Evaluating Presidential Support in the Brazilian House of Representatives Through a Network-Based Approach

TL;DR

This paper compares traditional and network-based methods to evaluate presidential support in the Brazilian House of Representatives, demonstrating that both approaches yield valid and consistent results over a period from 1998 to 2019.

Contribution

It introduces a novel network-based approach for assessing legislative support at the party level and compares it with traditional roll-call voting analysis.

Findings

Both methods produce consistent results across the studied period.

Network-based approach effectively captures party-level support dynamics.

Results align with previous studies using different methodologies.

Abstract

Conflicts between the executive and legislative powers are a common, and even expected, characteristic of presidential systems, with some governments being more successful in the activity of obtaining support from the Congress than others. In the case of Brazil, specifically, this factor is considered crucial in the so called "coalition governments", with direct positive or negative consequences for the president, in terms of government performance during his (her) term. In this work, we investigate this problem by testing and comparing two different methods for evaluating the government support in the Brazilian House of Representatives. The first method is a more traditional one, being based on roll-call voting data, and measures the presidential support at the legislators level. The second method uses a network-based approach, and performs the same type of analysis but at the parties level. The obtained results, when applying both methods on legislative data comprising the period from 1998 until 2019, indicate that both methods are valid, with common features being found not only between the results provided by the two of them, but also when comparing their results with those obtained by previous and relevant studies in this field, by using the same type of data but different methodologies.