| Sumario: | This paper proposes an econometric approach to temporally disaggregate and timely estimate different panels of time series by extracting latent variables using the Partial Least Squares method. Specifically, the procedure is based on estimating common factors by maximizing the comovements of different frequencies of time series, which allows a temporally disaggregate and timely estimate the series of the interest. The empirical application is carried out for some series of the Ecological Accounts of Mexico, and we conclude that the approach generates accurate nowcasts, and, by providing longer time series with quarterly frequency, it allows the generation of public policy. Finally, by implementing a Monte Carlo experiment, we conclude that the procedure can be extended to several sets of cointegrated time series.
|
|---|
