Journal of Official Statistics, Vol.19, No.3, 2003. pp. 215235

Current Issue
Personal Reference Library (PRL)
Personal Page

Monthly Disaggregation of a Quarterly Time Series and Forecasts of Its Unobservable Monthly Values

The temporal disaggregation problem consists of deriving high frequency data from less frequent observations of a time series. This problem usually occurs when carrying out analysis of the economic situation. In this article, a direct solution is first proposed to disaggregate historical values of an aggregated time series in one step. A recurive approach is then used to estimate current disaggregated values of the series and a method is proposed to predict future isaggregated values. The procedures are derived from a statistical model that links the unobserved data with a preliminary estimated series and with the series of aggregated values. It is assumed that the preliminary series can be estimated from data on related variables. Some results already established in the literature are employed to derive a theoretical solution that produces the Minimum Mean Squared Error Linear Estimator of the unobserved series. Mixico's GDP is used as an illustrative example.

ARIMA models; combatibility testing; minimum mean squared error; preliminary series.

Copyright Statistics Sweden 1996-2018.  Open Access
ISSN 0282-423X
Created and Maintained by OKS Group