Abstract
By use of input–output tables for three different years, the accuracy of the biproportional adjustment of input–output coefficients through time is investigated by comparing estimated and observed matrices. Major and minor input–output coefficients are distinguished. The contribution to the overall deviation by each coefficient category is found to be different. Relative errors are smaller for the major coefficients. The results suggest that the biproportional-adjustment technique provides a reasonable estimate of the more significant coefficients of an input–output matrix.