Technical Appendix to Air Demand in a Dynamic Competitive Context with the Automobile (2019)

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.

60 CHAPTER 8. TRAVEL BEHAVIORS AND ATTITUDES TOWARD THE LONG-DISTANCE TRIP: RESULTS OF THE ACRP 2017 SURVEY 8(A) INTRODUCTION This chapter of the Technical Appendix presents some findings about how Americans make the decision between taking a long-distance trip by air and taking a long-distance trip by car. In fact, the choice of the air mode is relatively rare in American trip-making as a whole. Data collected by the FAA and the BTS reveals that today the average American citizen takes less than one round trip per year by plane. While the rate of air travel is increasing, it is increasing gradually. The number of air trips per capita today is approximately 15% higher than it was in 1995. This represents an incremental growth rate stronger than for auto travel of all kinds (long distance and otherwise), which has grown approximately 7% during the same period. This chapter examines the factors that help to explain the choice between using the car or plane for trips longer than 300 miles. The central role of trip distance was established in Chapter 1, which demonstrated that an increase in trip distance is associated with a decrease in travel by car. This chapter builds on that central relationship and examines the interrelationship between trip distance and the following:  Trip cost.  Trip purpose.  Trip party size.  Traveler income.  Perceived need for an auto at the destination end of the trip. This chapter reviews many other factors (e.g., age, gender) using the same analytical format (i.e., mode share by distance and the candidate factor), but the five factors listed above emerged as the most determinant factors in the ACRP analysis. The results of the 2017 ACRP survey suggest that Americans make decisions about longer-distance travel in a utilitarian manner—but that understanding, and interpretation of those decisions can benefit from the careful (and cautious,) integration with values, preferences, and attitudes toward those trips and toward the places in which they live. The research team constructed a basic structural equations model that integrates both “hard” and “soft” factors in the travel decision-making process. Analysis of the output of that model suggests that the decision between air and car is first and foremost a mental process which weighs the problem of the cost of air versus the problem of the unpleasantness of the car’s long travel times. The purpose of this research is to help develop understanding of the trade-offs between car and air in mode choice among different population groups. The research identifies key indicators of attitudes and behavior by different market segments. It should be noted that this section does not include an analysis of the service characteristics of different airports in the survey. How people respond to options— the way they react to fares, travel times and pre-boarding times—is very transferable across sociodemographic groups. The specific service characteristics, such as

61 average business or economy fares, the number and times of direct flights to the destinations, and average peak and non-peak security times, are not similarly transferable. HIGHLIGHTS Highlights of the findings from Chapter 8 include:  People value long-distance travel positively, and enjoy doing it − There is strong support for the idea that the air trip is “exciting” − There is somewhat weaker, but still positive, support for the idea that the long auto trip is “appealing.”  People are making more trips per capita than in the base year of 1995 and drive more miles per capita over the same period.  Our survey results suggest that the rate of growth of the auto in the long distant trip is higher than the rate of growth by air, resulting in an increase in overall auto mode share for the long-distance trip since the base year.  Millennials had a higher propensity to choose air for short trips in our study, and a lower propensity to choose air for the longest trips  The role of the auto in the very long-distance non-leisure trip needs to be examined further, as the auto seems to play an important role in work-oriented multi-destination tours which last several days − The need for the auto on such work-based trips helps to explain why the mode share of auto never approaches zero, even for very long-distance trips  Similarly, the need for the auto for leisure trips with multiple children also explains why air is not chosen for some trips, even for very long-distance trips.  The perceived need for a car at the destination is a powerful explanatory factor in the choice of mode.  All demographic subgroups disagree with the concept that they would rather share a car than own it, with Millennials being somewhat more open to the idea.  Younger people are more likely to agree that driving for more than a day is unpleasant, and that they would like to use a smart device while traveling. Younger respondents express more concern about personal safety, uncertainty, and costs than older respondents.  Women are more likely than men to dislike driving for more than one day, and those with more income are more likely to dislike it than are those with less income.  Over 60% of our survey respondents agreed with the premise that they would “prefer to drive to a larger airport than take a feeder flight from a closer airport.” Stated preference exercises revealed that travelers would add an extra hour of driving to save one-half hour of flying. STRUCTURE The chapter addresses influential factors that influence the decision between the car and the airplane for the longer-distance trip. The chapter is organized in two major sections. Section 8(b) reviews the findings about the nature of long-distance trips, including some brief conclusions

62 about how that behavior has changed between the base year of 1995 and the present year. The analyses presented in this chapter are based on the preliminary results of the ACRP 2017 survey, which have since been applied to a national framework of data. Section 8(c) presents the research team’s initial examination of the role of values, preferences, and attitudes in the selection of the long-distance travel mode. Several major programmatic themes are explored through the grouping of several survey questions of similar content. Each of the survey’s attitudinal questions is reviewed simultaneously for the role of age, gender, and income level to better understand the possible role of traditional demographic variables in the formation of attitudes toward the trip, modes, and residential location of the trip maker. 8(B) THE CHOICE BETWEEN THE CAR AND THE PLANE FOR THE LONG-DISTANCE TRIP (2017) THE COSTS OF THE LONG-DISTANCE TRIP The Price of the Air Trip The price of air travel has a powerful effect on mode choice for long-distance trips as does the number of hours of driving that would be necessary for an alternative trip by auto. “Air Trip Cost” was derived from the survey results as the sum of the reported round-trip airfare, reported per-person parking costs, and reported baggage fees. Importantly, the price per mile paid by the consumer varies substantially by the length of the trip for the four market areas covered in the survey. As trip length increases, the price per mile offered by the airlines becomes more of a bargain for the customer. However, the air trip of under 700-mile distance (one way) is starkly higher in cost per mile. Figure 8-1 shows how different the short air trip cost is from the others. FIGURE 8-1: AIR COSTS PER MILE, BY TRIP DISTANCE Airlines for America reports that the average airfare, after adjustment for inflation, fell from approximately $0.20 per mile in 1995 (the years of the ATS) to slightly less than $0.16 per mile $- $0.05 $0.10 $0.15 $0.20 $0.25 $0.30 $0.35 $0.40 $0.45 300-700 700-1100 1100-1500 1500-1900 1900-2300 Above 2300 To ta l p er m ile C os t o f A ir Tr ip One-way Trip Distance, in Miles

