The STAMP software for state space models

This paper reviews the use of stamp (structural time series analyser,modeler and predictor) for modeling time series data using state-space methods with unobserved components. stamp is a commercial,gui-based program that runs on windows,linux and macintosh computers as part of the larger oxmetrics system. stamp can estimate a wide-variety of both univariate and multivariate state-space models,provides a wide array of diagnostics,and has a batch mode capability. the use of stamp is illustrated for the nile river data which is analyzed throughout this issue,as well as by modeling a variety of oceanographic and climate related data sets. the analyses of the oceanographic and climate data illustrate the breadth of models available in stamp,and that state-space methods produce results that provide new insights into important scientic problems.