ii-a The Transportation Futures Project: Planning for Technology Change

2.3       Identification of New Activities in-Motion Using ATUS Data 

Analysis in this section uses publicly available data from the 2011-2014 American Time Use Survey (ATUS). ATUS involves computer-assisted telephone interviewing in which survey respondents are interviewed on the next day of a pre-selected day about how they spent their time from 4 AM on the pre-selected day to 4 AM of the interview day (days are selected to ensure proportional distribution across the days of the week and even distribution across the weeks of the year) (Basner, et al., 2007). Each activity described by the respondent is coded using a three-tiered scheme, going from a top-level category of activities, to subcategories, and then to third-tier activities that describe very specific actions. To provide a specific example, the top-level two-digit “traveling” category includes a second-tier four-digit category of “traveling related to caring for and helping household members” under which “travel related to caring for and helping household children” is a third-tier activity with a six-digit activity code.  

ATUS data has a total of 465 six-digit activity codes, of which 6 indicate missing data status and 459 indicate activity information. Kenyon and Lyons’ (2007) had suggested that three attributes of activities may be important to the extent to which activity in-motion is both possible and desirable, including the degree of locational dependence, the degree of continuity of engagement, and the degree of active attention. Based upon their framework, the following two sets of activity attributes are important in identifying possible activities in-motion during the era of self-driving and better mobile technologies;

 The degree of locational dependence (e.g., park use) and the necessity of importable equipment (e.g., billiards)

 The degree of engagement continuity and desired comfort level (e.g., uninterrupted sleeping) 

The degree of active attention will eventually be no longer important for enabling activity inmotion because, with fully self-driving cars, travelers are able to pay full attention to and be fully engaged in other activities.  It is worth noting that, because most existing research have been about activities during bus/rail trips in public vehicles, the number of possible activities in-motion is likely to increase significantly after self-driving cars.  Appendix A identifies possible activities that could be done in private vehicles. These activities have rarely being studied in the travel time use literature.  They are:

 A limited set of personal care activity including dressing & grooming, health-related selfcare, personal/private activities;

 A limited set of child care activities including reading to/with children, home schooling, and arts and crafts with children; 

 Eating and drinking;

 Tobacco and drug use; and 

 Participation in religious practices. 

Using the 2011-2014 ATUS data, Figure 2.3 presents the average daily time spent on the above activities by U.S. population aged 15 or older. On average, the U.S. population spent 118 minutes (roughly two hours) on these identified activity types. Note that child care and personal care activities included in Figure 2.3 are a limited set of these activities. Many child care and personal activities are location dependent and/or require importable equipment that cannot be conducted in travelling cars.  

ATUS.

2.4       Impacts of the New Technologies

1.      Unclear implications on mode choice. 

Evidence has shown that use of information and telecommunication devices allows transit riders to stay connected, improves transit experience, potentially increase transit mode choice and ridership. Yet, there is also evidence that information and telecommunication technologies (ICTs) are the least important mode choice factor among people older than 30.  Whether increasing available ICTs on public transit vehicles can shift people from driving to taking public transit is dependent upon the general performance of transit services such as service frequency, coverage, and reliability.

2.      Bigger and smaller cars. 

Most existing prototype self-driving cars appear to be small (see Figure 2.4 below). This is largely due to the early state of the technologies and the recognition that most travel is functional, for short distance trips, and involves one or two people. Further electric vehicles are presently constrained by batteries, and so there is a desire to limit vehicle weight to enable EVs.  Once the technologies are mature, there could be large market demand for more space to increase activity options in the cars, particularly for longer distance (intercity) trips.   As shown in Figure 2.3, most of the new activities require spaces for comfort levels, such as child care, personal care, and eating and

Toyota Swagger Wagon Supreme as an example of demand for more activity options in private cars.

3.      Longer trip distances and durations.

Mokhtarian and Salomon (2001) suggested that excess travel is more likely to occur as people increase the perceived positive utility of activities that can be conducted while traveling and the utility of the activity of traveling itself.  Because mobile technologies and self-driving cars will make travelling effortless, increase activity options in the car, and increase the productivity of travel time (as illustrated in Figure 2.5), it is highly likely a significant amount of extra travel will occur. 

Figure 2.5: Illustrative frequency distribution of “utility” of travel time by mode (adapted based upon Lyons and Urry, 2005)

           

4.      Larger spatial footprints with more trip chaining behaviors

Mobile technologies offer more opportunities to replace hub-and-spoke trips toward a central point of gravity, with circular trips chained together (Dal Fiore et al, 2014). The saved travel time and the increased utility of travel are likely to encourage visits to more distant but more attractive destinations (Dal Fiore et al, 2014). 