63 today (including baggage fees) 1 —but the data shown on Figure 8-1 includes the price of parking at the airport. For the four survey areas, the average cost per mile is approximately $0.26. Based on self-reported costs, survey respondents reported paying approximately $0.10 per mile for the longest trips, which reveals that airlines may have a competitive advantage for longer-distance trips. The “average” survey respondent paid $437 for his or her reference air trip flight, and approximately 10% of that cost was attributable to parking costs and baggage fees. TABLE 8-1: AIR TRIP COSTS TRIP PURPOSE TOTAL COST Primarily Business $451 Primarily Leisure $377 As might be expected, the costs associated with a business trip are approximately 20% higher than costs associated with a nonbusiness trip. As discussed and defined in subsequent sections, the total cost per person of a round-trip business trip is estimated at $451 while the total cost of the round-trip nonbusiness trip is $377, which is shown in Table 8-1. Over 85% of primarily business trips in the 2017 ACRP survey were fully or partially paid for by employers. Figure 8-2 illustrates the fully allocated costs of the airplane versus the car as a function of trip distance and party size. The overall flatness of the air cost curve over distance is noticeable. The “fully allocated” costs of the one-person auto trip are also noteworthy. 1 https://www.theatlantic.com/business/archive/2013/02/how-airline-ticket-prices-fell-50-in-30-years-and-why- nobody-noticed/273506/

64 FIGURE 8-2: COST MODEL FULLY ALLOCATED ROUND-TRIP, BY MODE, AND PARTY SIZE ($0.54 PER MILE) The Price of the Auto Trip This section and its analysis explore the implications of the way in which the traveler interprets auto costs, as opposed to what has been established as fully allocated costs. Specifically, the traveler has a much narrower view of the incremental costs of driving the private vehicle for long-distance trips. The combination of gas and tolls are often referenced in the popular literature as the “out-of-pocket costs.” However, as expressed in Figure 1-6 earlier in this report, such costs might represent approximately half of the actual mileage-based costs (Figure 8-2). The research team first calculated “car trip cost” for a single-party trip. The research team initially calculated auto travel costs using the data in Figure 1-6 from BTS to generate per-mile vehicle costs for all categories that vary by mileage—excluding the cost of purchasing the vehicle. To update the data, the research team used a fully allocated cost of $0.54 per mile, which is consistent with the auto costs for travel expenses allowed by the Internal Revenue Service (IRS). The IRS accepts a similar number for travel cost calculations for income tax purposes. To account for the cost of staying overnight on longer trips, the research team assumed a night with lodging for every 700 miles, based on the analysis of this survey data. Approximately 85% of two-day travelers are taking trips of fewer than 1,200 miles, which would suggest that 700 miles is a reasonable distance for the first night’s lodging. A brief review of chain motels suggested that $90 would cover one basic room, with taxes and other costs included. Both Figure 8-2 and Figure 8-3 reflect the assumption of no lodging for the first 700 miles, two nights of lodging for 1,500 miles, and three nights of lodging for anything over 1,500 miles. In Figure 8-2, the auto trip costs for a one-person trip party are divided by two for a two-person trip party, and $- $500 $1,000 $1,500 $2,000 $2,500 $3,000 $3,500 300-700 700-1100 1100-1500 1500-1900 1900-2300 Above 2300 R ou nd T rip C os t p er P er so n One-way Trip Distance, in Miles One Person In Car Two in Car Four in Car Air Trip Cost

65 by four for a four-person trip party, assuming that all trip members could utilize a single motel room. FIGURE 8-3: COST MODEL, NONFIXED ROUND-TRIP COSTS, BY DISTANCE CAR ($0.23 PER MILE) Variable costs include gas at $0.11 per mile, maintenance at $0.05 per mile, and tolls at $0.04 per mile. Concerning the value of depreciation, the American Automobile Association calculates2 that a 5,000 increase in mileage equates to $180 in increased depreciation, or approximately $0.04 per mile. The research team used a nonfixed cost of $0.23 cents per mile in the calculation for Figure 8-3 that utilizes the average party size of 2.3 persons, derived from the 2017 ACRP survey. Attitudes/Perceptions Toward the Real Cost of the Auto Trip Analysis of attitudinal questions about how the user perceives cost support the notion expressed in Figure 8-4, which suggests that it is only after a 1,500–1,900-mile distance range that the plane is perceived to be cheaper than the car trip. 2 http://exchange.aaa.com/wp-content/uploads/2015/04/Your-Driving-Costs-2015.pdf $- $100 $200 $300 $400 $500 $600 $700 $800 300-700 700-1100 1100-1500 1500-1900 1900-2300 above 2300 R ou nd T rip C os t p er P er so n One-way Trip Distance, in Miles Car Average Party Size Air Trip Cost

66 FIGURE 8-4: ATTITUDINAL BELIEF ABOUT WHEN THE AIR TRIP IS CHEAPER The trade-off in Figure 8-3 is largely based on the difference between the per-person cost of the airplane trip (which does not vary much by trip party size) and the per-person cost of the car trip, which does vary by trip party size. In addition, increase in distance has a larger (and more linear) effect on total costs for the car than for the plane. Figure 8-2 shows that for a person traveling alone the air ticket is cheaper than the fully allocated costs of the auto at any range included in this study—with even the four-person air trip competitive at over 1,000 miles of one-way distance. Reasons to Look at Fully Allocated Costs Policy implications can be inferred when comparing Figure 8-2 and Figure 8-3. If society could implement a cost allocation system that charged all costs on a per-mile basis, then the individual traveler would have a greater incentive to choose a more cost-effective mode. Despite difficulties associated with the implementation of such a system, analysts writing about autonomous vehicles often assume that auto ownership would be collective, and that fees would be solely on a per-mile basis. Advocates of more efficient transportation are also approaching this more incrementally with the concept that insurance would be priced by usage rather than in lump sum, for example. In addition, comparing Figure 8-2 and Figure 8-3 shows how sensitive the price-estimation process is to assumptions about what costs vary with incremental driving and what costs do not. While Figure 8-3 was created based on research into the true costs that are not fixed—and vary by mileage—extensive reference is made in the literature to “gas and tolls” as the proxy for “out- of-pocket” costs. Using the same base assumptions, such costs would look like $0.15 cents per

