Cooperation between members of political institutions, such as parliament and congress, is necessary for healthy democratic processes. Vote trading is one of the most discussed yet least understood types of cooperation. Empirical evidence on vote trading on a large scale is practically non-existent, limiting researchers' understanding of political processes. This project is developing crucial quantitative methods to measure and understand vote trading in democratic institutions. The methods can also be applied to study institutions composed of non-elected members such as courts, international organisations, and even boardrooms of private enterprises.
Explaining the science
Suppose that an elected representative supporting the farming sector and another one supporting the financial industry agree to vote in favour of two bills: one introducing farming subventions and another deregulating the stock market. This is a typical vote trading situation in which two legislators agree to vote for each other’s preferred bills despite the fact that, under normal circumstances, they would not support each other’s position.
The representatives involved in vote trading tend to keep these agreements in secret, in large part due to the ‘unusual’ nature of their votes. Such secrecy provides a safe space for cooperation, for example, in situations where extreme opposing ideologies complicate negotiations in public. Then, cooperation through vote trading can help overcome legislative gridlocks in highly polarised environments, a critical component to socioeconomic progress.
The secrecy around vote trading makes this political process extremely difficult to measure. For instance, in spite of qualitative evidence from interviews, there is no quantitative evidence of vote trading on a large scale. Furthermore, there are no datasets that identify traded votes (in machine learning parlance, there are no training data).
However, political scientists and economists have developed extensive theoretical work on vote trading for almost a century. This project combines this body of knowledge with network science in order to measure vote trading. The idea is to capture the level of reciprocity between ‘unusual’ voting behaviour through explicit network representations. Since traditional statistical and machine learning validation techniques are not viable, Monte Carlo simulations of political behaviour are employed in order to construct various statistical tests.
The project aims to improve our understanding of cooperation in democratic institutions. In particular, it will shed new light on cooperative behaviour that is intentionally ‘hidden’. To achieve this aim, the project develops state-of-the-art methods to analyse newly available large-scale data on voting behaviour from different institutional contexts. These methods allow for a general and flexible framework to be built for the empirical analysis of vote trading.
The framework of vote trading networks can be used to study voting behaviour in a wide range of democratic institutions. At the moment, the project focuses on identifying and quantifying vote trading in the US Congress, the US Supreme Court, the United Nations General Assembly, and the European Parliament.
As the project evolves, the analysis will be extended to other institutional settings where vote trading is an important determinant of democratic decisions.
This framework will also be used to study institutional design and to identify concrete measures that can be adopted if one wants, for example, to encourage cooperation in a polarised institution.