Determinants of Voting Behavior on the Keystone XL Pipeline

Department of Economics Working Paper Series Determinants of Voting Behavior on the Keystone XL Pipeline Joshua Hall and Chris Shultz Working Paper No. 15-35 This paper can be found at the College of Business and Economics Working Paper Series homepage: http://be.wvu.edu/phd_economics/working-papers.htm

Determinants of Voting Behavior on the Keystone XL Pipeline Joshua C. Hall Associate Professor of Economics College of Business and Economics Faculty Research Associate, Regional Research Institute West Virginia University joshua.hall@mail.wvu.edu 1601 University Ave., PO Box 6025 Morgantown, WV 26506-6025 Chris Shultz Ph.D. Candidate Agricultural & Natural Resource Economics Division of Resource Management West Virginia University c.shultz@live.com 101 Research Way, P.O. Box 6108 Morgantown, WV 26506-6108 Abstract After lengthy debate, the Keystone XL Pipeline bill passed in January 2015. We use this event to better understand the determinants of Senator voting behavior. Specifically, this paper attempts to examine the relative impacts of political and economic influences. This is accomplished through the use of a binary logit regression model with legislator vote as the dependent variable. Results indicate that while legislators do appear to be representing their political constituency, the role of campaign funding plays an important role as well. The economic effect of such funding, controlling for other factors, is quantitatively small. Keywords: Keystone XL, Keystone Pipeline, voting behavior, campaign financing JEL Codes: D72, D73 Acknowledgements: This study was funded by the West Virginia University Experiment Station as well as a hatch fund through the Division of Resource Management. Hall would like to acknowledge general research support from the Center for Free Enterprise at West Virginia University. A portion of this research was conducted while Hall was a Big XII Fellow at Texas Tech University.

Determinants of Voting Behavior on the Keystone XL Pipeline 1 Introduction The Keystone XL Pipeline is a proposed project that would transport crude oil from a supply point in Alberta, Canada to various delivery points in Oklahoma and Texas. This serves as an extension of the existing Keystone pipeline that links the Alberta oil sands to refineries in Illinois and Oklahoma. While estimated project costs are high, with a range from $7 to $8 billion (Environmental Analysis of the Keystone XL Project, 2015), the benefits in terms of additional pipeline access are large. The various delivery points of the extension such as Nederland Texas and Cushing Oklahoma provide access to additional pipeline systems, including those reaching the Gulf Coast. In September of 2008, TransCanada applied to the U.S. State Department for a permit to cross the U.S.-Canada border with the Keystone XL project. The following six years have seen considerable regulatory and political wrangling due to environmental and economic concerns. President Obama opposed the Keystone bill and expressed his desire to give the State Department time to finish its review. Republicans, however, tried to force approval of the Keystone XL pipeline through legislation. In a vote on November 19, 2014, proponents of advancing the pipeline proposal fell one vote short with a tally of 59 to 41 with all 45 Republicans in support. On Monday, January 12, the U.S. Senate advanced a bill to approve the pipeline with a 63-32 tally to allow debate and the offering of amendments on the bill. The Keystone XL Pipeline bill was passed on January 29, 2015 with a spread of 62 to 36. A history of house action on the Keystone pipeline (updated to January 5, 2015) can be found on the House website (House Action on the Keystone XL Pipeline, 2015). Congressional Digest has also provided a timeline on the legislative background (Keystone XL Timeline, 2015; Legislative Background on the Keystone XL Pipeline, 2015.) 1

This legislative gridlock is interesting given that the majority of Americans favor the construction of the pipeline based on recent polls. For example, a CNN/ORC poll revealed that of 1,011 Americans surveyed, 57% are in favor. 28% of respondents are against the Pipeline, while 15% are unsure. This support is strongest in the South and Midwest, where 65% and 63% are in favor, respectively (Bradner, 2015). Given that the United States Senate was designed to be a more deliberative body than the U.S. House, it is not a surprise that the Senate was the final legislative domino to fall. Given populist views on the project, we are interested in better understanding which factors played a role in Senator voting on the final bill. Are legislative decisions on issues such as Keystone based primarily on costs/benefits, environmental safety and public opinion, or are other factors at play? This is the primary question which is empirically addressed in this article. 2 Model and Data A logit binary choice model is fitted to the Yea and Nay decisions of the legislators who voted in the January 29, 2015 election to pass the Keystone XL Pipeline Act. Using this as a dependent variable allows for the consideration of factors which influence legislative decision making on issues such as the Keystone XL pipeline. We then employ two types of independent variables used in our analysis. The first are related directly to the legislator in question. It has been shown that ideological differences can play a large role in the difference between observed behaviors even when voter constituencies are identical (Sobel and Lawson, 1995). Energy Funding is the variable of focus for this paper 2

and it is defined as the proportion of campaign funding from the energy sector to the total campaign funding for the legislators most recent cycle (Politicians & Elections, 2015). The role of campaign funding has been found to be statistically significant in explaining legislator votes (Yueng, 2008). ADA is a measure of democratic action taken by each legislator (Americans for Democratic Action, 2015). Same Party is a binary variable representing whether the legislator is of the same party as the incumbent president (1=yes, 0=no). This variable has been used in similar studies probably most notably that of Lee and Tkachyk (1987) in their analysis of voting on Farm Bill Legislation. Finally, O Roark and Wood (2010) demonstrate that members who majored in economics as undergraduates had a significant impact on voting behavior in regard to minimum wage legislation compared to other undergraduate majors. We therefore coded any member of the U.S. Senate with an undergraduate degree in economics as a one, and everyone else a zero. The second type of variable analyzed considers the characteristics of each Senator s constituency. Using the return from presidential elections as a proxy for the ideology of a given constituency is commonly practiced amongst researchers (Leogrande and Jeydel, 1997). Political climate is a percentage of a state s constituency which voted Republican in the 2012 presidential election (Federal Elections, 2012). Union membership is analyzed since union members are likely to be highly affected by the possibility of job creation as a result of the Keystone XL pipeline (Hirsch and Macpherson, 2015). Other demographics analyzed include the percentage of a state s population over 25 with a bachelor s degree or higher and per capita income and these are obtained from the BLS. Descriptive statistics for all non-binary variables can be found in Table 1. Since each State has two Senators, each Senator shares a constituent base. 3