67 mile, instead of the $0.23 cents used this chart. The logical implications of the two figures are important. The cost advantage of air travel is obvious at distances over 1,500 miles, even when the traveler does not interpret the real costs of the automobile to be as high as the IRS estimate, reported here as $0.54 cents per mile. THE IMPORTANCE OF TRIP DISTANCE IN INTERACTION WITH OTHER FACTORS This section reviews major factors that interact with trip distance in the explanation of mode choice between air and auto. As discussed in Chapter 1, income level is a strong predictor of the propensity to choose auto for the long-distance trip. The Role of Income, by Trip Distance In this analysis, the sample was divided into three income groups, with Figure 8-5 showing the difference between the highest and lowest-income groups. The curves for both groups reflect the observation that longer distances are associated with lower auto mode share and higher air mode shares. The higher-income curve seems to flatten at the 1,900–2,300-mile distance category, at which point approximately 10% of travelers continue to choose to travel by car. For trips of the same distance, approximately 20% of the lower-income group chooses to travel by car. At the highest distance lengths, variables other than income and trip length are needed to explain why 10–15% of travelers remain with auto. The following sections address the question of who is still choosing the auto at these distances. FIGURE 8-5: AUTO SHARE BY TRIP DISTANCE, FOR TWO INCOME GROUPS Employment Status, by Trip Distance Conceptually, employment status is a proxy for income. In general, as shown in Figure 8-6, auto share declines as trip distance increases. For trips up until the 1,500–1,900 mile range, individuals without a job (either full or part time) take the auto more than others (including students and retired). For trips over 1,900 miles, the ratio of car trips for both groups flattens, 0% 10% 20% 30% 40% 50% 60% 70% 80% 300-700 700-1100 1100-1500 1500-1900 1900-2300 above 2300 Au to M od e Sh ar e One-way Distance, in Miles Under $75k Over $100k

68 with the curve for the employed group staying quite flat through the highest distance categories. Clearly, in the key distance range between 700 and 1,700 miles, lack of employment is a more powerful explanatory factor than income alone. FIGURE 8-6: INTERACTIVE EFFECT OF DISTANCE AND EMPLOYMENT STATUS ON AUTO MODE SHARE The Role of Age, by Trip Distance As with all demographic subgroups examined, travel by auto among Millennials decreases as trip distance increases, as shown in Figure 8-7. However, distance seems to play a less dramatic role for the younger group than for the older group since the decline in auto travel as distance increases is less substantial for Millennials. Thus, Millennials are less likely than the older group to choose the car for shorter trips but are more likely to choose the car for trips longer than 1,700 miles. The fact that Millennials are less likely to travel by car for shorter trips is consistent with much of the attitudinal data presented in this report, and in other recent CRP studies; the fact that Millennials have a higher propensity to choose the auto for the longest trip is worthy of further research in the next phase of this project. In our sample, 86% of Millennials are working for pay, while only 63% of the older groups are working for pay. Thus, the younger group is more likely to be on a business trip, whose modal implications are explored here. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 300-700 700-1100 1100-1500 1500-1900 1900-2300 Above 2300 Au to M od e Sh ar e One-way Distance, in Miles Not Working for Pay Working for Pay

69 FIGURE 8-7: EFFECT OF TRIP DISTANCE ON AUTO MODE SHARE, TWO AGE GROUPS The Role of Trip Purpose, by Trip Distance Compared to those traveling for leisure, those traveling on business have a far greater propensity to take the airplane for the short-distance trip, and a far lower propensity to take it for trips over 1,900 miles (Figure 8-8). Again, for trips up to approximately 1,700 miles, costs associated with the trips can likely explain the pattern. Shorter-distance airplane tickets cost more per mile (Figure 8-1) than taking the same trip by car—and someone other than the traveler is paying for this far-faster business mode. However, the auto seems to play a key role in many long business trips, including trips with multiple destinations. (See Figure 8-13 later in this chapter.) 0% 10% 20% 30% 40% 50% 60% 70% 80% 300‐700 700‐1100 1100‐1500 1500‐1900 1900‐2300 above 2300 Au to  M od e  Sh ar e One‐way Trip Distance, in Miles Older Groups Millennials

70 FIGURE 8-8: AUTO MODE SHARE, BY DISTANCE BY TRIP PURPOSE The Role of Travel Party Size, by Trip Distance But what about larger trip party units? In the 2017 ACRP survey, the average travel party size was 2.3. Figure 8-9 shows the auto mode share by distance for one or two persons in a travel party (orange, solid line) compared with three or more (blue, dashed line.) For those traveling alone or with one other person, the curve looks familiar, with the longest trips showing an approximately 12% share for the auto. Conversely, among those who traveled with two or more individuals, approximately 30% took the car for trips over 2,000 miles. 0% 10% 20% 30% 40% 50% 60% 70% 80% 300-700 700-1100 1100-1500 1500-1900 1900-2300 Above 2300 Au to M od e Sh ar e One-way Trip Distance, in Miles Primarily Business Primarily Leisure

71 FIGURE 8-9: AUTO MODES SHARE, BY TRIP DISTANCE AND PARTY SIZE The Interaction between Trip Purpose and Party Size Figure 8-10 shows that business trips are far more likely to be taken alone than nonbusiness trips. Fifty-six percent of business trips by air are taken alone, compared to only 31% of nonbusiness trips by air. Similarly, 32% of auto trips for business are taken alone, compared to only 14% of auto trips for nonbusiness purposes. When all party sizes are taken together, the averages are less dramatically different, but business trips with both modes are still associated with a smaller party size, as shown in Figure 8-10; for both trip purposes, party size for air is also smaller than the party size for auto. The party size for business trips by plane is approximately 25% lower than the party size for leisure trips by car. 0% 10% 20% 30% 40% 50% 60% 70% 80% 300-700 700-1100 1100-1500 1500-1900 1900-2300 above 2300 Au to M od e Sh ar e One-way Trip Distance, in Miles Three + One or Two

