Econometrics and inconsistencies in the national accounts

Economic data are typically inconsistent with national accounting identities, contain measurement errors and are sometimes unavailable. A generalized conditioned least square procedure for the adjustment of data is proposed to deal with these problems. It is shown how the resulting data may be used for FIML estimation of the parameters of dynamic models (and the data themselves) with the aid of the Kalman-Bucy filter. An illustrative application of the proposed techniques to UK data is reported.