Files
Abstract
This piece of work was carried out objectifying to adjust the time series models to the series of
price indexes for the consumer ( PIC ) of Lavras MG. It aimed at verifying the occurrence of differences in
trend as for the period which preceeded, as well as for the one which succeeded the Plano Real, and if there
was a simple linear trend along with Plano Real ( 1994-1999 ) as well as if any seasonality occurred. It also
objectified to comparing the series of both Lavras and the Federal District. It was verified that there was no
trend along the whole series, nor there was any seasonality, notwithstanding the fact that a level change
occurred, which was characterized as an intervention, rather than a trend. It was necessary to fit a model
featuring an intervention for a better model adjustment and forecast. The parameters of the model which
featured interventions were estimated and those coincided with the periods of governmental interventios in
the Brazilian economy. The adjustment of a AR (1) model with intervention, proved to be more efficient
than the one which would not consider the intervention in the forecast of PIC from May to December, 1999.
To Brasília, the model fitted was an AR (2) with a permanent gradual intervention on the time 168 and a
permanent sudden intervention on time 266 which correspond to two months, respectively the one following
the introduction of Plano Cruzado and the other one, to the exact time when Plano Real was implanted. As
for the period from August 1992 to December 1998, the series of both Lavras and Brasília are considered as
being equal.