1 / 20

Freight Demand Modeling Using Econometric Models

Freight Demand Modeling Using Econometric Models. Sept 14, 2010. Econometric Models…. Use regression based approaches to estimate demand Typical statistical techniques used are - Ordinary Least Squares (OLS) - Panel Models

jasia
Download Presentation

Freight Demand Modeling Using Econometric Models

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Freight Demand Modeling Using Econometric Models Sept 14, 2010

  2. Econometric Models…. • Use regression based approaches to estimate demand • Typical statistical techniques used are - Ordinary Least Squares (OLS) - Panel Models - Others (Two Stages, etc) • Use historical data

  3. Are particularly useful when… • Long period of historical data exists • Socio-economic factors are important, for example at a major interstate with significant proportion of thru-traffic • No significant network changes are expected (e.g. competing route construction)

  4. Advantages of Econometric Models • Easy to develop and estimate • Simulation of economic variables, toll and other scenarios can be developed in a statistically rigorous way • Relatively inexpensive to update and recalibrate

  5. Disadvantages of Econometric Models • Theoretically relevant independent variables can be highly correlated (multi-collinearity) • Assumption that historical data is a good predictor of future trends (tenuous when structural change might be taking place)

  6. Typical Model Structure • ….generally consists of… Truck Transactions / AADT = f (tolls, diesel prices, economic factors (e.g. employment, industrial production, inventory accumulation, inventory sales ratios), seasonal dummies, special one-off events)

  7. Michigan / Canada Border Crossing An Example Application

  8. Michigan / Canada Border Crossing • Freight Corridor model of Ambassador and Blue Water Bridges

  9. Independent Demand Drivers • Diesel Price • Foreign Exchange Rate • US Industrial Production • US Inventory Sales Ratios • US Light Vehicle Sales • Seasonal Dummies • Blue Water Deck Replacement

  10. The Challenges… Correlation =0.9 High Correlation between Foreign Exchange Rate and Diesel Price Solution: Principal Component Analysis

  11. Principal Component Analysis • Used to address high correlation of independent variables (or multi-collinearity) • Technique captures “underlying” trend of the data (by computing a weighted average) • Derived components are uncorrelated to each other

  12. Why do we use Principal Component Techniques here? • Reduce multi-collinearity between Diesel Price and the exchange rate • Incorporate more information: Principal component is derived for Inventory Sales Ratio (wholesale, retail, manufacturing, total business)

  13. Model Results • Sample 1995Q3 – 2007 Q4

  14. Do the Models Forecast the Recession? • Models Backcast very well with MAPE of 2.3% for Inventory based models • For the naïve GDP model MAPE of 7.5%

  15. New York State Thru-way An Example Application

  16. The New York Thru-way Model • Models estimated for different sections of the New York State Thruway • In particular models focus on sections near Buffalo and Lake Eerie, providing connections across the border into Canada

  17. Panel Models • Panel Models are estimated for different sections of the New York State Thruway • Panel models are jointly estimated for a common set of coefficients • Fixed effect methods control for section specific (unobserved) characteristics

  18. Model Specification • Inventory Sales Ratio Principal Component (includes retail, manufacturing, wholesale, total) • Industrial Production • US Diesel Prices • Monthly dummies and other event dummies

  19. Model Backcasts

  20. Conclusions • Inventory based models work well and value compared naïve models based on a broad based macro-economic indicator • Principal component techniques can be successfully used to address multicollinarity issues • When well calibrated, econometric models can successfully capture even unprecendent declines in activity

More Related