Are consensus democracies more environmentally effective?

Abstract

Are consensus democracies more likely to enact environmentally friendly policies than majoritarian democracies? The relationship between democratic type and environmental effectiveness is more complex than previously assumed. What matters are not only institutions but also the nature of the policy issues, the types of institutions and veto players in the system. Four measures of effectiveness emerged from the data: mundane environmentalism; environmental taxation; conservation; and nuclear energy production. Environmental effectiveness in democratic systems differs depending upon which measure of effectiveness is examined and whether or not one looks at representational or organisational aspects of democracy.

Keywords:

Scholars interested in democratic systems have tended to classify democracies into two types – majoritarian and consensus – based upon the predominant institutional configurations found within the countries (Lijphart 1984, 1999; Powell 2000). 1 While debate has focused on the virtues and vices of each system, less attention has been given to the relationship between environmental effectiveness and a country's majoritarian or consensus status. 2 The purpose of this study is to explore the relationship between majoritarian and consensus democratic institutions and environmental effectiveness. I ask if consensus democracies are more likely to enact environmentally friendly policies by looking at policy responses through the 1990s and matching these with Lijphart's (1999) scales of democracies. Why is this topic important? Many scholars believe aggregated preferences are funnelled through institutional configurations to arrive at macro-level outcomes (e.g. Coleman 1987). If this is true, it may not be enough for the citizens of advanced industrial democracies to want environmental protection; the institutional configuration may constrain what outcomes are possible. The debate in newly democratic states about the ‘best’ type of democracy is real (Agh 2001), and the institutional choices made now may impact environmental performance in the future.

Of the 23 advanced industrial countries 3 studied, 18 are European. Public attention to environmental matters has remained widespread throughout Europe over the past three decades (Dalton 1994; Dalton and Rohrschneider 1998; European Opinion Research Group 2002); further, the data confirm a trend of increasing popular interest in environmentalism (Dalton 1994; European Opinion Research Group 2002). There is therefore – at least in Europe – widespread public support for environmental issues. These countries nevertheless differ in the degree of their environmental effectiveness (Palmer 1997). Why do we see this difference? I propose that these differences may be due to the institutional characteristics of each country.

After testing the relationship between environmental quality and institutional configurations, Lijphart (1999) claimed that consensus democracies are more environmental. The link, however, between the environment and institutions is more complex than Lijphart and most analysts assume. I suggest that what matters are not only institutions, but also the nature of the issues, the type of institutions and the degree to which green parties exert public pressure on policy-makers. Lijphart (1984, 1999) identified ten traits that typify each type of democratic system. Majoritarianism is characterised by the concentration of executive power in one-party and bare majority cabinets, majority party control of the cabinet, a two-party system, disproportionality of elections, interest group pluralism, unitary and centralised governments, unicameralism, constitutional flexibility, absence of judicial review and executive control of the central bank. Consensus democracies are characterised by executive power-sharing in broad coalition cabinets, balance between executive and legislative power, proportional representation, interest group corporatism, federalism, bicameralism, constitutional rigidity, judicial review and independence of the central bank. Each type – majoritarian and consensus – is an ideal type. Very few democracies approximate the ideal; discrepancies will be dealt with in the operationalisation of the variables.

Clarifying the Debate

Several questions surround these two types of democracies. Why has there been a debate over which type is ‘best’? What sides have been taken, and what evidence is there for choosing sides? What impact may this have on environmental effectiveness? To date, most scholars working within this area have ignored environmental issues.

The argument for majoritarian democracies proceeds on the assumption that one-party majority governments produced by plurality elections are more decisive and more effective policy-makers. As early as 1896, Lowell claimed that one-party majority cabinets were necessary for effective policy-making, implying that the representative nature of a government comes at the expense of effective government (Beer 1998). This suggests that environmental issues somehow get ‘lost’ among the multiplicity of issues found in consensus democracies. In other words, majoritarian democracies would be more decisive in instituting environmental policies.

Furthermore, veto player theory (Tsebelis 1999) states that the number of significant laws produced by a coalition government is significantly lower than the number of laws produced by single-party governments or by coalitions with partners that agree (Tsebelis 1999, p. 3). One would thus expect to find more significant pieces of legislation in countries with fewer veto players (Tsebelis 1999, p. 6). At-large electoral systems with proportional representation (characteristics of consensus democracies) provide extensive veto capabilities (Herzberg and Ostrom 1986). This implies that in consensus democracies, one would expect to see less environmentally significant legislation. Conversely, majoritarian democracies decrease the number of veto players, thus favouring environmental legislation.

