Estimating Seemingly Unrelated Regression Models from Incomplete Cross-Section / Time-Series Data
MetadataShow full item record
Most of the theoretical contributions to models for handling combined cross-section/time-series data are based on the assumption that complete time-series of equal length exist for all observation units. This paper is concerned with the estimation of multi-equation models in situations where the observation units "rotate" over time. The data then become incomplete cross-section/time-series data. The situation with complete cross-section/time-series data emerges as a special case of this specification. Econometric work with the Norwegian Surveys of Consumer Expenditure, which are based on rotating panels, has been a main motivation for exploring these problems. The conclusions, however, have for wider applicability. A substantial part of the work reported in this paper was carried out during the author's visit s i to Institut National de la Statistique et des Etudes Economiques (INSEE), Paris, in the autumn 1980.