Past event
Department of Economics Brown Bag with Professor Massimo Franchi, Sapienza Universita di Roma Autoregressive modelling of persistent change on a sphere, with application to global near-surface air temperature
Abstract: This paper studies representation and inference for spherical autoregressive processes with one or more unit roots. The cointegrating structure of such processes is shown to be remarkably simple, being completely determined by the unit root multiplicities on harmonic subspaces. We develop pooled augmented Dickey–Fuller tests for distinguishing unit roots from trend-stationary alternatives. An application to global near-surface air temperature favors trend-stationarity, with weaker evidence against a unit root in the global mean. The estimated trends describe geographically heterogeneous warming, particularly pronounced at high northern latitudes. A sparse harmonic reconstruction preserves the main geographical features.