Statistical Modelling 1 (2001), 143–159

Modelling Global Temperature Trends Using Cointegration and Smooth Transitions

David Harvey
Department of Economics, Loughborough University
Loughborough, Leics, LE11 3TU, UK

Terence Mills
Department of Economics, Loughborough University
Loughborough, Leics, LE11 3TU, UK
eMail: T.C.Mills@lboro.ac.uk

Abstract:

We analyse annual northern and southern hemisphere temperature data from 1856 to 1998 using models that contain either stochastic linear trends or deterministic nonlinear trends.  We find that changes in southern hemisphere temperatures lead to changes in northern hemisphere temperatures, with no evidence of feedback, in both types of model.  Where the models differ is that the assumption of stochastic linear trends leads to a common trend with constant growth throughout the sample period, whereas with deterministic nonlinear trends there are divergent trend growth paths in the two hemispheres after the early part of the 20th century.

Keywords:

global temperature; trends; cointegration; smooth transitions

Downloads:

Data sets and programs in zipped archive (See readme.txt for details.)
back