Environmental policy-making will also be influenced by interest-group arrangement. Majoritarian systems are characterised by pluralistic interest-group politics. Pluralists typically see policy as being made through interactions among interest groups that are capable of pressuring governments (Dahl 1982). ‘In such an open and competitive interest group community, the success of environmental policy should depend on the skills and resources of environmental interest groups compared to those of their competitors’ (Strom and Swindle 1993, p. 7). More important to this research is that pluralistic environments would allow environmental interest groups the opportunity to compete with other interest groups for governmental influence.

A corporatist interest group environment – characteristic of consensus democracies – provides mixed opportunities for environmental effectiveness. The privileged position of business and labour in this model may be inhospitable to environmental efforts (Strom and Swindle 1993, p. 7), and corporatism has been said to follow a materialist logic of economic growth that is in fundamental disagreement with ecological concerns (Crepaz 1995). Corporatist states, however, may be more flexible in adjusting policy, and this could benefit environmental reform (Strom and Swindle 1993, p. 7; see also Katzenstein 1985). Several other researchers have determined that corporatism has a positive effect on environmental performance (Jahn 1998; Scruggs 1999, 2001, 2003).

The central argument for consensus democracies builds around the idea of representation (Lijphart 1999). Scholars have argued that proportional representation in consensus democracies provides better minority representation (e.g. Lijphart 1984, 1999). Moreover, consensus democracies offer broader inclusion of interests in decision-making situations. ‘The basic justification of the [consensus] system … is to provide an accurate reflection of public opinion in the councils of government’ (Herzberg and Ostrom 1986, p. 440). This could be particularly important in the area of environmental effectiveness, with consensus democracies giving voice to green/ecological parties. Consensus systems may support more veto players, but they also facilitate formal representation of green issues in government.

Lijphart (1984, 1999) has developed two measures – executive-parties and federal-unitary – to describe consensus and majoritarian democracies. The executive-parties measure deals specifically with the representational aspects of democracies. Therefore we can expect to find a significant and positive relationship between Lijphart's executive-parties dimension and environmental effectiveness.

While much attention has focused on the representative nature of consensus democracies, fewer studies have investigated their federal as opposed to unitary governmental structure. Lijphart does provide a measure for this organisational aspect of government (the unitary-federal dimension), but fails to test it. In other research, scholars have failed to find a relationship between environmental effort and decentralised governance. Strom and Swindle (1993) examined the degree to which environmental effort is related to unitarianism, finding no significant results. Their measure of unitarianism was, however, weak.

Critics of federalism have argued that federalism creates a situation in which we have tyranny of the minority (Riker 1964). In this view, the politics of federal systems becomes dominated by local authorities more likely to have vested interests in political outcomes (Maddox 1976, 1996). This could potentially lead to greater environmental policy response if we assume local authorities are interested in promoting these issues. Critics suggest, however, that federalism interferes with coherent policy-making and leads to excessive waste and duplication (Maddox 1996; Davis et al. 1988; Riker 1964). Such problems are particularly vexing for environmental policies because inefficiencies waste scarce resources (Chappell 2001).

Functional federalists argue that environmental concerns such as clean air or clean water are best addressed at the national or supranational level because of spillovers (Rose-Ackerman 1995). In decentralised systems environmental quality levels can be ‘bid down’ by competing regional and local interests. Opponents have assumed that federalism results in the absence of a strong national government, which could lead to conservatism in policy-making and decreased social innovation (Dicey 1897; Maddox 1996). It has also been argued that federalism involves ‘a large number of protagonists with considerable ideological polarisation [which] increases the transaction costs necessary to implement a policy change, therefore enhancing the probability of retaining the status quo’ (Tsebelis, in Obinger 1998, pp. 245–6).

The critiques of federalism can be challenged on a number of levels (Chappell 2001, p. 63). Social policies in federal states are influenced by several variables, such as the distribution of powers and responsibilities between constituent units, representation of these units at the central state level and the extent of commitment to fiscal equalisation across the units (Pierson 1995). Rather than obstructing the development of progressive policies, federalism can in fact facilitate them. Chappell (2001), examining domestic violence policies in Australia, found that federalism helped to augment the success of the policies because federal systems promoted increased competition, experimentation and innovation in government. Federal systems also display more sensitivity to subnational concerns and provide more access points to government (Kincaid 2001). For example, Weidener (1995) and Czada (1993) found in Germany, and Knoepfel (1997) found in Switzerland, that federalism promoted nuclear power safety primarily because promoters of the environmental interest have been able to make use of the multiple access points in the German and Swiss federal systems (Wälti 2004). Therefore we can expect that federal governmental structures will be positively related to environmental effectiveness.

Given the research to date, more sophisticated analysis is needed to examine the relationship between democratic type and environmental effectiveness. In short, the literature presents an overly simplistic view of the relationship, lacking the proper operationalisation of environmental effectiveness. Given these criticisms, the following sections will elaborate on the operationalisation of the variables and the analyses conducted in this research.

