able to control for econometric problems (unobserved heterogeneity and endogeneity, in particular) and these typically find some evidence of a minor pollution haven effect due to environmental regulations. The other attracting factors are market size and labour costs.
• To discourage the ‘race to the bottom’ in face of threats of undercutting by countries that lower standards to compete for investment, host countries could be induced to strengthen their environmental policies via trade agreements that promoted appropriate global safeguards and taxes.
• In sum, the mixed evidence for the pollution haven hypothesis including the recent work that supports the hypothesis, does not necessarily apply to environmental tax reform. In particular, the manner of ETR implementation can help to avert industrial relocation by careful targeting of the recycled revenues to mitigate the effects of the tax side of the ETR.