Measuring the Public Antecedents of Institutional Change: Quantifying Issue Salience and Tone of Public Discourse through Content Analysis

The institutional change literature has struggled with measuring the antecedents of change, i.e. the preferences/interests that are generally understood to lead policymakers, corporations, and other economic actors to change their behavior and consequently their institutional environment. Consensus has formed around institutions as shaping and constraining preferences, which in turn shape institutions. This inescapably cyclical logic has triggered the search for an extraneous variable that can explain changes in preferences and ultimately institutions. This lynchpin may have been found in the role of ideas, as proposed by constructivist approaches (e.g. Blyth 2002; Hay 2004; Schmidt 2010). Here, preferences are linked to ideas, the former being derived from the latter. Ideas can be thought of as basic conceptions of how, for instance, the economy or the political system, work and how they should work. Preferences are formed based on these ideas and the goals the firm wants to realize. Ideas are less internalized than preferences, and formed much more through public discourse, making them capture and thus a useful target for study as public antecedents of institutional change.

This paper contributes to this endeavor by measuring and quantifying ideas in terms of issue salience and tone of public discourse in three broad fields: finance, corporate governance, and labor relations. Issue salience enables to track when issues and associated ideas rise and fall in the public conscience, for instance when audit rotation, executive pay, or minority shareholder rights become salient issues. The tone of public discourse related to those issues reveals how positive or negative an issue is portrayed. Taken together, these two measures could illustrate that executive pay turned from a neutral, marginal topic of public debate into a highly salient topic with negative coverage. From this, we can infer that the idea that “excessive executive pay is acceptable” has been tarnished and is being challenged and potentially supplanted by the idea that “excessive executive pay is inappropriate.”

The method used to measure issue salience follows the approach of Culpepper (2011), Jones and Baumgartner (2005) and Smith (2000), which measures the number of articles found in national newspapers for a given set of keywords. Relevant keywords are defined through an inductive, iterative process for 15 issue areas across the three institutional spheres mentioned earlier. The newspaper search is conducted across four countries (US, UK, Germany, Switzerland), using one center-left and one center-right national quality newspaper in each country, in yearly intervals between 1995 and 2013. The tone of discourse is measured through content analysis following the approach of Tetlock (2007) and Tetlock et al (2008). Here, the full-text articles returned in the issue salience search are analyzed for the occurrence of ‘negative’ words using a dictionary created for this purpose by Loughran & McDonald (2011).