Figure 2.6: A two-step hypothesis on the implications of mobile technology for spatial behavior. Step 1: circular trips replace hub-and-spoke trips to and from a point of gravity;

Step 2: the number of trips increases (and/or chosen destinations for existing trips change) due to increased efficiency and information. (Figure downloaded from Dal Fiore et al, 2014)

5.      Behavioral impacts of technologies are likely to differ by gender, age, and SES.

Complex interaction effects are likely to occur when it comes to how the technologies may affect trip distance, duration, activity space, trip chaining, and car preferences across population groups. For example, women have traditionally commuted shorter distances but travel more for household support than men, which had led to limited economic opportunities for women. Self-driving cars may enable women travel farther for jobs because an increased number of personal care and child care activities can be done in self-driving cars.  Mobile technology and self-driving cars may also enable women to travel less for errands because more errands can be done online and self-driving cars encourage trip chaining activity. 

6.      Societal implications 

Mobile technology and self-driving cars are likely to have fundamental societal impacts such as transforming urban form and improving transportation equity. For example, although technologies may encourage extra travel for people with resources, they could also eliminate the needs for travel and the needs for auto ownership among people with lower incomes (see Chapter 4).   

Contingent on continued private vehicle ownership, people will be more likely to make housing location choices based on their residential preferences (such as school quality, neighborhood security, neighborhood cohesion etc.) than based upon spatial accessibility. As people who own autonomous vehicles are more willing to travel farther for more attractive destinations, fewer but larger business centers are expected.

           

Chapter 3: Information and Communication Technologies

3.1       Introduction

It is well perceived that information and communication technologies (ICT) have had pervasive impacts on modern society - they are changing how and where we work, shop, and in other ways live our lives.  ICT, particularly the Internet and mobile service, make virtual activities a viable alternative to traditional physical activities.  The growing penetration of ICT has important implications on transportation system.  Unsurprisingly, transportation policy-makers and planners expect tele-activities to replace some activities that require travel between places and hence alleviate transportation and related problems that many large metropolitan areas are confronting.  However, research conducted during past decades indicates that the impact of ICT on activity participation and travel is more complex than it appears at first glance. 

This report summarizes the effect of telecommuting and teleshopping on individuals’ travel behavior based on the literature, and discusses their potential impact on transportation system by 2040. It then summarizes the effects of other tele-activities on travel.  The final section offers an overall summary and reviews reasons why ICT are unable, by themselves, to solve transportation problems, but do bring about new challenges to our transportation systems.   

3.2       Telecommuting

Telecommuting has grown considerably during the past few decades.  Employees who primarily work at home increased by 61% from 2005 to 2009 and in 2010 about 16 million employees worked at home at least once a month, an increase of 62% from 2005 (Lister and Harnish 2011).  Telecommuting has particular appeal because of its potential to reduce commute travel and traffic congestion during peak hours, and energy consumption, and air pollution.  

3.2.1 The impacts of telecommuting on transportation 

Conceptual and empirical studies in the field of ICT and transportation suggest that telecommuting may interact with travel behavior in four ways:  substitution, complementarity, modification, and neutrality (Salomon 1986, Mokhtarian 1990).  Substitution denotes that an individual works at home instead of making a physical trip to her workplace.  Complementarity means that telecommuting generates new demands for other non-work trips.  Modification denotes that telecommuting does not affect the total amount of commute travel but changes the characteristics of trips such as mode choice, timing, and chaining.  Neutrality means that telecommuting has no impacts on travel behavior.  Among the four ways, transportation planners are most interested in substitution. 

The relationships between telecommuting and travel behavior vary based on the measures of travel behavior.  For example, an individual replaces a commute trip by working at home, but she makes a nonwork trip because of time savings from not making the commute trip.  In this case, the former represents a substitution effect and the latter is a complementary effect.  The net effect of telecommuting on total travel depends on the characteristics of the commute and nonwork trips. If the nonwork trip is longer than the commute trip, the net effect is complementarity.  If the nonwork trip is shorter than the commute trip, the net effect is substitution.  If the two trips have the same length but take place at different times (such as peak vs. non-peak hours), the net effect can be classified as modification.  

Significant research has been conducted to understand the impact of ICT on where work is done and how this affects travel.  Not surprisingly, previous studies offer mixed results.  Pendyala et al. (1991) found that telecommuters not only reduced commute trips, but also chose non-work destinations close to their home.  By contrast, Gould and Golob (1997) found that telecommuters generated new non-work trips, which offset the benefits of saved commute trips.  Using the 1991 Caltrans Statewide Travel Survey, Mokhtarian and Henderson (1998) found that home-based workers and non-home-based workers made a similar number of trips:  the savings from commute trips of home-based workers are almost completely offset by the increase in their non-work trips.  

