Abstract
Using pooled OLS analysis, early literature found that incumbent and challenger campaign spending exhibited asymmetrical marginal productivities. That is, early scholars found that challenger campaign spending was more productive than incumbent campaign spending. More contemporary literature found this asymmetry was the result of endogeneity bias in the campaign spending covariates. Using 2014 U.S. House election data, we examine the nature of now-ubiqitous independent expenditures and whether these expenditures also exhibit an asymmetry in pooled OLS analysis. In addition to finding that independent expenditures can be differentiated from traditional campaign expenditures in their scope, dispersion, and magnitude, we find that the marginal impact of incumbent and challenger independent expenditures in pooled OLS analysis is symmetrical. We attribute this finding to the limited scope of independent expenditures. Moreover, using our estimated campaign spending effects, we find that political expenditures are rarely pivotal in determining election outcomes.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
As defined by the Federal Election Commission in 11 CFR 100.16(a), an independent expenditure is a “communication expressly advocating the election or defeat of a clearly identified candidate that is not made in cooperation, consultation, or concert with, or at the request or suggestion of, a candidate, a candidate’s authorized committee, or their agents, or a political party committee or its agents” (Coordinated Communications and Independent Expenditures 2015).
- 2.
In contrast to independent expenditures in 11 CFR 100.16(a), campaign expenditures are made by and/or coordinated with the campaigns themselves; they are filed with the Federal Election Commission and published in 2-year reports.
- 3.
United States v. Carolene Products Co. 304 U.S. 144 (1938) established the framework of different levels of judicial scrutiny. Cases involving a fundamental right or suspect class invoke strict scrutiny. Under such scrutiny, in order to be constitutional, the government must demonstrate a compelling government interest, and moreover, the means to pursue that interest must be narrowly tailored. Most economic legislation is subject to a less stringent rational basis examination, which holds that the government must simply demonstrate that the regulation is rationally related to a legitimate state interest.
- 4.
While these expenditures are technically independent, these independent expenditure groups are frequently managed by close advisors of the candidate and/or party they support.
- 5.
While theory implies both the incumbent and challenger campaign spending covariates should be biased, the following literature demonstrates that while the coefficient on incumbent campaign spending is consistently biased downward, the estimates for the coefficient on challenger campaign spending do not appear overestimated. This contention is substantiated by a comparison of the OLS and TSLS models’ estimates of the challenger campaign spending variable; the estimates for challenger campaign spending are typically not significantly attenuated in the TSLS model. This suggests there could be factors that prevent the theoretical expectation of upward bias from being an empirical reality. Erickson and Palfrey (2000) discuss a number of explanations for the challenger campaign spending covariate being insensitive to endogeneity bias; these include inexperienced challengers, challenger’s expectations being less accurate than incumbent’s expectations, and the fundamental asymmetry in the nature of the endogeneity bias for challengers and incumbents. For a thorough discussion of these explanations, see Erickson and Palfrey (2000, pp. 603–604).
- 6.
This measure of expectations was likely used due to data availability. Election forecasts in October, which occur near the end of the election cycle, are likely partially comprised of year-to-date fundraising numbers. If this is the case, the impact of the campaign expenditure variables is likely mis-estimated.
- 7.
This article also explored this possibility using different expectations data. Rather than use Congressional Quarterly’s October election forecast, this article used Stuart Rothenberg’s election forecasts from April, August, and October. Unfortunately, confirming the findings of Erickson and Palfrey (2000), this approach did not solve the endogeneity problem.
- 8.
Since U.S. House elections are not staggered like U.S. Senate elections, the previous race almost certainly includes the same incumbent and/or challenger; a criticism could be that specific personal characteristics of the candidates are correlated with the error term.
- 9.
Independent expenditure groups cannot contribute directly to campaigns; these groups are only allowed to spend monies independently of the candidates. One could argue that donors are making a choice between contributing to independent expenditure groups or campaigns; this is not supported anecdotally and is inconsistent with intuition, however. Donors likely use independent expenditure groups to influence election campaigns in excess of limits on contributions to candidates. The correlation between incumbent campaign spending and incumbent independent expenditures is r = 0.45. Similarly, the correlation between challenger campaign spending and challenger independent expenditures is r = 0.65. While these positive correlations aggregate over the individual-level story, they provide suggestive evidence that campaign contributions and contributions to independent expenditure groups are complements for major donors.
- 10.
During each election cycle, Stuart Rothenberg publishes election forecasts for congressional elections; these are well-known in the media and can be considered an adequate measure of expectations determining the level of spending observed in elections. For the archive of Rothenberg’s ratings, see http://www.rothenberggonzales.com/ratings/house/2014-house-ratings-october-3-2014
- 11.
Spending data were collected from the Federal Election Commission. The FEC has an independent expenditure database providing independent expenditures spent on behalf of all candidates for federal election. Moreover, the FEC has a database of campaign disbursements in each 2 year election cycle for all federal candidates.
