Forecasting Generation of Freight Groups with Regression Models for Traffic Analysis Zones in Iran

In this research, linear regression models are developed for Iran’s inland freight production & attraction over a 20-year horizon until 2041. The dependent variables of these models are the total road and railway freight transported to/from 418 counties across the country in each of the 10 commodity groups. In these models, general population and employment variables are implemented together with binary variable of significant industrial and borderland counties to explain variations in the response variable. To calibrate these models, independent variables with a causal relationship to freight generation are chosen such that the highest coefficient significance level is achieved. Regarding the models’ goodness of fit, R-square statistic of the calibrated models stands between 0.85 and 0.98 which is appropriate considering the limited variables employed. To predict independent variables over the study horizon, age profile of the base year is developed in a 25-year timeline starting from 2016, using time-varying birth rates and constant mortality and migration rates. Then assuming four unemployment scenarios, employment of each county is projected using the last predicted populations. According to the models’ estimation, the total freight produced/attracted is expected to reach 545/551 million tons in 2021 and 668/660 million tons in 2041 with a 12.5 percent unemployment rate. Furthermore, with unemployment rate rising to 25 percent, the total produced/attracted freight is expected to fall 8.6/2.2 percent in 2021 and 9.4/2.4 percent in 2041.