Data and methods

The dependent variable: Environmental effectiveness

Environmental effectiveness at the national level has been operationalised in a variety of ways. Lijphart (1999) used the Palmer Index as his measure of environmental effectiveness. Palmer's (1997) index is a composite of measurements of concern for the environment based mainly on carbon dioxide emissions, fertiliser consumption and deforestation. Lijphart (1999, p. 297) also used energy efficiency as a ‘better overall measure of environmental responsibility’. Other studies have focused solely on pollution (water and air), nuclear energy production and, in some cases, recycling in order to measure environmental effort – e.g. Crepaz 1995 (pollution); Jahn 1998 (pollution); Rohrschneider 1993 (nuclear energy); Scruggs 1999, 2001 (pollution and recycling); Wälti 2004 (pollution). These measures can be improved. The dependent variable in this study is a more robust measure of environmental effectiveness that focuses specifically on the policy responses of countries to environmental pressures.

Measures of environmental effectiveness need to encompass a wide array of policy responses. Dalton (1994) and Dalton and Rohrschneider (1998) found that it is possible to categorise Europeans according to their willingness to support environmental policies. Large numbers support more mundane activities such as recycling, whereas fewer are willing to undertake activities that more greatly impinge upon their lives, such as reducing driving or paying ‘green’ taxes. It is plausible that nation states also differ in their degree of environmental commitment along the same dimensions.

It is important to include a robust measure of environmental effectiveness because operationalisation may influence results. This study's environmental effectiveness measure thus includes a full range of variables: glass recycling, paper and cardboard recycling, public expenditure on waste water treatment, market share of unleaded fuel, percent of territory designated as protected lands, nuclear energy production, percentage of land in strict nature reserves, national red data books for endangered species (amphibians, reptiles, mammal, birds, fish, invertebrates and plants), and taxation of energy use, water use and nitrous oxide production. Data are drawn from the Organisation for Economic Co-operation and Development (OECD), Towards Sustainable Development: Environmental Indicators (www.oecd.org), unless otherwise indicated. 4 If consensus democracies deliver better environmental policies, then we should expect those countries to have higher scores of environmental effectiveness based on this more robust measure.

Paper, cardboard, and glass recycling

Recycling is one response to increased waste and is generally the result of evolving waste management and minimisation practices (OECD 1998, p. 40). Recycling is also an effort that, at the micro level, has been found to have widespread support among the public (Dalton 1994; Dalton and Rohrschneider 1998) and tends to be associated with more mundane, less radical environmentalism. The two variables used to measure recycling in this study are paper and cardboard recycling and glass recycling.

Public expenditure on waste water treatment

Agricultural, industrial, and human wastes affect the quality of surface water, marine water, ground water and drinking water (OECD 1998, p. 42). One response of OECD countries has been increased expenditures on waste water treatment; like recycling, it is associated with more mundane environmentalism. Waste water treatment is measured here in US dollars per capita spent on waste water treatment.

Market share of unleaded fuel

Governments can require that automobile fuel be unleaded, or that unleaded fuel comprise a certain percentage of the total fuel in the country. For example, fuel in Canada, the United States, Japan, Austria, and Sweden is 100 per cent unleaded, whereas less than 10 per cent of the fuel in Portugal and Spain during the time period was unleaded.

Nuclear energy

The production of nuclear energy has been used by some scholars to measure a country's environmental effort (e.g. Rohrschneider 1993). Others consider it to be related to a more fundamental environmental programme (e.g. Dalton 1994). It is difficult to judge whether commitment to nuclear energy is ‘environmental’. Its acceptance in the literature, however, necessitates including it in this study to see if it is connected to other types of environmental efforts.

Percentage of territory in designated protected lands

This measure addresses the percentage of a country's territory that has been designated as a major protected area, including all protected areas of at least 1,000 hectares in the five World Conservation Union (IUCN) management categories. It includes totally protected areas as well as partially protected areas that may be managed for specific uses such as recreation or tourism, or areas that provide optimum conditions for certain species or communities of wildlife. As a result of the human impact allowed in these areas, this measure captures a more traditional approach to environmentalism. Some extractive use within these areas is allowed (World Resources Institute 1999, p. 328), and the focus of the management of these lands is not necessarily on conservation.

Percentage of land in strict nature reserves, wilderness areas, and national parks

This variable is more stringent than the previous percentage of territory in designated protected lands in that lands placed into strict reserves fall into the strictest IUCN categories I, II and III. Lands in these categories have strict limitations on human use, are maintained in a natural state and are closed to extractive use. This variable, as with national red books below, measures a more conservationist-oriented environmental programme.

