A recent paper by Pitfield (1978) concluded from an analysis of freight flows, employing a range of tests, that there was no means of predicting whether a linear-programming solution or a gravity model provided the better approximation to an actual flow matrix. However, it can be shown theoretically that the gravity model should always be at least as good as the linear-programming solution. The source of Pitfield's inconsistent results is traced and it is argued that plural tests should be used to identify specification errors rather than to promote agnosticism.
Get full access to this article
View all access options for this article.
References
1.
ChisholmMO'SullivanP, 1973Freight Flows and Spatial Aspects of the British Economy (Cambridge University Press, London)
2.
EvansAW, 1971“The calibration of trip distribution models with exponential or similar cost functions”Transportation Research51538
3.
GordonIR, 1974“The gravity hypothesis and transportation cost minimisation: A comment”Regional and Urban Economics419
4.
HartwickJM, 1972“The gravity hypothesis and transportation cost minimisation”Regional and Urban Economics2309326
5.
HeymanR, 1971“Initial attempts at modelling road freight flows” Note 211, Mathematical Advisory Unit, Department of the Environment, London
6.
MeraK, 1971“An evaluation of gravity and linear programming models for predicting inter-regional commodity flows” in Techniques of Transport Planning7 Ed. MeyerJR (Brookings Institution, Washington, DC) pp 297308
7.
O'SullivanPRalstonB, 1974“Forecasting intercity commodity transport in the U.S.A.”Regional Studies8191195
8.
PitfieldDE, 1978“Freight distribution model predictions compared: A test of hypotheses”Environment and Planning A10813836
