Forecasting of Migration Matrices in Business Demography

This paper demonstrates that the forecast of migration matrices can be conducted by means of updating procedures, well-known in the I-O theory. The authors use some of the most popular I-O updating procedures (RAS and some non-biproportional approaches) and calculate measures of the ex-post error of predictions. While taking into account the measures of distance between two matrices, a ranking of forecasting methods of migration matrices (forecast horizon one) is established. Finally, the advantages and drawbacks of particular forecasting methods with respect to one-step ex-post forecasts of migration matrices are discussed.