National red data books

This variable measures the number of redbooks (or endangered species lists) in a country. To be counted, the redbook had to be backed up with laws regarding the management and recovery of threatened and endangered species. Redbooks counted included books for mammals, birds, fish, invertebrates, reptiles, amphibians and plants. 5 A country received a score of 7 if it had redbooks for all families. Scores diminished with each redbook that was missing.

Taxation policies

Countries can also choose whether or not to enact environmentally related taxes. Three types of taxes were included for analysis in this study: energy taxation; water taxation; and nitrous oxide taxation. Data come from a cooperative effort between the OECD and European Environment Agency. The variables measured the percentage of total revenues from environmentally related taxes as related to total tax revenue. Environmentally related taxes are defined by the OECD as ‘any compulsory, unrequited payment to general government levied on tax-bases deemed to be of particular environmental relevance’ (see www.oecd.org).

Factor analysis: Creation of the dependent variable

The above measures of environmental effectiveness were subjected to a principal components factor analysis with a varimax rotation. Four unique factors emerged in the analysis: mundane environmentalism; taxation; conservation; and nuclear energy. All components had Eigen values above one, and together they explained 79.41 percent of the variance. The factor loadings for each factor can be found in Table 1.

Table 1. Factor loadings for policy response variables that measure environmental effectiveness.

Variable	Factor I	Factor II		Factor IV
	Mundane environmentalism	Environmental taxation	Factor IIIConservation	Nuclear energy
Glass recycling	0.61	−0.06	−0.63	0.02
Paper and cardboard recycling	0.73	0.30	−0.06	−0.37
Waste water	0.83	0.01	0.06	0.32
Unleaded gasoline	0.65	0.02	0.12	0.12
Protected areas	0.62	−0.56	0.11	0.02
Strict reserves	0.19	−0.53	0.56	0.22
Redbook	0.20	0.07	0.72	−0.23
Nuclear energy	0.08	0.33	−0.01	0.91
Energy tax	−0.31	0.64	0.30	0.13
Water tax	0.22	0.76	−0.22	0.07
Nitrous oxide tax	0.23	0.61	0.17	−0.36
Variance explained	29.02	20.59	15.10	8.54

Factor I: mundane environmentalism 6

The first factor to emerge from the analysis measures a more mundane form of environmentalism or the ‘everyday’ activities that advanced industrial democracies are increasingly performing: glass and paper recycling, clean water, unleaded fuel and maintenance of recreational areas. In addition, energy taxation was weakly and negatively correlated to this factor.

Factor II: environmental taxation

The second factor to emerge from the analysis was an environmental taxation factor. This factor measures a country's willingness to use economic instruments related to environmental issues. Willingness to engage in energy taxation, water use taxation and nitrous oxide production taxation can be thought of as variables that require more of an environmental commitment than those above. In addition, the land use variables, national parks and strict land use, are weakly and negatively correlated to this factor. This factor is interesting because the use of economic instruments to promote environmental sustainability (such as environmental taxation) has been a theme in ecological modernisation discourse (see, e.g., Barry and Paterson 2004).

Factor III: conservation

Factor III is dominated by the number of national redbooks in a country and the percent of land set aside in strict reserves. 7 Both variables represent conservation issues. Glass and paper recycling are negatively correlated to the factor. This shows that countries differentially apply policies for mundane environmental affairs and conservation policies. 8 Factors II and III, then, can be expected to measure a more committed environmental programme as contrasted with Factor I which focuses on much more mundane and less personally costly environmental issues.

Factor IV: nuclear energy production

The fourth factor is overwhelmingly dominated by a single variable (percentage of total energy production that is nuclear). 9 As a result, this factor does not create a very good scale. Percentage of territory in major protected areas and percentage of land in strict reserves, however, are both negatively correlated to the factor.

The results of the factor analysis mirror what scholars have found at the micro level. Just as individuals provide differing levels of support to different environmental policies, nation states also divide their efforts between policy responses. Dalton and Rohrschneider (1998) found that people are more willing to support environmental activities that do not threaten to alter their current lifestyle. For example, while perceived threat of car pollution to the environment in general ran from 25% (Netherlands) to 64% (Germany), when asked if they avoided using a car, the percentage of people who said that they always or often do ranged from only 4% (Ireland) to 29% (Germany). On the other hand, those who responded that they always or often recycled glass ran from 39% (Ireland) to 87% (Germany) (Dalton and Rohrschneider 1998). It appears that the environmental activities associated with greater personal cost are less likely to be undertaken and undertaken at a different rate than the less costly environmental activities. Macro-level policy responses, therefore, mirror micro-level attitudes. Countries as well as individuals respond differentially to environmental issues.

