Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 17
17 Table 12. List of explanatory variables used in air taxi analysis. Commercial Air Taxi Air Taxi Air Taxi Light Automobile Variable Air Piston Prop Jet Constant Term (Mode-specific) Cost (Mode-specific) Total Travel Time Avg Time Between Departures (Commercial Air) Income (Mode-specific) log(Based GA Piston + Prop Aircraft) log(Based GA Jet Aircraft) Table 13. Explanatory variable calculations. Cost: AAA total driving cost per mile for intermediate car (2006) x mileage computed from Microsoft MapPoint software. Automobile Travel Time: Drive times from Microsoft MapPoint software (+ periodic rest stops for trips greater than 3 hours). Cost: Top decile DB1B FY2007 market-specific fare from nearest commercial airports + access/egress cost to airports from Census locations. Travel Time: Weighted-average travel time for market-specific non-stop/one-stop/two- Commercial Air stop services (from May 2007 OAG schedule) + access/egress times to airports + airport terminal wait times. Average Time Between Departures: Market-specific estimate based on number of daily service offers assuming 16-hour day. Cost: See Model Assumptions section. Travel Time: Based on average aircraft speeds + access/egress times to airports. Air Taxi Based Aircraft: The number of based aircraft was measured in logs, allowing the positive impact of this "attractiveness" proxy to lessen at the margin. Explanatory Variables passengers selecting automobile was allowed to be different and Model Estimation from the effect of a change in the cost of commercial air for those selecting that mode. Additional details about the assump- Travelers choose from among five existing modes: auto- tions and calculations employed are shown in Table 13. mobile, commercial air, air taxi piston, air taxi turboprop, Although cost and travel time are typically primary factors in and air taxi light jet.12 Each mode is characterized by a set of any mode choice analysis, it is important to recognize that other attributes that affect travelers' choices. Aside from the stan- more difficult-to-measure attributes such as travel conve- dard cost and time attributes applicable to each mode, a nience, reliability, and unforeseen congestion may also be variable representing average time between departures was significant factors in choosing among available modes. included for the commercial air mode to account for the These are accounted for indirectly in the model by includ- effects of service availability of this mode. In addition, the num- ing mode-specific constant terms that, in principle, pick up ber of based prop or jet aircraft (averaged between the depar- average unobserved effects of each individual mode.13 ture and arrival airports) was calculated to serve as proxies The model is estimated with a "multinomial logit" specifi- for airport amenities that may affect the attractiveness of cation, and attempts to ascribe coefficient values to the attri- the three air taxi modes. Finally, personal income was also butes that best fit the observed choices made in the datasets included as a user characteristic that may affect mode travel (ETMS, DB1B, and ATS) described above. Because there are choice. A small number of different specifications were con- literally hundreds of thousands of observations derived from sidered before settling on the list of explanatory variables these datasets, it was not feasible to incorporate all of them shown in Table 12. into the estimation process. Instead, a sample of observations As indicated in the table, the effect of the constant term, was drawn from each dataset in such a way as to ensure that cost, and income variables was allowed to vary by mode; for example, the effect of a change in the cost of driving for those 13 As with the income variable whose value for a given individual does not vary across modes, it is necessary to "normalize" on one of the modes as 12 The automobile mode is excluded from the choice set when the auto the baseline (statistically it does not matter which one). Then, the income alternative would involve driving more than 400 miles. or constant term effects are interpreted relative to the baseline mode.