Dividend Yield as a predictor for stock pricing in the electricity sector: application of ARIMA and VAR models

Resumo

Companies in the Brazilian electricity sector, unlike other sectors, have their own characteristics that place them in a different group from conventional economic theory. One of the main financial decisions taken by companies refers to the definition of a dividend distribution policy. It is up to the company to decide on net income: retain it, to reinvest in its own activity; or distribute it to its shareholders. The objective of this work is 1) to verify if the use of ARIMA and VAR models can predict the prices of shares in the electricity sector from the payment of dividends and 2) if dividends have a causal relationship on the price of assets in the electricity sector. Results and conclusions: Dividend-based VAR and ARIMA models were not able to predict asset prices. The Granger test showed a causal relationship between the dividend yield on the variation of asset prices in ELET3 and CMIG4, but not in TRPL4. One of the reasons why the Granger test had different results between TRPL4 and the others may be related to the former acting only in the transmission sector, unlike ELET3 and CMIG4, which operate in different segments of generation, transmission and distribution. In addition, ELET3 and CMIG4 are companies with strong state influence, which would explain the need to pay dividends as a way of managing agency conflicts, signaling and even protecting minority shareholders.