Ecological modernisation argues that economic growth and environmental protection are not mutually exclusive (Keil 2006; Barry and Paterson 2004). While perhaps a ‘weaker’ form of environmentalism (as distinguished, for example, from deep ecology), yet stronger than the more environmental health indicators discussed above, the ecological modernisation variables loaded onto a unique factor. This suggests that countries that use economic instruments related to environmental issues do not necessarily engage in the other environmental policies responses to the same degree. In other words, ecological modernisation variables, or economic instruments, are a unique measure of environmental effectiveness.

Also telling is that nuclear energy production does not load onto a factor with any of the other indicators. Nuclear energy production has been used as a surrogate measure for environmentalism because it has been presumed that the subject evokes strong responses from the public, especially in Europe. Nevertheless, when examining environmental effectiveness at the level of the nation state, nuclear energy production may not be the best surrogate for environmentalism as it is unrelated to any other environmental policy area.

In sum, the results of the factor analysis show us that the way in which we measure environmental effectiveness is important and may influence our results. Countries differentially engage in environmental policy responses. From the variables examined here, we know that there are at least four unique policy responses related to environmental issues.

Independent variables

The independent variables of interest here include measures of a country's majoritarian or consensus institutional configuration. Using Lijphart's (1999) data to measure institutional configurations on two dimensions – executive-parties and federal-unitary – helps to alleviate the problem of classifying particular countries as strictly one ideal type or the other. Furthermore, the use of two dimensions provides a more accurate representation of the democratic configuration of the individual countries. 10 Table 2 presents the factor loadings for each variable on each factor.

Table 2. Factor loadings for variables that distinguish consensus from majoritarian democracies.

	Factor I	Factor II
Variable	(Executive-parties)	(Federal-unitary)
Effective number of parliamentary parties	−0.90	0.02
Minimal winning one-party cabinets	0.93	−0.07
Executive dominance	0.74	−0.10
Electoral disproportionality	0.72	0.09
Interest group pluralism	0.78	−0.01
Federalism-decentralization	−0.28	0.86
Bicameralism	0.06	0.74
Constitutional rigidity	−0.05	0.71
Judicial review	0.20	0.73
Central bank independence	−0.07	0.71
From Lijphart 1999, p. 246.

Control variables

Presumably a country's environmental effectiveness is determined by more than its institutional configuration. Numerous studies have attempted to link environmental quality with economic and social development (e.g., Arrow et al. 1996; Radetzki 2001). Some posit a linear relationship between development and environmental quality, arguing that as countries have become more developed, their environmental performance has increased (Radetzki 2001). More developed countries are able to provide services such as clean water, recycling and unleaded fuel. The development process, however, places strains on a country's natural resources and may impact the ability for deeper ecological preservation. Others argue that there is an inverted-U relationship between economic development and environmental quality, with poorer and richer countries performing better environmentally than their developing counterparts (Grossman and Krueger 1992, Lucas et al. 1992, Panayotou 2001, Radetzki 1992).

The argument that high levels of economic development benefit the environment has been challenged in other works (e.g. Lekakis 2000). Even those who advocate economic growth admit that the resulting pressures on the environment will find their solution in the adoption of additional environmental policies (Anderson and Blackhurst 1992). De Bruyn (1997, p. 498) has concluded that ‘while the structure of the economy may not impede economic growth from an environmental perspective, the scale effect of economic growth does result in more environmental pressure when not corrected with appropriate environmental policies’.

Others have linked development specifically to policy responses. One work examining OECD countries has claimed that as per capita income increases, public spending for environmental protection also grows (Komen et al. 1997). Carson et al. (1997) found that in the United States local environmental policy effectiveness and regulatory structures appear to correlate positively with absolute income levels across the various states. The precedent exists in the literature for the linking of development and environmental policy. Controversy mainly focuses on whether this relationship is positive, negative or curvilinear. In addition to examining the relationship between the institutional measures and environmental effectiveness, these analyses also explore the degree to which economic development – measured here as the Human Development Index (HDI) 11 – is related to environmental effectiveness.

Other factors have also been put forward to explain variances in environmental effectiveness. Ever since Thomas Malthus's essay on population growth, scholars have been wary of the negative effects of increasing human populations. In The Population Bomb (1978), Paul Ehrlich warned of the dangers of an increasing world population to the world's natural resources. More recent research has also established a link between population density and environmental effectiveness (e.g. Dasgupta and Mody 2001; Fonseca and Wong 2000). Increasing human pressures on natural environments, such as the building of roads, can have serious adverse effects on the natural environment (Trombulak and Frissell 2000). In addition, Strom and Swindle (1993) found that population density was positively correlated with paper and cardboard recycling.

