TL;DR
This paper presents a theoretical and empirical case for "political posturing," where incumbents prioritize divisive ideological issues over consensus-driven "common-value" policies to signal their alignment with voters. Surprisingly, the authors find that increased transparency in news coverage can actually exacerbate this behavior, leading to more polarized rhetoric rather than better governance.
What Problem Does This Paper Address?
We often lament that politicians spend a disproportionate amount of time arguing over "wedge issues" (like gun control or abortion) while neglecting broadly popular goals like infrastructure repair or economic stability. Standard political science models often suggest that electoral pressure should pull candidates toward the "median voter" (the center).
Elliott Ash, Massimo Morelli, and Richard Van Weelden tackle the gap between this theory and the reality of a polarized "culture war." They ask: Why do politicians choose to focus on divisive issues even when consensus policies might be more socially efficient? The authors hypothesize that when voters are uncertain about a politician’s true preferences, the politician has a "signaling" incentive to over-provide effort on divisive issues to prove they are ideologically "congruent" with their base or the majority.
Methods
The authors employ a two-pronged approach:
- Theoretical Modeling: They develop a two-period game-theoretic model where an incumbent allocates a fixed budget of "effort" between a common-value issue (where everyone agrees on the goal) and a divisive issue (where preferences differ). They use the D1 criterion to refine the Perfect Bayesian Equilibrium, analyzing how re-election motives and the observability of effort (transparency) distort a politician's choices.
- Natural Language Processing (NLP): To test the theory, the authors analyze the Congressional Record from 1973 to 2012. They represent "effort" by the amount of floor speech dedicated to specific topics.
- Econometric Identification:
- Senate Analysis: They exploit the staggered six-year election cycle of the U.S. Senate. Since only one-third of the Senate is up for election at a time, they can compare the speech of Senators facing an imminent election to those who are not, using legislator fixed effects to control for individual traits.
- House Analysis: They use an instrumental variable for transparency—the "congruence" between congressional districts and newspaper markets (developed by Snyder and Stromberg, 2010). This measures how much local media actually covers a specific representative.
Datasets
- The Congressional Record (1973–2012): Raw text of floor speeches, processed into a vocabulary of 3,000 high-information bigrams and trigrams.
- Snyder and Stromberg (2010) Media Data: A dataset mapping the overlap between newspaper circulation and U.S. House districts to proxy for political transparency.
- Jensen et al. (2012) Dataset: Used as the primary source for the segmented and processed text of the Congressional Record.
Key Insights and Ideas
- Divisiveness as a Signal: Divisive issues are "high-information" signals. If a politician works on a consensus issue, it doesn't tell the voters much about their ideology. Working on a divisive issue, however, clearly signals which "side" they are on, which is valuable for winning over a majority-type electorate.
- The Transparency Paradox: We usually assume transparency is a "sunlight" that improves behavior. However, the model shows that if voters can see effort (what a politician talks about) but not necessarily outcomes, transparency increases the "marginal return" on posturing. This can lead to a "pooling equilibrium" where all politicians focus on divisive issues, making it impossible for voters to actually distinguish between types.
- Pareto Domination: In some scenarios, the incentive to posture is so strong that politicians choose effort allocations that are strictly worse for both the politician and every single voter compared to a consensus-focused allocation, simply because the consensus-focused path would lead to electoral defeat.
Key Results
- Electoral Proximity: U.S. Senators significantly increase their use of divisive language as their election year approaches. Moving from the first two years of a term to the election year increases speech divisiveness by approximately 5.8%.
- Transparency Effects: In the House of Representatives, higher news transparency is positively correlated with more divisive speech. A 1% increase in the transparency index leads to a 0.08% to 0.10% increase in speech divisiveness.
- Qualitative Validation: The "most divisive" phrases identified by their $\chi^2$ metric align with intuition: Republicans use "partial birth abortion" and "capital gain tax," while Democrats use "woman’s right to choose" and "tax breaks (for the) wealthy." "Least divisive" phrases include "federal highway administration" and "law enforcement community."
Limitations and Future Work
- Speech vs. Action: The authors use floor speech as a proxy for "effort." While they argue speech correlates with legislative focus, critics might argue that "talk is cheap" and doesn't necessarily reflect how a politician spends their behind-the-scenes time or staff resources.
- Issue Importance: The paper ranks issues by divisiveness but not by objective importance. Future work could integrate public opinion data to see if politicians specifically ignore "important but consensus" issues in favor of "unimportant but divisive" ones.
- Institutional Design: The findings suggest that delegating "common-value" tasks (like infrastructure or technical regulation) to politically insulated bureaucracies or "city managers" might be more efficient than leaving them to electorally accountable officials. This remains a fertile ground for further institutional research.
Summary updated