72 FIGURE 8-10: TRIP PARTY SIZE, BY TRIP PURPOSE AND MODE Need for the Auto at Trip End, by Trip Distance Those who report that they need the flexibility of a car at the destination of the trip have a significantly higher propensity to choose the car in the referenced trip, as shown in Figure 8-11. The minority of the sample reported that they do not need a car choose it as a mode, even for relatively short-distance trips. Among those who stated that they did not need a car, mode share for car continued to decrease as distance increased, up until a trip distance of approximately 1,700 miles. For trip distances over 1,700 miles, distance ceases to be explanatory, as the residual 10% of the group continued to choose the auto even with increasing trip distance. To the contrary, among those who stated that they needed a car, the relationship between trip distance and car mode choice behaves in a more linear manner, such that car mode share continues to decrease as distance increases. 0 0.5 1 1.5 2 2.5 3 Non-business Business Tr ip P ar ty S iz e Trip Purpose Car Plane

73 FIGURE 8-11: AUTO MODE SHARE BY DISTANCE—NEED CAR VS. DON'T NEED CAR Number of nights in total trip Figure 8-12 shows the number of nights spent on the round trip. The fiftieth percentile of trip duration fell at four nights: half the respondents stayed less than that and half stayed longer. But trips of longer duration were well represented—18% of the travelers stayed away one or two weeks, and another 5% stayed away for more than two weeks. The number of nights in the total trip was not found to be a predictive factor in the explanation of choice of long-distance mode. FIGURE 8-12: NUMBER OF NIGHTS AWAY 0% 10% 20% 30% 40% 50% 60% 70% 80% 300-700 700-1100 1100-1500 1500-1900 1900-2300 Above 2300 Au to M od e Sh ar e One-way Trip Distance Need Car Don't need car 2% 7% 18 % 21 % 15 % 10 % 6% 18 % 5% I t w a s a d a y t r i p o n l y 1 n i g h t 2 n i g h t s 3 n i g h t s 4 n i g h t s 5 n i g h t s 6 n i g h t s 7 - 1 4 n i g h t s 1 5 o r m o r e n i g h t s

74 WHO IS TRAVELING BY CAR OVER 1,700 MILES? In the examination of the explanatory variables, increasing trip distance is associated with decreasing share for the auto—until approximately 1,700 miles, where the auto mode share remains near 10%. As shown in Figure 8-3, the price advantage of the car disappears at this trip distance; however, a consistently significant portion of the sample remains with the slower, multi-night auto trip. Clearly, some travelers continue to choose or need the car for these long- distance trips. For very long-distance trips (over 1,700 miles), compared to air travelers, the following tends to be true for auto travelers:  More likely to be male.  More likely to be younger.  Have less income.  Have fewer years of education.  More likely on a business trip.  Have more people in the travel party.  Have more household members in the travel party.  Have more children in the travel party.  Report more need for a car at the destination.  More likely to be on a multi-destination trip. The last traveler behavior is important, as nearly 75% of respondents using the car for the trip over 1,700 miles were going to multiple destinations. By comparison, fewer than 25% of air travelers for this trip distance were going to more than one destination. WHO IS TAKING A MULTI-DESTINATION TRIP? Approximately 43% of survey respondents reported going to more than one destination on their reference trip. The reality of several “stops” on the trip has a profound impact on mode choice, such that 42% of our sample on a single-destination trip chose the car, but that percentage increases to 63% for those with multiple destinations (the average mode share for car was 51%). Figure 8-13 shows auto mode share by trip distance for travelers with single destinations and travelers with multiple destinations.

75 FIGURE 8-13: RELATIONSHIP BETWEEN NUMBER OF DESTINATIONS AND MODE CHOICE, BY DISTANCE Travelers with multiple destinations are more likely to be male, young, away for more days, and in a larger travel party—with more household members and more children. (It is not resolved here as to whether the decision to visit several relatives in one trip causes the choice of the car, or whether the choice of the car causes the traveler to visit several relatives.) In addition, those traveling to multiple destinations are more likely to be to be traveling on business, which suggests that there may be two separate markets for long-distance multiple destination trips by car. The propensity for a given trip to have multiple destinations is highest in the lower-distance trip length categories (300–700 miles); this connection flattens for the rest of the distance spectrum at approximately one-third of trips (not shown). 8(C) UNDERSTANDING THE CHOICE OF AIRPORT FOR THE LONG-DISTANCE TRIP FACTORS AFFECTING AIRPORT CHOICE The project has explored what is known, and what is not known, about travelers’ propensities to drive longer distances to begin their trips at more distant airports. Air travelers’ choices among alternative airports can involve trade-offs across multiple attributes that vary by airport, including ground access/egress times and costs, airport flight service frequency, types of flights and airfares available, and likely flight delays and the design/operation of the airports, among others. Any one attribute may motivate an air passenger to choose an airport that is less attractive than other alternative airports along all of those other dimensions. Furthermore, there are certainly many cases in which air travelers find that one airport dominates all others with respect to all of these attributes. But in most regions with multiple airports, air travelers likely find that different available airports are preferable to others for some—but not all—of the important attributes. The result is a choice that involves trade-offs among several attributes. 0% 10% 20% 30% 40% 50% 60% 70% 80% 300-700 700-1100 1100-1500 1500-1900 1900-2300 above 2300 Au to M od e Sh ar e One-way Trip Distance Mulitple destination Single destination

