Distribution and interpolation revisited: a structural approach

Abstract

The paper revisits the traditional problem of distributing an annual aggregate over the seasonal periods and of interpolating an annual series when information on related series is available. Our approach relies upon the vector of seasons representation of a multivariate SUTSE model for the aggregate series and the indicators. It is argued that a more careful consideration is attributed to the time series properties of the variables involved, and important features, such as seasonal cointegration, can be brought into the interpolation and distribution process.