Quantification of spatial effects on freight generation and freight trips

Abstract

FG/FTG models for Bangladesh were estimated by Ordinary Least Squares (OLS) using multiple explanatory variables including spatial variables. Four different functional forms were used for estimation (linear, logarithmic, power, and exponential). The models were estimated for six industry sectors (Agriculture, Manufacturing, Retail, Transportation, Wholesale and Food) as well as 32 industry sectors at 2-digit Bangladesh Standard Classification System (2-dig BSIC). The data used were collected in Bangladesh from a survey conducted for establishments on a national level. The spatial variables used included the travel time to Dhaka (Bangladesh’s capital and largest city) and to Chittagong (the country’s largest port and second largest city). Another added time related variable was the time to the closest port to capture the influence of ports on the freight activity of establishments. Using the difference in sum of squared errors and the adjusted R2 values, the estimated models were then compared with previous FG/FTG models estimated for Bangladesh with only employment as an explanatory variable to assess the significance of the considered spatial variables in explaining freight activity in Bangladesh.

This thesis studies the effects of spatial attributes of establishments and other relevant attributes, e.g., employment, fleet size by type, and establishment area on the estimation of nationwide Freight Generation (FG) and Freight Trip Generation (FTG) models. Spatial attributes can help explaining the influence of ports, and major cities on freight activity of establishments across the nation. Although employment has proved to be highly significant in producing unbiased estimates for freight activity, some FG/FTG models that only consider employment have low explanatory power, exhibited by low R2 values. Hence, the need to include other attributes, to increase the explanatory power of the model. Among these attributes, spatial attributes have proved to be highly significant in explaining FG/FTG patterns.