The classifications used by statistical agencies are sometimes updated. Hence, for the sake of comparability, it is necessary to estimate data from past periods according to the new classification. A frequently used method to calculate the estimates is through the use of Conversion Matrices. We present a theoretical analysis of this method and show with a practical example that it is possible to obtain useful estimates of the error.
Palabras clave / Key words: Change of Classification, BackcastingOn the error of backcast estimates using conversion matrices under a change of classification (Pdf 212 KB)