ÍNDICE DE ATIVIDADE ECONÔMICA: OS MODELOS DE FILTRO DE KALMAN E BOX-JENKINS COMPARADOS
This paper has as objective to build a composite economic activity index for the local economy, created as form of measuring the economic activity. We use the factor analysis technique to determinate the components and their weights. This local index is then compared to national ones. As a result, the index behaves nicely as a local coincident index of the economic activities. Two techniques are used in their forecast; the first was the Kalman Filter and the second one the Box-Jenkins model. The presence of outliers required that we use a new technique in order the coefficients of the Kalman Filter model to be stable. The two techniques are then compared using a statistic test developed by Diebold and Mariano (1995). As a final result, the two models' forecasting are the same.