76 PREVIOUS EXPLORATIONS OF AIRPORT CHOICE Several past studies have analyzed the relationship between airports’ attributes and the share of air passengers that they attract in regions with multiple airports. One of the earliest published studies on this topic resulted in the development of the Multiple Airport Demand Allocation Model. This model was originally developed to analyze potential traffic at the New York region’s Stewart Airport, but it was subsequently applied to the metropolitan Washington region and several other multiple airport regions.3 The model assumed that the choice among alternative airports depended on three primary sets of factors: 1) access/egress times and costs; 2) time spent at airport; and 3) the schedule frequency. The model converted access costs into equivalent minutes and added this, along with access times, to the time spent at each airport based on expected terminal walking, waiting, and delay times. Finally, the model calculated expected wait times based on flight schedules to create a total time for each airport. The effect of this total airport time on air shares was calibrated based on observed airport shares. While this model reflects several of the key airport attributes that affect airport choice, it does not consider airfares and flight service differences (other than flight frequency), nor does it consider traveler type differences among airports. In addition, the relative weightings of the included attributes were asserted rather than statistically estimated. A later study of the San Francisco region used disaggregate data from airport surveys to statistically estimate parameters of discrete choice-based (multinomial logit) airport choice models for that region.4 That study confirmed the importance of ground access times and costs and of the frequency of air service (particularly direct service) on airport choices and provided statistical estimates of the relative importance of these factors. The study also demonstrated that business and nonbusiness travelers make distinct trade-offs among these attributes. This was the first of several such studies using San Francisco airport survey data, each making successive improvements to the model.5 While previous work could determine the effects of obvious factors such as ground access travel time and flight frequency, the use of observed airport choices—revealed preferences (RP)— alone imposes inherent limitations on the ability to determine trade-offs among attributes such as airfares and airport access travel time. These include the fact that available airfares and air travel options vary significantly for any given itinerary and cannot be accurately estimated from survey responses. In addition, in many markets such as Atlanta, a single airport dominates and there are few if any viable alternatives. As a result, much of the more recent work on airport choice has employed stated preference (SP) surveys to estimate trade-offs among air travel attributes.6 This 3Brian Campbell and Associates, Description of the Multiple Airport Demand Allocation Model, US DOT/FAA Office of Aviation Policy, September 1977. 4 Harvey, Greig, “Airport Choice in a Multiple Airport Region,” Transportation Research Part A, Volume 21, Issue 6, November 1987, Pages 439-449. 5 See, for example, Hess, S. & Polak J.W. (2005), Mixed Logit modelling of airport choice in multi-airport regions, Journal of Air Transport Management, 11(2), pp.59-68. 6 Hess, S., Adler, T. & Polak, J.W. (2007), Modelling airport and airline choice behaviour with the use of stated preference survey data, Transportation Research Part E, Volume 43, pp. 221-233.

77 approach, used also in this research project, can provide more robust estimates of the ways in which air travelers trade off attributes when making airport choices. IMPLICATIONS FROM THE ACRP 2017 SURVEY RESULTS The following sections of this chapter deliver early observations from the results of the ACRP 2017 survey, illustrating the ways travelers value each of the attributes associated with available airports. Attitudes Toward Choice of Airport More survey respondents agreed than disagreed with the statement, “Assuming that the door-to- door travel times were similar, I would prefer to drive to a larger airport to get more flight choices than take a feeder flight from a closer airport.” Sixty-two percent of survey respondents agreed with the statement, with almost no variation by the length of the referenced trip. The level of agreement with the statement was higher for Millennials than for older age groups, higher for the higher-income group than the lower group, and somewhat higher for the males than for the females. Indeed, of those who did not fly out of the closest airport, 88% were going to a larger airport, and only 12% were going to a smaller one. Those choosing the more distant airport tended to be younger than those departing from closer to home. Willingness to Pay for Attributes Some early observations about the choice of the departure airport can be made from the estimates of “willingness to pay” derived from the project’s models. Willingness to pay is the amount someone would be willing to pay to have more desirable levels of specific attributes. In these models, the willingness to pay is calculated as the amount of additional airfare travelers would absorb to improve their air travel. 1. On average, travelers are willing to pay an additional $17 on airfare to save one hour of access time but are willing to spend approximately double that amount to save one hour of flight time. This confirms results from many past studies that flight time is considered more onerous than ground access time. It implies, for example, that travelers would drive one extra hour to an airport if it saves them 30 minutes of flight time. 2. Travelers place a much higher value on additional flights in airports that currently have fewer flights, or in the reverse, are willing to pay much more for airports that have more flights. The effect is strongest for direct flights but also applies to connecting itineraries. However, this effect of a single additional flight diminishes when comparing airports that both have more flights. 3. On average, travelers are willing to pay $50–$60 to fly from an airport that has direct flights to their destination rather than connecting itineraries. Many travelers are also willing to pay much more to avoid connections. 4. There are also general preferences among travelers for specific airports due to their designs, services, and amenities. This research found that travelers are willing to pay $15–$120 to fly from a particular airport in each region compared to other alternative