- 12.
Data for U.S. House districts were collected from the 2013 American Community Survey sponsored by the U.S. Census.
- 13.
For additional information regarding the PVI, see http://cookpolitical.com/story/5604. Of course, a more direct measurement of district-level partisanship is desirable. Such a measurement would be more direct and would allow lagged presidential vote shares to be used as an instrumental variable. Unfortunately, such a measurement was unavailable for this project. Future research could use Multi-level Modeling with Post-Stratification to develop such a measurement.
- 14.
Data can be retrieved from http://www.census.gov/mycd/
- 15.
Statistical significance is determined using one-tailed probability values when there is a theoretical expectation about the sign of the coefficient.
- 16.
For clarification, it is important to note that the semi-log model only takes the natural logarithm of the campaign spending variables. The non-spending determinants of the vote remain in level form. Since the outcome variable is the same across models, coefficients for these covariates are comparable across models.
- 17.
As previously mentioned, the functional forms described above have notable advantages and drawbacks; accordingly, it was important to estimate the model with a variety of functional forms, and determine whether the conclusions drawn are model dependent. Fortunately, the estimation results suggest the conclusions are not dependent on functional form.
- 18.
As previously noted, a problem with the quadratic model is the possibility of having a negative marginal impact, or even a negative total impact, at high levels of spending. The figure provided above suggests the maximum effect on vote share occurs at approximately seven dollars of independent expenditures per person. There are a smattering of observations exceeding this threshold, unfortunately. Moreover, the total impact becomes negative at approximately 14 dollars of independent expenditures per person; there are no observations with this level of spending.
- 19.
The outcome of the AZ-1 being changed by political expenditures is the result of a statistical artifact. In addition to the natural logarithm of zero being undefined, the natural log of very small values is strongly negative; given the dearth of campaign spending in the AZ-1, this latter problems skews the findings; in particular, minimal challenger campaign spending yielded a 7.42 percentage point increase in incumbent vote shares; this is obviously an inaccurate depiction of reality.
- 20.
The NY-25 suffers from the same statistical artifact detailed in the previous footnote.
- 21.
This could be considered a counter-factual society in which campaign expenditures are outlawed.
- 22.
As previously noted, the estimate for the NY-25 is a statistical artifact.
References
Abramowitz, A. (1988, June). Explaining senate election outcomes. The American Political Science Review, 82(2), 385–403.
Bipartisan Campaign Reform Act of 2002 (“McCain-Feingold”), Section 203 (2002). Federal Election Commission. Accessed 23 July 2015. http://www.fec.gov/press/bkgnd/bcra_overview.shtml
Citizens United v. Federal Election Commission. Oyez: Chicago-Kent College of Law at Illinois Tech. Accessed 23 July 2015. https://www.oyez.org/cases/2008/08-205
Coordinated Communications and Independent Expenditures. Federal Election Commission. Accessed 23 July 2015. http://www.fec.gov/pages/brochures/indexp.shtml
Dionne, E. J., Jr. (2012, February 5). The Citizens United catastrophe. Washington Post.
Egan, T. (2014, October 23). The disgust election. New York Times.
Erickson, R., & Palfrey, T. (2000, September). Equilibria in campaign spending games: Theory and data. The American Political Science Review, 94(3), 595–609.
Gerber, A. (1998, June). Estimating the effect of campaign spending on senate election outcomes using instrumental variables. The American Political Science Review, 92(2), 401–411.
Green, D., & Krasno, J. (1988, November). Salvation of the spendthrift incumbent: Re-estimating the effects of campaign spending in house elections. American Journal of Political Science, 31(4), 894–907.
Jacobson, G. (1978, June). The effects of campaign spending in congressional elections. The American Political Science Review, 72(2), 469–491.
Op-Ed. (2014, October 13). A trickle down effect of Citizens United. New York Times.
Pelosi, N., & Sarbanes, J. (2014, February 4). Reversing the grievous error of Citizens United. Washington Post.
SpeechNOW.Org v. FEC. Federal Election Commission. Accessed 23 July 2015. http://www.fec.gov/law/litigation/speechnow.shtml
Super PACs. OpenSecrets.org. Center for Responsive Politics. Accessed 24 July 2015. https://www.opensecrets.org/pacs/superpacs.php
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendix
Rights and permissions
Copyright information
© 2016 Springer International Publishing Switzerland
About this chapter
Cite this chapter
Barutt, B., Schofield, N. (2016). Measuring Campaign Spending Effects in Post-Citizens United Congressional Elections. In: Gallego, M., Schofield, N. (eds) The Political Economy of Social Choices. Studies in Political Economy. Springer, Cham. https://doi.org/10.1007/978-3-319-40118-8_9
Download citation
DOI: https://doi.org/10.1007/978-3-319-40118-8_9
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-40116-4
Online ISBN: 978-3-319-40118-8
eBook Packages: Political Science and International StudiesPolitical Science and International Studies (R0)