Population density may be positively correlated with the more mundane environmental measures because more people would create more need for trash recycling and water sanitation. Overcrowding can lead to environmental health issues that prompt the government to take action (Yassi et al. 1999). Nevertheless, increasing numbers of people, it could be argued, would actually impede the government's ability to enact conservationist policy responses. In areas that are more densely populated, one could expect minimal land to be set aside for non-human use. Some have argued that the human population is exceeding carrying capacity in many areas of the world, creating stress on the natural environment (e.g. Pimentel and Giampietro 1998). It is necessary then to determine the relationship between the target variables and environmental effectiveness while controlling for population density. 12

In democracies with consensus institutions, green parties would presumably be the parties to push the environmental agenda. The argument for consensus democracies is based on the notion that these institutions somehow benefit environmental interests because such interests can be realised through the participation of green parties in open elections. If this is the case, we should expect to see a positive relationship between the strength of green parties 13 and environmental effectiveness. In addition, the stronger the green party, the more likely there will be environmental policy responses. Bomberg (1996) and Grogan (1999) have both argued that the presence of green parties in the parliaments of western Europe, particularly Germany, was essential in the institutionalisation of recycling programmes. Furthermore, in western Europe, green party platforms have mainly focused on natural resource issues (Kamieniecki 1991).

Green party representation in parliament is nevertheless not a prerequisite for environmental legislation or policy response (Shipan and Lowry 2001; Yishai 2001), nor are green parties the only parties that propose environmental legislation (Allmendinger and Tewder-Jones 2000) or represent the environmental movement in all countries (Kamieniecki 1991). Under certain institutional arrangements (such as a first-past-the-post electoral system), public pressure for environmental reform does not produce a new left or green party, but instead the old left party absorbs the environmental issue (Rohrschneider 1993). In addition, it is possible to have an effective environmental movement (as in the UK) without a strong green party (Poloni-Staudinger 2005).

The introduction of this variable into the equation is of further importance because it allows assessment of the relationship between environmental effectiveness and the institutional variables independently of the effect of green parties. This is important for two reasons: there are majoritarian countries with strong green parties (New Zealand), 14 and this provides a better test of the executive-parties dimension.

Results

Four models corresponding to the four factors uncovered during the factor analysis were tested using robust regression. 15 The factor scales served as the dependent variable in each equation. The independent variables served as predictor variables in all equations (see Table 3).

Table 3. Predicting environmental effectiveness.

	Mundane environmentalism	Environmental taxation	Conservation	Nuclear energy
Constant	−19.49	13.57	0.15	−0.16
Executive-parties	0.40*	.18	−0.37**	0.35
Federal-unitary	0.35**	.15	0.15*	0.22
HDI	20.27	−14.62	0.37**	0.15
Population density	0.01	.01	−0.01	0.56
Green party strength	0.09*	−.01	−0.11**	0.17
F	2.35	.26	6.04	0.59
Prob > F	0.04	.93	0.004	0.75
Number of observations	23	23	23	23
Note: Entries are robust regression coefficients.
*p < .05.
**p < .01.

The relationship between environmental effectiveness and democratic institutions is not as straightforward as previous analyses have suggested. If one were to test only the relationship between mundane environmentalism and democratic institutions, one would conclude that consensus democracies – when measured solely on their executive-parties dimension – do make better environmentalists (0.40, p < .05). Consensus democracies allow more parties with environmental preferences into the decision-making process, making it less of a surprise that consensus democracies – measured on their executive-parties dimension – are positively correlated with the mundane environmentalism scale. The fact that the strength of green parties was also positively correlated lends credence to this claim (0.09, p < .05). The green/environmental parties are the most likely parties through which environmental demands would be made.

These results, however, only tell part of the story. The executive-parties variable was significantly and positively correlated with the mundane environmental measure, but it was significantly and negatively correlated with the conservation measure (–.365, p < .05). A further complication is that green party strength is also negatively correlated to conservationist-oriented policy responses (−.111, p < .01). Why is there a difference between the mundane and conservation equations?

It seems counter-intuitive that increased green party strength is associated with decreased conservationist-oriented policy responses. Veto player theory may shed light on this. It is possible that where green parties are strong, other parties and interest groups are also strong, and they veto the conservation legislation. The executive-parties dimension measures in part the effective number of parties in a given country. The more parties there are in a country, the more consensual the country, and the greater the number of potential veto players.

This implies differing costs for the two types of policies – mundane versus conservation. The costs associated with mundane environmental policies are lower; thus, this type of policy is enacted more easily. The costs associated with conservation legislation are greater, and other veto players may be more willing to veto the legislation. Just as people are more willing to undertake environmental activities that do not require lifestyle changes, other interests in society may act in the same way. For instance, conservation legislation could potentially threaten hunters, foresters and agriculturalists to a greater degree than does mundane environmental legislation. Recycling glass would probably not upset a farmer or a timber industry official. In contrast, habitat restrictions due to endangered species laws would affect the ability of a farmer to develop certain areas for agricultural production, and it would put more restrictions on forest products operations. These groups might allow mundane environmental legislation to pass without a fight, but would veto the more controversial legislation. Thus, various stakeholders can in a sense ‘save face’ with environmental interests by allowing the passage of some environmental legislation while still blocking the legislation that is most threatening to economic development.

