Freight Generation Models

This paper conducts a comparative analysis of two alternative approaches to freight generation modeling: ordinary least square (OLS) and cross classification. OLS models were estimated to identify a functional relationship between the number of freight deliveries per day and a set of company attributes used as independent variables. Cross-classification techniques aim at identifying a classification structure that provides a good representation of the freight generation process. To that effect, multiple classification analysis was used to identify groups of independent variables explaining freight generation, which provided the basis for constructing cross-classification tables. In both cases freight generation is explained as a function of company attributes. The model estimation process used data obtained from commercial establishments located in Manhattan and Brooklyn, New York. More than 190 different variables were tested as predictors for the number of deliveries received or carried per day. Six linear regression models found to be statistically significant and conceptually valid are discussed. Establishment attributes such as industry segment, commodity type, facility type, total sales, and number of employees were found to be statistically significant. The OLS models indicated that industry segment and commodity type are strong predictors of freight generation. It was also noticed that many of these variables played a significant role when interacting with economic variables such as total sales or employment. Twelve different groups of independent variables predicting freight generation were found to be significant as part of the cross-classification models. Results indicate that as in the case of regression models, commodity type, industry segment, and employment are strong predictors for freight generation.