Discutants: Quentin Hoarau, Marc-Antoine Faure

Abstract : « Latent profile analysis (LPA) is carried out to identify different behaviour of firms in the European Union Emissions Trading System (EU ETS). The goal is to identify profiles based on firms’ transaction activity and not considering other characteristics, thus without any a priori bias on the type of behaviour a firm may have. This is especially relevant in the current context of suspicion of speculation in the market. The aim is to contribute to the debate on the potential detrimental role played by certain actors. Firm transaction behaviour is summarised using centrality and popularity indicators stemming from network theory. Estimated indicators include degree, strength, betweenness and eigenvector centralities, harmonic closeness and PageRank. Based on these indexes, LPA is performed to investigate the existence of different firm profiles and identify them. Firm characteristics are then added as covariates to explore the determinants of belonging to a certain group. Visualisation of the transactions network along with the estimation of indicators reveal that the network is highly polarised around a handful of financial firms that play the role of intermediary for compliance firms. The LPA further leads to the identification of three relevant profiles, with notably a « popular » profile characterised by extreme centrality values. Firm regulatory status and sectoral activity seem to be playing a significant role in determining profile assignment. »