How is the unitary-federal dimension related to environmental effectiveness? Researchers have paid less attention to this aspect of democracies, and the debate surrounding the benefits of federalism has been mixed. Federalism, second chambers, rigid constitutions, strong judicial review and independent central banks – characteristics of consensus democracies – can all be assumed to inhibit the decisiveness, speed and coherence of the central government's policy-making compared with a majoritarian system's unicameralism, flexible constitutions, weak judicial review and weak central banks. However, despite this – or perhaps because of it – consensus democratic institutions appear to foster increased levels of environmental effectiveness measured in both its mundane (.353, p < .01) and conservation forms (.153, p < .05). How can this be explained?

Perhaps of greatest importance is the impact of federalism on the increased environmental performance of consensus systems. In consensus democracies, central governments can devolve policy responsibilities to subnational units; for example, it may be easier to coordinate recycling across the Länd of Bavaria than across the entire country of Germany. Federal systems provide greater responsiveness to citizen preferences (Kincaid 2001). Many environmental issues are local in nature; therefore, they benefit from a federal system. Majoritarian systems do not offer this benefit.

Devolution of responsibility could also lead to increased legitimacy of the policy and better enforcement. These findings confirm a growing literature claiming a positive link between federal systems and environmental effectiveness (see Wälti 2004 for a review of this). It should be noted, however, that these findings stand in stark contrast to Campbell and Thomas's (2002) claim that federalism inhibited biodiversity conservation in Canada; however, the Canadian study focuses on just this one case.

One final finding worthy of mention is the positive correlation between HDI and the conservation scale, indicating that as development increases, scores on the conservation scale correspondingly increase (.37, p < .05). This suggests that as countries increase in both social and economic development they will expand environmental attention beyond environmental health issues to issues of conservation.

Results related to environmental taxation

There were no significant results associated with the environmental taxation measure. This is not to say that countries do not engage in environmental taxation; in fact, the data and factor analysis show that they do. This is simply to say that democratic type is not significantly related to variation in the use of environmental taxation. Future research into the issue will need to focus on non-institutional (as measured here) explanations for the differential use of these taxation policies in advanced industrial democracies.

Relevance of nuclear energy as an environmental indicator

As was found with the environmental taxation measure, none of the independent variables proved to be significantly correlated to the nuclear energy scale. 16 Likely, nuclear energy production is due to human pressures for energy, a lack of other energy sources and other non-institutional variables. The lack of significant predictors in this equation is itself important. Nuclear energy is often used as a surrogate measure of environmentalism. Not only does nuclear energy not scale with any of the other environmental measures; the relationship between it and democratic institutions differs from the relationship found with the other measures of environmental effectiveness. This suggests that nuclear energy is not the best measure for examining environmental effectiveness.

Conclusions

Several conclusions emerge from this study. First, countries differentially apply policies for environmental issues. Policy responses fall into at least four categories: mundane environmentalism; environmental taxation; conservation; and nuclear energy production. The split between mundane environmentalism and conservation mirrors a split found in attitudes at the micro level. Second, the federal-unitary variable is significantly and positively correlated with the mundane and conservation measures. It was suggested that federal systems allow greater ease of coordination of environmental policy. Third, the executive-parties and strength of green parties variables behave in the same way – they are both positively and significantly correlated with the mundane environmental measure, but they are negatively and significantly correlated with the conservation measure. This may be due to the higher costs associated with conservation policies, which increase the likelihood that these policies will be vetoed by other actors in the decision-making process. This result would benefit from future research, perhaps in the form of interviews with relevant stakeholders in order to delve deeper into cost-based explanations for these findings. Finally, results for both the environmental taxation and nuclear energy equations were inconclusive, and future research is needed in these areas.

In sum, the mundane environmental results from this study provide support for the results of Lijphart and others who have focused on only one aspect of democratic institutions. This is not surprising given that the measures in other studies also examined mundane forms of environmentalism. Nevertheless, there appears to be a trade-off between mundane environmental policy responses and conservation policy responses. Furthermore, different democratic configurations perform differently on each of the measures. Consensus democracies (executive-parties dimension) appear to do better when environmental effectiveness is measured in the mundane form. Majoritarian (executive-parties dimension) systems do better when environmentalism is measured in terms of conservation. In addition, consensus democratic characteristics related to the federal-unitary dimension are in both instances positively related to environmental effectiveness. So are consensus democracies more environmental? This study shows that although there is support for this claim, the evidence is mixed and is dependent upon how one measures environmental effectiveness.

