Ministries that We Want: Parties' Agendas and Signaling of Portfolio Salience

• In Event: Interbranch Relations in Latin America

Abstract

In this paper, we propose a new measure of portfolio salience to coalition members in multiparty governments based on a measurement strategy that takes into account the portfolio payoffs parties receive. This stream of literature has evolved from a scholarly divide regarding the criteria driving portfolio allocation and the parties’ rewards that stem from them: whether it is the dominance of proportionality criteria, or the advantage of the coalition formateur in getting a disproportional ministerial bonus.

Studies on portfolio allocation made progress by exploring either: the structural asymmetries among ministries and their centrality to executive politics; or, the relative importance or salience of portfolios to the parties. Previous studies have focused on structural features of ministries, such as policy jurisdictions, budgetary resources, and staffing in order to measure portfolio salience. We argue that this attempt is insufficient, as it reflects the absolute value of the portfolio itself without considering who will get this position.

Other studies have estimated parties’ rewards by exploring the relational value of portfolios. They have addressed the convergence of parties’ high priorities in policy formation and the potential policy influence of each ministry. Portfolio salience matters to the parties, as it shows us which coalition members receive payoffs that best enable them to achieve their political priorities (Evans, 2020). The more salient a ministry is for a particular party, the greater the chances for this party to achieve its policy goals, influence the governing agenda, exert patronage and feed its partisan networks. Some studies have tried to fill this gap by measuring portfolio salience through surveys with specialists and/or politicians, in an attempt to classify portfolios according to their importance through ratings or rankings (Druckman; Warwick, 2005; Laver; Hunt, 1992; Zucco; Batista; Power, 2019; Evans, 2020).

We argue that a disadvantage of using surveys for this purpose is the risk of idiosyncratic judgments and the lack of variation in time. Therefore, this is not a sufficiently reliable measure to address the issue, especially when the research makes qualitative inferences about the salience of payoffs covering a long period and disregards contextual factors and the evolution of voter and party preferences (Evans, 2020). It is noteworthy that salience changes over time, as a consequence of exogenous or endogenous factors.

In this paper, we offer a new measurement strategy to estimate parties' yearly preferences for ministerial portfolios. First, we employ text-as-data to measure the proportion of floor speech made by parties' members on each of the Brazilian portfolios, which we use as a crude measure of their importance to each party. Second, we combine this information with data on bills sponsorship and pork-barrel earmarks by policy area, which we relate to their respective cabinet portfolios. Finally, we estimate a latent dimension from this data that we interpret as a portfolio salience measure.

To better grasp the utility of our measure, we have built a novel dataset, based on portfolio allocation in the coalitional presidentialism in Brazil from 1995 to 2020, which allows us to analyze the effect of parties’ preferences for portfolios on coalition formation. Notwithstanding, we do not intend to generalize our findings. Rather, we have built an indicator that can be replicated in other contexts so that, later on, we can advance in the comparative analysis. Through a quantitative research approach, we hope to contribute to the coalition-building literature by incorporating criteria not analyzed in previous studies and to provide a valid and reliable measure of salience that is sensitive to different parties.

Authors

• Magna Inácio, Federal University of Minas Gerais

• Rosiene Guerra de Andrade, Federal University of Minas Gerais

• Fernando Meireles, Federal University of Minas Gerais