78 airports. In all cases, these specific airports were the larger hub airports in the regions studied, reflecting the higher levels of amenities and service offered in those airports. These observations all suggest that there are compelling factors that can cause air travelers to choose a more distant airport. In addition, although airport flight schedules and available airfares are crucial factors in airport choice, other airport amenities and services can have a significant effect on airport choice. 8(D) LONG-DISTANCE TRAVEL BEHAVIOR IN 2017 COMPARED WITH 1995 COMPARISON WITH EARLIER DATA Early examination of the 2017 survey results reveals a considerable level of stability in American long-distance travel patterns when compared with the 1995 base-year data. The preliminary results suggest that Americans’ long-distance trip-making by car has increased demonstrably alongside the trip-making rate on domestic flights. An early analysis of the 2017 survey data shows a mode share of 51% for the automobile for trips exceeding 300 miles, which is somewhat higher than a directly comparable mode share of 46% from the 1995 American Travel Survey data. The issue of the growth in the role of auto will be further refined and examined in the final phase of this ACRP project, where the results of this most recent survey data will be integrated into the larger national model of long-distance travel. Important differences exist between the content of the 1995 ATS and the recent 2017 ACRP survey. First, the surveys were undertaken for different purposes. The 1995 ATS sought to understand all long-distance travel in the United States. This research project’s survey sought to understand the travel out from four market survey areas—areas chosen to improve the understanding of the choice between smaller airports and larger airports in major markets. In addition, the 1995 ATS was undertaken in all areas. In contract, this project’s survey was undertaken in four areas with major, successful airports with extensive feeder air services in many instances. The 1995 ATS results were reexamined for trips beginning in the US Census geographic districts, which include Boston, Washington, Chicago, and Denver. All the data in the present survey was collected within four US Census divisions: New England, South Atlantic, East North Central, and Mountain. Data from those four areas in the 1995 ATS survey is presented in the dotted line labeled “Four Divisions” in Figure 8-14, which also includes a line summarizing 1995 data from the entire nationwide sample. The 1995 ATS’s average mode share to auto is approximately the same as this project’s more limited sample and benefits from the stability of a larger sample size. In other words, utilizing 1995 ATS data results in curves that are remarkably similar and that support the kind of early comparison with the 2107 curve presented here. The research team restructured the 1995 ATS data to only reflect trips by air and auto to facilitate comparison with the data collected for this project, which only queried for those two modes. Thus, the 1995 “auto mode share” in all cases represents the ratio of auto travel to auto plus-air travel.

79 FIGURE 8-14: AUTO MODE SHARE IN 1995 AND 2017 SURVEYS, BY TRIP DISTANCE A significant increase in the mode share of auto between 1995 and 2017 is evident in Figure 8-14. Many attributes that can be expected to influence mode choice for long-distance trips have changed between 1995 and 2017, and some of those changes are briefly reviewed in this chapter, including changes in basic prices. With these factors reviewed, the basic logic of a significant increase in the auto for the long-distance trip can be reviewed. FIGURE 8-15: CHANGE IN AUTO USE PER CAPITA OVER TWO DECADES DOES AN INCREASE IN CAR USE MAKE SENSE? The research team has concluded that the national per-capita VMT is about 7% higher in 2017 than it was in 1995. Figure 8-15 suggests that VMT per capita today is about the same as it was in 1997/1998, which had risen rapidly since 1995. While VMT per capita did decline between 0% 10% 20% 30% 40% 50% 60% 70% 80% 300-700 700-1100 1100-1500 1500-1900 1900-2300 above 2300 Au to M od e Sh ar e One-way Trip distance, in Miles Auto Mode Share, by Distance New 2017 compared with Two 1995 Samples 2017 1995 All Four Divisions 0 2,000 4,000 6,000 8,000 10,000 12,000 VMT per Capita, 1995- 2017

80 2007 and 2013, it has largely rebounded in the past four years. The data in Figure 8-15 suggest that there is remarkably little change in our propensity to generate VMT over the past 20 years, at just under 10,000 miles per person. Changes in Gas Prices Even though the price of gas has increased significantly, Americans are driving at a rate about 7% higher than in 1995. Gas prices at the pump were higher (adjusting for inflation) at the time of ACRP 2017 survey than at the time of the 1995 ATS survey: we estimate that adjusted for inflation, gas at the time of the ACRP 2017 survey was about 34% higher than in 1995.7 This reinforces the well-documented observation that changes in gas prices cannot predict changes in car use well. In general, while the gas price rose by approximately one-third, the consumption rate rose by approximately 7%. FIGURE 8-16: COST OF AIRLINE TICKETS AND GAS OVER TIME Changes in Air Travel Fares The relationship between frequency of air travel and the price of air travel is more intuitive. Reviewing several sources8 of data produces multiple estimates of how much the cost of air has decreased on a per/mile basis, but the latest updates from the airline industry suggest that compared to the estimates for the 1995 base year, the cost has dropped by more than half. This general decrease in air costs correlates well with the rise in enplanements shown in Figure 1-7 of Chapter 1; the rate of enplanements per capita has risen during this time by about 18% 7 Source: FHWA 8 Several sources were reviewed. Those in Figure 16 came from http://www.planetickets.com/airfare.html. Gas prices over time in Figure 16 came from www.randomuseless.info/gasprice/gasprice.html 0 0.1 0.2 0.3 0.4 0.5 0.6 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 C os t p er m ile (d ol la rs ) Air (ticket price only) Auto (gas price only)

81 between 1995 and 2016. Phrased differently, the average American is boarding planes about 18% more than he/she was in 1995 (Table 8-2.) Implications for the Shorter-Distance Airline Market In 2016, the reported air trip length was 15% longer9 than it was in the base year, reflecting a higher portion of long trips and a lower portion of short trips in the total. This new data is generally consistent with the data shown in Figure 1-8, which showed the airlines placing a larger portion of their domestic flights in for longer trips, and smaller portion of the flights for the shorter trips. Our preliminary data suggests that the airlines have a lower market share for the medium- distance trips, and that the auto share is markedly higher than 20 years ago. All of this suggests a major subject area for exploration is the evident change in the importance of the automobile in most distance-based air markets. TABLE 8-2: CHANGE IN AIR TRIP BEHAVIOR BETWEEN 1995 AND 2017 1995  2016  INCREASE  2016/1995  Enplanements10 565,235,549 817,964,902 45% Enplanement per capita 2.2 2.5 18% Trips at 1.4 segments per trip 403,739,678 584,260,644 45% Round trips 201,869,839 292,130,322 45% Round trips per capita 0.77 0.90 18% THE EVIDENT CHANGE IN DISTRIBUTION OF TRIP DISTANCES Based on the available data, we can observe that the role of the auto for long-distance trips has increased significantly over the past two decades. Given that enplanements per capita has grown by about 18% this implies that long-distance auto trips have grown at a rate even higher than that. Using the preliminary numbers in this chapter, the mode share for auto is now somewhat higher than 50%, implying a major rise in absolute volume of trips. However, the changes in mode share do not occur uniformly for all types of trips. For very long trips, the share may be rising modestly, but the difficulties in competing with the far-cheaper air trip suggests something of a cap of growth in this market. For the shortest of the trips, the auto modes share in the base year were already quite high. This suggests that the auto has increased in importance especially between approximately 700 and 1,500 miles – a distance in which one en- route overnight is essential for most trips. 9 Source: BTS. https://www.bts.gov/content/average-length-haul-domestic-freight-and-passenger-modes-miles 10 From TAF summaries