Acknowledgements

I would like to give special thanks to Robert Rohrschneider for his encouragement on earlier versions of this manuscript. Thanks are due as well to Melissa Mueller, Matthew Loveless and Steven Hoffman and especially to the anonymous reviewers and editors of Environmental Politics for their valuable comments.

Notes

 1. While ‘democracy’ is broader than the division between majoritarian and consensus, Lijphart's categories are narrow models deriving from the Schumpeterian definition of democracy (Schumpeter 1962). Debates surrounding democracy and environmentalism have also focused on direct democracy, deliberative democracy and radical democracy (e.g. Zwart 2003; Eckersley 2004; Baber and Bartlett 2005; Smith 2004).

 2. Authors who have looked at the impact of institutions on environmental performance include Lundqvist (1974), Vogel (1992), Strom and Swindle (1993), Crepaz (1995), Jahn (1998) and Scruggs (1999, 2001, 2003).

 3. Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Japan, Luxembourg, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, United States.

 4. The 1998 OECD data correspond most closely, temporally, with Lijphart's majoritarian/consensus scales. Data on the OECD indicators have been found reliable for measuring environmental efforts (see Crepaz 1995; Jänicke and Weidner 1997; Jahn 1998).

 5. Data for this variable came from Europe's Environment: The Second Assessment, available online at www.biodiversity.environment.gov.au/plants/threatened/list.esp-lists/sched1-ind.html (Australia); www.speciesatrisk.gc.ca/ (Canada); www.eic.or.jp/eanet/ns/nsj-pls3.html (Japan); www.doc.govt.nz (New Zealand); personal communication with Paula Warren, Principal Policy Analyst, Department of Conservation (New Zealand); http://endangered.fws.gov/boxscore.html (United States).

 6. Factor I has an Eigenvalue of 2.943 and explains 26.75% of the variance.

 7. The Eigen value for this factor was 1.65 and explains 15.00% of the variance.

 8. While there may be stricter measures of environmentalism unexamined in this paper, this measure does measure a more committed environmental programme. Traditional conservation or the ‘wise use’ approach to wildlife and habitat management would allow for some extractive use (as captured in the protected lands measure included in the weaker factor). Further, while not a radical concept, species preservation through endangered species legislation does measure a more committed environmental programme than something such as recycling, particularly as we see endangered species lists move beyond protection for only mammals and into protection for less ‘sexy’ species such as amphibians, reptiles and plants. This is not to say, however, that those countries that engage in conservation efforts are more ‘liberal’. Conservative governments (e.g., in the UK) have a long history of conservation practices.

 9. This factor had an Eigen value of 1.0 and accounted for 8.10% of the variance.

10. The executive-parties dimension factor scale measures representation with the following variables: effective number of parliamentary parties; minimal winning one-party cabinets; executive dominance; electoral disproportionality; and interest group pluralism/corporatism. The unitary-federal dimension factor scale uses federalism-decentralisation; bicameralism; constitutional rigidity; judicial review; and central bank independence. High positive scores on each factor scale indicate consensus-like democracies. High negative scores indicate majoritarianism.

11. Lairson and Skidmore (1997) identify HDI as a better measure for capturing some of the social dimensions of a nation's socio-economic development that are neglected using income measures alone. Per capita income measures have disadvantages as a measure of development or welfare because they reflect only current income and not the amount of wealth accumulated over previous years; they provide no information about equity; and they do not reflect the value of goods and services not exchanged for money in a legal economy. The ecological modernisation debate focuses specifically on the relationship between development (not just income) and environmental quality. HDI is an alternative welfare indicator to account for not just per capita income, but also other important dimensions of welfare. It embraces Nussbaum and Sen's ‘capability approach, where human development is based on enhancing and widening people's choices is seen as the primary objective of social and economic change’ (1993; see also Daniels 2004). Differences related simply to measurement of the variables were accounted for with computations that included PPP and GNP as the measure of economic development; there were no statistically significant differences in any of the results.

12. Population density was measured as number of inhabitants per square kilometre as of 1996.

13. Data for this variable were taken from the Political Data Handbook OECD Countries (Lane et al., 1998. This variable was an index of electoral strength based on percentage of the vote in national elections between 1990 and 1994.

14. I use data that align most closely with the Lijphart indices, therefore predating the 1996 change in voting laws in New Zealand.

15. Robust regression is the preferred method when a research design is cross-national in focus because it reduces problems caused by heteroskedacicity.

16. There were also no significant relationships when the dependent variable was measured solely as percentage of nuclear energy.