Green means stop: veto players and their impact on climate-change policy outputs

Abstract

Political institutions can hinder the adoption of certain policies. Veto-players theory suggests that more political institutions in a state will lead to lower rates of policy adoption. Extending this argument to climate-change policy, I contend that more political institutions will lead to lower overall climate-policy adoption rates, lower adoption rates for cost-concentrated climate-policy tools, and lower adoption rates for significant climate policies. Using new data on climate policies and legislative passage rates in 23 OECD states between 1996 and 2010, I find empirical evidence demonstrating that the number of political institutions intrinsic to a state negatively affects climate-policy adoption.

Keywords:

• climate-change policy
• policy adoption
• outputs
• veto players
• institutions
• OECD

Notes

^1. The research on VPs has seen a heavy emphasis on partisan VPs, especially in Tsebelis’ own work (1999, 2002, Tsebelis and Chang 2004). It focuses more on ideological aspects of the parties in government/opposition. While some of this research suggests different outcomes between partisan and institutional VPs, more recent scholarship (Duffield and Hankla 2011, Cao and Prakash 2012) adopts a more holistic approach by including both types of VPs in their research. This paper follows this approach, seeking to provide a broader overview of how political institutions act as barriers to policy change in a comprehensive manner.

^2. Though these policies typically focus on large emitters of GHGs, there are some that deal directly with small emitters. For example, the Regulatory Energy Tax of 1996 in the Netherlands focused a fossil-fuels tax primarily on households and medium-to-small-sized enterprises while also exempting green energy usage from this tax scheme.

^3. It is worth noting that this measure has been adopted and made universal across the entire EU with the Green Cars Initiative (http://ec.europa.eu/research/transport/road/green_cars/index_en.htm ).

^4. Sample states include Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Japan, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, Turkey, the UK, and the United States.

^5. Because of missing and/or uncertain data, the time period examined for Canada and Turkey is truncated to 2001 and 1999 respectively.

^6. The coding of policy tools in this study comes from Salamon (2002). I use concentrated cost and major policy percentages as dependent variables because they are mirror images of the distributed and minor policy terms. An effect for distributed or minor policy is the opposite of the displayed concentrated and major policy measures.

^7. Coding ‘major’ and ‘minor’ policies presents some challenges, but I code them in a manner to produce consistent outcomes. The Appendix provides more details on coding. The coding involves two questions: 1.) Is the policy passed a new policy/significant amendment to an existing policy? 2.) Did the policy have any media or any other visible coverage? An answer in the affirmative for both questions is a policy coded as major. A ‘yes’ answer was needed in both questions as these are necessary conditions that must be met to be a major policy. There is no middle category for policy satisfying only one of the two criteria as policy scholars since Lindblom have not identified any theoretical or empirical reasoning for a moderate policy category and differentiating between the two is essential for conceptual clarity.

^8. In terms of measurement choice, a count of institutional and partisan VPs accounts for both sides of the VP equation. While much previous work emphasises partisan VPs and ideology, I strive for a balanced analysis which necessitates the count measure. Using an ideological indicator, such as the environmental stance of parties in government and/or ideological distance, belies the institutional VPs where ideological differences are most easily observed. My measure provides the best opportunity for balanced analysis. Even though these are not my measures for VPs, I have included them as controls in the empirical analysis.

^9. Scholars like Tsebelis and Money (1997) recognise a strong relationship between federalism and bicameralism. Though this is generally correct, there are aberrant cases which do not conform to this observation: Canada, Austria, Belgium, and others. These institutions are correlated with a coefficient of 0.56, but not at a high enough level to be concerned about multicollinearity. Tests confirm that it is not present.

^10. Electoral formula is an important VP, as it deals directly with the notion that proportional systems are better at providing public goods – of which climate policy is – than majoritarian systems (Persson and Tabellini 1999). Essentially, majoritarian systems give parties the incentive to provide excludable goods to localities to win elections, whereas party competition in proportional systems is centred around public-goods provision. Consequently, parties can get a higher percentage of the national vote to win elections.

^11. Coding for federalism, electoral formula, and bicameralism range from: 0 (unitary state; proportional representation [PR]; unicameral legislature), 1 (medium federalism; mixed PR; and weak bicameral), and 2 (strong federal state; single member district; strong bicameral legislature). Coding for presidentialism, citizen initiative/referenda, judicial review, and pluralism is a 0 for absence of the institution/has no veto power. It is a1 if the institution is present and has formal or a high degree informal veto power vis à vis influence over VPs with formal veto power.

^12. For judicial review, each state’s constitution was reviewed to determine the jurisdiction of the highest court in each country. If they were able to overturn or modify legislation unequivocally, they were given a 1. For pluralism, I utilised Siaroff (1999) on corporatism. The cut-off of a score of 4 and above defined a state as corporatist and all below 4 are coded pluralist. Countries not included in the Siaroff piece were coded as pluralistic/corporatist based on a review of their constitution and secondary sources.

^13. The data for GDP per capita, percentage of workforce in industry and services, and consumption (population) come from the UN World Development Indicators database.

^14. Neumayer (2003) finds that more Greens improve environmental outcomes, meaning they may also positively impact climate-policy adoption as well. These data come from Duane Swank’s Comparative Parties data set. The data only go to 2006, but I was able to update the data through 2010. Data can be accessed through Swank’s website: www.marquette.edu/polisci/faculty_swank.shtml.

^15. Because of their slow-moving nature, the controls for Consumption and EU are included in the FEVD error term.

^16. For instance, the UK, Netherlands, Greece, and others vary once or twice in the 15-year span, having serious consequences for linear regressions, as the variables do not vary.

^17. The statistical software package used for this research is STATA 12. The Wooldridge test used the xtserial command. To test for a unit root process, I used the Fisher’s Augmented Dickey–Fuller test (xtfisher).

^18. From reviewing the policies in the data set, EU member states are extremely active in policy adoption. Many of the major policies adopted by these states are because of EU directives, but these states craft a wide variety of minor policies tailored to their state of origin that may not be adopted by other states via EU directive.

^19. Models run without the eta_bh term turn up weakly significant results for the VPs variable. This provides further evidence that the FEVD modelling technique is appropriate for these data. Email communication with A. Cooper Drury supports the judgement that the coefficient of ‘1’ for the eta_bh term is common and does not indicate a problem with the estimation.