In Brazil, currently, one of the main efforts made by the Sistema Brasileiro de Defesa da
Concorrência (SBDC) is to fight against collusive associations between firms in the brazilian
fuels market, such as which incur in this type of association both in retail (gás stations) and not
wholesale (distribution sector). In the state of Pará, in particular, there are few studies dedicated
to analyzing the municipalities that this research will endeavor to study. Added to this is the
fact that the Conselho Administrativo de Defesa Econômica (CADE) increasingly needs studies
such as this one to support investigations into complaints of price parallelism between firms.
Thus, this study used screen statistical methods using the ARCH (Autoregressive Conditional
Heteroscedasticity), GARCH (Generalized Autoregressive Conditional Heteroscedasticity),
EGARCH (Exponential Generalized Autoregressive Conditional Heteroscedastic) and
TGARCH (Generalized Authoregressive Conditional Heteroscedas)) volatility models
(Exponential Generalized Autoregressive Conditional Heteroscedastic) and TGARCH
(Threshold Generalized Autoregressive Conditional Heteroscedasticity) on the series of
gasoline C prices in the municipalities of Belém/PA and Santarém/PA, using weekly data
provided by the official website of the National Agency of Petroleum, Natural Gas and Biofuels
(ANP). The series covers the period 2004-2020. In order to apply the models, two periods were
selected in which there was supposed to have been a cartel. These periods were determined
based on the application of resale margin and price variation filters, following the selected
methodology. The models did not show signs of a cartel, according to the equation for the
average, as there is no price increase during the hypothetical cartel period in the two
municipalities. Furthermore, it was verified in the variance equation of the Belém ARCH model
that there was a reduction in price variation, configuring an indication of a cartel. There was no
evidence of the presence of asymmetric shocks in the Belém series. In the Santarém models,
there was no evidence of a cartel by the equation of variance of the models. The TGARCH
model for Santarém showed evidence of the presence of asymmetric shocks in the
municipality's series.
|
