Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
sures may be closer to the true causal effect. A problem with this approach is that most travel sur- vey data sets do not collect attitudinal data. â A second method is to use a two- stage instru- mental variable approach in which the endogenous explanatory built environment attributes are first regressed on instruments that are related to built environment attributes but have little correlation with the randomness of the primary travel behav- ior of interest. The instrumental variable method is not applicable to a nonlinear structure, however. Also, ignoring the sampling variance in the pre- dicted values of built environment attributes can lead to incorrect conclusions. â A third approach is to examine travel patterns of households immediately before and after a household relocation. This approach assumes that households move because of factors unrelated to their built environment attribute preferences. One potential problem with this approach is that relo- cating households are themselves a self- selected group. ⢠A proposed modeling framework was developed to address these concerns. The general methodology con- trols for residential sorting due to observed and unob- served attributes. It considers the direct and indirect effects of individual and household attributes on travel decisions as well as recognizes unobserved taste varia- tion. It focuses on automobile ownership as a travel deci- sion because automobile ownership impacts almost all aspects of daily activity- travel patterns. The framework does not consider attitudinal variables and it uses traffic analysis zones as surrogates for neighborhoods. ⢠There are a number of reasons for studying auto- mobile ownership related to the built environment. Automobile ownership is an intervening variable in the effect of the built environment on travel decisions. There is less research on the effect of built environment charac- teristics on automobile ownership. Automobile owner- ship impacts almost all aspects of daily activity- travel patterns. ⢠The joint residential choice and automobile owner- ship model was developed and tested on an empirical analysis of residential choice and automobile ownership decisions in the San Francisco Bay area. The analysis indicates that the built environment attributes affect res- idential location decisions, as well as automobile owner- ship decisions. There are random variations in sensitivity to built environment attributes, however. Household demographics appear to have a more dominant effect on automobile ownership than built environment factors, although both are important. Use of the population or density measures or both as proxy variables for built environment measures, such as street block density and transit accessibility, appears appropriate. There is varia- tion in sensitivity to build environment attributes due to both demographic and unobserved factors in both resi- dential choice and automobile ownership decisions. INNOVATIVE METHODS FOR PRICING STUDIES Arun Kuppam, Maren Outwater, and Rob Hranac Arun Kuppam discussed the innovative methods for examining pricing strategies used in the Washington State Comprehensive Toll Study. He provided an overview of the limitations of using traditional modeling techniques in pricing studies, the approach used in the study, and preliminary results. Volume 2 contains a paper on the topic.3 The following points were covered in his presentation. ⢠Current forecasting models have limitations for use with pricing studies. One concern relates to possible inaccurate traveler values of time by trip purpose, mode, and time period. The lack of temporal detail in time- of- day choice models is also a problem. There is also a need to model strategies to optimize tolls for pricing studies. ⢠This element of the Washington State Comprehen- sive Toll Study had five objectives. The first objective was to apply values of time for different market segments in a trip- based model. The second objective was to cap- ture variations in time of day by 30-minute time periods. The third objective was to develop an approach that is sensitive to pricing scenarios. The fourth objective was to capture travel behavior that reflects the tendency to shift to nearby time periods. The fifth objective was to develop a tool to optimize tolls by time periods. ⢠Market segmentation was used in the model. The four general categories were work trips by income group, nonwork trips by purpose, truck trips by class, and auto- mobile trips by mode. Four income groups were used: income less than $25,000, $25,000 to $45,000, $45,000 to $75,000, and more than $75,000. The nonwork trip purposes were home- based college, home- based school, home- based shop, home- based other, non- home- based work, and non- home- based other. The truck trips by class were light duty, medium duty, and heavy duty. The auto- mobile trips by mode were single- occupant vehicles, high- occupancy vehicles (HOV-2 and HOV-3+), and vanpool. ⢠The value of time by market segment was calculated for peak and off- peak periods. The equivalent minutes for a $3.00 toll were also estimated. For example, the value of time for drive- alone work trips by individuals in the over $75,000 income group was $37.04 and the equivalent minutes for a $3.00 toll were 4.9. For individuals in the 48 INNOVATIONS IN TRAVEL DEMAND MODELING, VOLUME 1 3 See Kuppam, A. R., M. L. Outwater, and R. C. Hranac. Innovative Methods for Pricing Studies. Volume 2, pp. 142â149.