82 8(E) ATTITUDES TOWARD THE LONG-DISTANCE TRIP AND ITS MODES OVERALL OBSERVATION: PEOPLE STILL WANT TO TRAVEL FIGURE 8-17: HEDONIC ATTITUDES ABOUT THE LONG-DISTANCE TRIP, BY MODE AND AGE GROUP In this section of Chapter 8, we review a wide variety of subject areas concerning attitudes toward transportation and specific modes. Results are graphed to easily allow spotting trends or consistencies in responses by gender, age and income. Americans seem to be somewhat positive about longer-distance trips. For instance, they agree with both the statement that there is something “exciting” about taking a trip by air, and “appealing” about taking a long trip by car, with more positive feelings about the plane than the automobile. Concerning long trips by car, Millennials are more positive than older age groups as shown in Figure 8-17 and males more positive about the trip than females, as shown in Figure 8-18. As mentioned earlier, respondents with lower income are more likely to drive even for longer trips, and their hedonic attitudes toward long car trips are somewhat more positive than those of more affluent respondents. 0 0.2 0.4 0.6 0.8 1 1.2 To me, there is something appealing about taking a long trip by car To me, there is something exciting about taking a trip by air Millennials Older Groups

83 ATTITUDES TOWARD THE AUTO FIGURE 8-18: ATTITUDES TOWARD THE AUTO BY AGE, GENDER, AND INCOME Survey respondents generally report positive attitudes toward the car. Attitudes toward the freedom and independence that a car provides vary relatively little by demographic group. The one exception is age, such that Millennials are significantly less likely to attribute freedom and independence to owning a car. Importantly, none of these subgroups prefer to rent or borrow rather than own a car– but Millennials were more positive about the new forms than the older groups: the scaling of agreement with the statement as presented here for the older groups on this question was twice as high as that of the Millennials. Males seem to be more open to the idea of renting/borrowing than females, with richer group over the less rich group. Again, this must be seen in the context of a strong level of agreement from both groups about the preference to own the vehicle. This is reflected by the lack of difference in support for the concept that ‘I need to drive a car to get where I need to go,” except that respondents with a lower income were more likely to endorse this statement than respondents with a higher income. To me, there is  something  appealing about  taking a long trip  by car I love the freedom  and independence  I get from owning  one or more cars I would rather own  a car than borrow  or rent it (REV)  I need to drive a  car to get where I  need to go I feel I am less  dependent on cars  than my parents  are/were Total 0.36 1.4 1.5 0.73 ‐0.41 Millennials 0.43 1.22 0.83 0.7 0.27 Older Groups 0.34 1.45 1.67 0.74 ‐0.58 Female 0.26 1.38 1.68 0.76 ‐0.53 Male 0.46 1.42 1.29 0.7 ‐0.27 Less Income 0.52 1.39 1.4 0.8 ‐0.41 More Income  0.14 1.42 1.63 0.64 ‐0.42

84 ATTITUDES TOWARD CONGESTION AND STRESS FIGURE 8-19: ATTITUDES TOWARD CONGESTION ON STRESS Survey respondents strongly believe that the long-distance trip by air will be less stressful and tiresome than an equivalent trip by car. Individuals with higher income and Millennials especially agree with that statement. Congestion is of more concern to Millennials than to other subgroups in the sample, both for intra- and intercity trips. Contrary to what might be expected, there was little endorsement of the statement that going through airport security is stressful. Compared to individuals with lower income, individuals with higher income reported being less stressed at security or discomfort with airport crowds. I feel really  stressed when  driving for a long  time in congestion  in and around big  cities To me, getting  stuck in traffic  congestion on a  long trip is a big  concern Compared to  driving a car for  this trip, I would  be less tired and  stressed if I took  the trip by air For me, the  process of going  through airport  security is stressful Dealing with the  crowds of people  at the airports is  uncomfortable for  me Total 0.55 0.84 1.42 ‐0.01 ‐0.06 Millennials 0.75 1.04 1.53 0.05 0.06 Older Groups 0.5 0.78 1.4 ‐0.02 ‐0.09 Female 0.67 0.86 1.47 ‐0.14 ‐0.15 Male 0.41 0.81 1.37 0.14 0.05 Less Income 0.58 0.87 1.3 0.13 0.06 More Income  0.51 0.79 1.58 ‐0.18 ‐0.22

85 FIGURE 8-20: PREFERENCES ABOUT LONG-DISTANCE MODE Driving for several days is perceived as being more unpleasant by Millennials compared to other subgroups. Further, Millennials are more likely to say that even driving with family and friends is unpleasant, that finding a hotel and the costs of overnight stays is a concern, and that the level of uncertainty associated with a car trips makes them choose the plane. By much smaller margins, females tend to evaluate trips by car more negatively than males, although they do not rise to level of influencing the modal decision. The thought of  driving for several  days with  family/friends is  unpleasant To me, the basic  idea of driving for  more than a day is  unpleasant When planning a  long trip, finding a  good hotel is a big  concern When planning a  long trip, the costs  of staying  overnight and  meals along the  way is a big  concern The level of  uncertainty  associated with a  multi‐day auto trip  tends to  make me  choose the plane Total ‐0.08 0.25 0.48 0.51 0.26 Millennials 0.34 0.74 0.72 0.96 0.65 Older Groups ‐0.18 0.13 0.42 0.4 0.16 Female ‐0.09 0.29 0.54 0.59 0.25 Male ‐0.07 0.21 0.41 0.42 0.28 Less Income ‐0.13 0.15 0.53 0.69 0.2 More Income  ‐0.02 0.39 0.41 0.28 0.33

