ALTERNATIVE DEMAND MODELS AND THEIR ELASTICITY ESTIMATES.

THIS PAPER COMPARES SEVERAL COMMONLY-USED FREIGHT DEMAND MODELS USING CANADIAN INTER-REGIONAL FLOW DATA. THE MODELS EXAMINED ARE LINEAR, LOG-LINEAR, AGGREGATE MARKET SHARE LOGIT, AND TRANSLOG DEMAND SYSTEM. THE VARIABLE RETAINED IN THE FINAL MODEL AND THE ELASTICITY ESTIMATES ARE VERY SENSITIVE TO THE FUNCTIONAL FORM BEING USED FOR ESTIMATION. OF THE FIVE ALTERNATIVE MODELS (INCLUDING THE BOX-COX MODEL) EXAMINED, THE TRANSLOG DEMAND SYSTEM CLEARLY PERFORMS THE BEST IN ALL RESPECTS. THE BOX-COX MODEL PLACES SECOND IN THIS OVERALL RANKING WHILE THE LOG-LINEAR MODEL IS A CLOSE THIRD. THE LINEAR LOGIT AND LINEAR REGRESSION MODELS PERFORM POORLY IN THIS APPLICATION. (Author/publisher).