Understanding the behaviour of energy prices in Brazil Abdinardo Moreira Barreto de Oliveira 1  ORCID: 0000-0002-9377-6267 Anandadeep Mandal 2 1. Universidade Tecnológica Federal do Paraná, Departamento de Administração. Via do co- nhecimento, km 01, Pato Branco, Paraná, Brazil. ZIP: 85.503-390. abdinardom@utfpr.edu.br 2. University of Birmingham, Department of Finance, Birmingham Business School. Edgbas- ton, Birmingham, B15 2TY, United Kingdom. A.Mandal@bham.ac.uk Abstract. The objective of this paper is twofold: i) to identify the relationship, in a deterministic way, between the useful water volume in the Brazilian reser- voirs and the Price for Settlement of Differences (PSD) and; ii) to identify the parametric stochastic process which explains better the behaviour of the PSD series. Although there is a considerable amount of publications dealing with the commercialization of Brazilian electric energy through auctions, or dealing with models for predicting the increasing demand for electricity, little has been dis- cussed about the econometric or stochastic behaviour of these prices, even though it is admitted, in these same publications, that they occur. It was col- lected 882 weekly observations of the useful water volume (%) and the PSD (R$/MWh). Regarding the first question, it was found that the useful volume of water can explain between 54% and 89% of the variations in the log PSD. Re- garding the second question, it was found that a Vasicek (or Ornstein- Uhlenbeck) and a CKLS process, quite close to a Square Root Cox-Ingersoll- Ross (CIR-SR) process, better explain the stochastic behaviour of PSD. Lastly, futures studies are suggested at the end of this paper. Keywords: Energy prices, Markov Switching model, Stochastic processes. Proceedings ITISE 2018. Granada, 19-21 September, 2018. 146