86 ATTITUDES ABOUT ACCESS TO/FROM THE AIRPORT FIGURE 8-21:ATTITUDES ABOUT AIRPORT ACCESSIBILITY Most respondents in our survey did not feel that getting from the home to the airport was inconvenient, with Millennials showing the least satisfaction here. The needs at the nonhome end of the trip, however, are somewhat more complex. Subgroups generally agreed with the statement that transportation options are available at the destination airport, and subgroups did not vary substantially in this perception, although Millennials evaluated the options somewhat more positively. There is was also relatively little variation in the belief that one will need a car at the destination, with Millennials and females reporting less need for a car. However, most in the sample did not agree with the proposition that needing a car would make it more difficult to choose the plane over the personal car. Taken together, this implies that individuals think that some combination of car rental, and other forms access to vehicles can fulfill the need for flexibility at the destination. While Millennials had the highest proclivity to believe there are good public modes of transportation at the airport, there is little evidence that this belief explains the basic long-distance mode choice. Getting from my  home to the  airport is NOT  inconvenient I would need the  flexibility of a car  once I arrive in  <destination> Needing my car at  the destination  end of the trip  makes it more  difficult to choose  air for the trip At the airport of  <destination>,  there are good  taxis, vans and  buses to help me  get to my final  destination Total 0.34 1.01 ‐0.14 0.68 Millennials 0.14 0.86 0.05 0.92 Older Groups 0.39 1.05 ‐0.19 0.61 Female 0.42 0.89 ‐0.22 0.67 Male 0.24 1.16 ‐0.05 0.69 Less Income 0.28 1.06 0.01 0.64 More Income  0.42 0.95 ‐0.34 0.73

87 ATTITUDES ABOUT DISTURBING BEHAVIOR IN THE TRIP FIGURE 8-22: CONCERNS ABOUT THE AIR TRIP Note: Three questions in Figure 8-22 regarding disturbing behavior and the safety of air travel were reverse-scored to facilitate interpretation Individuals are less concerned about crime and disturbing behavior when traveling by plane compared to by car. However, survey respondents were more concerned about the lack of privacy when traveling by plane, with most moderately agreeing with the idea the having people seated so close is unpleasant. Generally, there was little variation in endorsing this statement across subgroups, though individual with lower incomes tended to agree slightly more than those with higher incomes. Even though individuals evaluate the lack of privacy negatively, they perceive air travel as safe. This is reflected in the widespread endorsement of the statement that they do not worry about personal safety or disturbing behavior when on the plane. Even though individuals on a whole agree that a plane trip is safer than a trip by car, there are subgroups such as Millennials and individuals with lower income who are less likely to agree. While they tend to agree that the plane trip has less disturbing behavior, Millennials, males and lower-income groups are less likely to agree older age groups, female respondents, and individuals with higher income. Females are more likely than males to worry about traveling with people they do not know, as are individuals with lower income. Millennials, compared to older individuals, are less worried about travel with people they do not know. Having people so  close to me in an  airline seat is  unpleasant to me I don't mind  traveling with  people I do not  know To me, making a  long distance trip  by car  exposes me  to MORE crime and  disturbing  behavior than  going by air I would NOT worry  about personal  safety or  disturbing  behavior if I went  by plane To me, taking a trip  by air is MORE safe  than taking that  trip by car.  Total 0.45 0.49 0.62 0.81 0.96 0 0 0 Millennials 0.49 0.7 0.48 0.53 0.66 Older Groups 0.44 0.44 0.65 0.88 1.03 0 0 0 Female 0.45 0.41 0.72 0.9 0.99 Male 0.45 0.59 0.5 0.71 0.92 0 0 0 Less Income 0.49 0.45 0.5 0.65 0.84 More Income  0.38 0.55 0.76 1.02 1.11

88 PREFERENCES AND CHOICE OF THE AIR TRIP FIGURE 8-23: AIR TRIP PREFERENCES Most Americans would prefer to fly for their trips over 300 miles. This especially holds for females, those with higher income, and for younger individuals compared to males, those with lower income, and those over 35. These subgroup differences also emerge for the belief that peers would choose air travel, with the exception that there are no gender differences in this regard. When asked the more abstract question about whether they would continue to travel by air if driverless cars were to become reality, most individuals respond affirmatively, with the exception of Millennials, who state that they would be less likely to choose air over autonomous cars. IMPLICATIONS FROM THE ATTITUDINAL DATA While the prior discussion of results focused on differences between demographic subgroups, it is important to keep the overarching similarities in mind: All subgroups would rather fly than drive, and all groups show some agreement that the air trip is safe, and that the issue of disturbing behavior is not a major concern with air travel. Getting through airport security and the issue of crowding at airports does not rise to be statistically relevant either way. On the other hand, sitting too close to others in airplanes is almost universally disliked. Even though the “basic idea of driving more than one day” is perceived moderately unpleasant by most, all but one subgroups (Millennials) reported liking driving with friends and family— with all groups finding some level of ‘appeal’ about a road trip. Some demographic patterns were consistent, with males more positive about driving than females, with lower-income respondents less worried about the long-distance trip than respondents with higher income. Within an overall pattern of trip approval, younger individuals tend to worry more about personal safety and disturbing behavior than those older individuals. Young individuals were also more worried about the details of finding lodgings on the way and paying for it. Consistent with expectations, younger respondents also reported a higher usage and need for information and communication technologies, and Millennials consistently gave lower ratings to the various concepts of auto dependence and auto need. To me, there is  something exciting  about taking a trip  by air When I have a  choice, I would  really prefer  to fly  REV People important  in my life would  choose the plane  over the drive by  car I would definitely  consider taking the  plane for a trip to  <destination> If driverless cars  were to become a  reality, I would  NOT be less likely  to travel by plane Total 0.72 0.48 1.18 1.6 0.33 Millennials 1.13 0.53 1.36 1.8 ‐0.18 Older Groups 0.61 0.46 1.14 1.55 0.46 Female 0.8 0.61 1.15 1.65 0.44 Male 0.62 0.32 1.22 1.54 0.19 Less Income 0.74 0.35 1.04 1.44 0.26 More Income  0.69 0.63 1.36 1.81 0.41