3 Empirical Findings Regression results are shown in Table 2. Three separate specifications were attempted in order to examine the relevance of all possible explanatory variables. Consistently throughout each specification, Energy Funding and Political Climate remained significant with coefficients ranging from 1.09-1.41 and 0.47-0.70, respectively. The impact of union membership, education levels, per capita income, political party of the legislator and economic background all remained statistically insignificant for each regression. This implies that while politicians in the vote on the Keystone XL Pipeline are quite likely acting to represent their constituency, on the margin they are responding strongly to the influence of funding from the energy sector. Unlike O Roark and Wood (2010), we find no evidence that an undergraduate economics training is related in a statistically significant way, either positively or negatively, to voting on the Keystone pipeline. In terms of marginal effects, we calculated a one-percent change in the proportion of funding coming from energy sources on the likelihood of voting yes on the Keystone pipeline. The marginal effects of energy funding are 0.0394 in the first regression, 0.0393 in the second regression, and 0.0325 in the third regression. These results suggest that while statistically significant, the effects of a percent increase in the energy funding ratio on the likelihood of a Senator voting yes on the Keystone pipeline while all other variables are at their means is relatively small. 4 Conclusion The results of this analysis indicate that the proportion of campaign funding a Senator receives from energy interests likely played a significant role in the voting decisions of policy makers on 4

the Keystone XL Pipeline bill. This supports some of the previous literature such as that of Gilens and Page (2014) who found that economic elites and organized groups representing business interests have substantial impacts on government policy, while average citizens and interest groups have little or no influence. It also is consistent with the literature, such as Wright (1989) and Sobel and Lawson (1995), showing that different Senators have different constituencies beyond the median voter in the state. 5

References Americans for Democratic Action. (2015). http://www.adaction.org Bradner, E. (2015) Poll: Majority of Americans Back Keystone Pipeline. CNN. Environmental Analysis of the Keystone XL Project. (2015) Congressional Digest, 94(1):5-7. Federal Elections 2012: Election Results for the U.S. President, the U.S. Senate, and the U.S. House of Representatives. (2012) Federal Election Commission. Gilens, M. and Page, B. (2014) Testing Theories of American Politics: Elites, Interest Groups, and Average Citizens. Perspectives on Politics, 12(3):564-581. Hirsch, B. and Macpherson, D. (2015) Union Membership and Coverage Database from the CPS. Union Stats. Keystone XL Pipeline Overview. (2015) Congressional Digest, 94(1):3-4. Keystone XL Timeline. (2015) Congressional Digest, 94(1):2. Lee, D. and Tkachyk, S. (1987) An Empirical Analysis of Congressional Voting on Farm Bill Legislation. Cornell Agricultural Economics Staff Paper Series, 87:21 Legislative Background on the Keystone XL Pipeline. Congressional Digest, 94(1):8. Leogrande, W. and Jeydel, A. (1997) Using Presidential Election Returns to Measure Constituency Ideology. American Politics Research, 25(1):3-18. O Roark, J. and Wood, W. (2010) Determinants of Congressional Minimum Wage Support: The Role of Economic Education. Public Choice, 147:209-225. Politicians & Elections. (2015) Open Secrets. http://www.opensecrets.org Sobel, R. and Lawson, R. (1995) Intrastate Differences in Representative and Senator Behavior. Journal of Public Finance and Public Choice, 13(1):3-17. Wright, G. C. (1989). Policy Voting in the U.S. Senate: Who Is Represented? Legislative Studies Quarterly, 14(4): 465-86. Yueng, I. (2008) The Influence of Campaign Contributions on Legislative Voting: a Case Study of the 1996 Farm Bill. Thesis, Department of Economics of Amherst College. 6

Table 1 Descriptive Statistics for Non-Binary Variables Variable Mean Standard Minimum Maximum Deviation Energy Funding 4.43 3.98 0.17 24.089 Political Climate 50.39 9.93 27.84 72.55 ADA 46.60 41.39 0.00 100.00 Union 10.05 5.33 1.90 24.60 Education 27.88 4.83 18.30 39.40 Per Capita Income 27.52 3.89 20.62 37.89 N=98. 7

Table 2 Logit model of legislator voting behavior on the Keystone XL Pipeline (1) (2) (3) Intercept -18.027-31.711-28.939 (-1.349) (-1.645) (1.406) Energy Funding 1.091 * 1.410 * 1.225 * (1.842) (1.885) (1.678) Political Climate 0.474 ** 0.6684 ** 0.705 * (2.125) (2.183) (1.938) ADA -0.066 ** -0.052-0.064 (-2.320) (-1.615) (-1.514) Union -0.133-0.122-0.037 (-0.826) (-0.639) (-0.171) Education -0.002-0.121 0.0603 (-0.007) (-0.305) (0.139) Per Capita Income 0.004 0.357 0.010 (0.008) (0.631) (0.015) Same Party -4.117-3.744 (1.777) (-1.428) Economics 2.315 (0.975) Pseudo R-Squared 0.91 0.93 0.93 AIC 36.25 33.89 34.75 Note: Z-statistics are given in parentheses. ***, **, * indicate significance at the 1%, 5% and 10% levels, respectively. 8