Several studies have found that telecommuting is positively associated with commute distance (Mokhtarian, Handy et al. 1995, Zhu 2012). Because researchers are unsure about which comes first, telecommuting or residential location, the association may result from two potential causes:  One mechanism is that individuals choose to telecommute because they want to reduce the cost associated with their long commute distance; and the rival mechanism is that individuals choose to live farther away from their workplaces because they are able to telecommute (Ory and Mokhtarian 2006).  If the latter prevails, telecommuting may have an adverse effect on VMT.  For example, telecommuters may choose to live in far-flung exurban or rural areas where they depend more on private vehicles for their daily life than living in urban areas.  Using the 2001 and 2009 National Household Travel Survey (NHTS), Zhu (2012) explored the impacts of telecommuting on travel behavior.  He used the use of the Internet as an instrument to predict the probability of telecommuting, and then used the predicted probability of telecommuting to explain travel behavior.  He found that telecommuting has positive associations with the following behavioral variables: one-way commute distance, one-way commute duration, total work-trip distance, total work-trip duration, total work-trip frequency, total non-work-trip distance, total non-work-trip duration, and total non-work-trip frequency.  Overall, he concluded that telecommuting did not reduce but increase travel. By contrast, an aggregate time series analysis showed that telecommuting reduced vehicle miles travelled (VMT) by about 0.8% or less (Choo, Mokhtarian et al. 2005).  

After reviewing more than 30 empirical studies, Andreev et al. (2010) concluded that

telecommuting tends to reduce various measures of travel in the short run, but its long-term impact is uncertain, partly because of induced travel demand and residential location choice impacted by the ability of telecommuting.  

3.2.2 Trend analysis

In the American Community Survey (ACS), respondents were asked to indicate “what was your primary means of transportation to work during the survey week?” The choice set includes:

 Car, truck, or van - driving alone  Car, truck, or van - carpooled

 Public transportation  Walked

 Taxi, motorcycle, or bike

 Worked at home

Work at home (WAH) is the third commonly chosen primary means of transportation, after driving alone and carpooling.  In 2013, the share of workers who worked at home was about 25% more than that of workers who took public transportation^[11].  Note that this underestimates the share of less used modes, for instance, for people who work-at-home twice a week and drive three times will be reported as driving by 100% of those respondents. 

Applying the logistic growth curve to the ACS WAH data from 2005 to 2013, three different scenarios with different saturation levels are presented in Figure 3.1.  The low-saturation scenario assumes that WAH workers will eventually account for 10% of the total workforce in the US.  The rationale for this scenario is that although WAH is generally growing, its growth is slow, on average at about 2.5% annually during the past decade.  The share of WAH workers was at 4.4% in 2013 and it will take more than three decades to reach the saturation level if the growth rate remains stable.  The high-saturation scenario assumes that the maximum level is at 40%.  This is consistent with Handy and Mokhtarian (1996). Their assumption was drawn from Nilles’s assumption that “information workers” account for about 50% of the work force, and 80% of information workers have the potential to telecommute (Nilles 1988). Using California data, Handy and Mokhtarian also showed that the assumption is consistent with “the characteristics of the California work force and with studies of telecommuters” (p. 171). This saturation rate is also consistent with the estimates of Global Workplace Analytics^[12]:  50% of jobs are compatible with telework and 79% of workers want to telecommute (50%*79%=40%).  The medium saturation rate is set at 20%.  As shown in Figure 3.1, the share of WAH workers will grow to 6.7% ~ 7.6% under the three scenarios by 2040.  This represents an increase of 2.3 ~ 3.2 percentage points over the 2013 level.  The growth is not substantial, but it is also non-trivial because the impact of travel reduction on congestion level is exponential. 

It is worth noting that WAH workers in the Census or ACS include self-employed workers such as farm, domestic, and service workers.  Their home-based work does not substitute for a commute so they are not telecommuters per se (Mokhtarian, Salomon et al. 2005).  Therefore, we are likely to misestimate the number of telecommuters.  In 2005, 3.62% of the work force worked at home.  Excluding the self-employed workers, about 1.37% of workers considered their home primary workplace.  That is, self-employed workers accounted for about 2.24%.  In 2012, employee teleworkers totaled about 3.3 million^[13], accounting for 2.3% of the work force, whereas selfemployed workers accounted for 2.03%. 

Further, the ACS does not consider occasional telecommuters, who do not choose home as their primary workplace.  That is, the ACS undercounts the number of telecommuters.  However, because there is no consensus on how to define telecommuting, reliable historical data on the number of telecommuters in the USA are not available (Mokhtarian, Salomon et al. 2005).  WorldatWork Survey estimated that about 16 million workers telecommuted at least once a month in 2010 (Lister and Harnish 2011).  This accounted for about 11% of the work force.  This number is expected to grow. However, in terms of telecommuting frequency, the followers tend to telecommute less often than early adopters of telecommuting (Handy and Mokhtarian 1996). However, telecommuting frequency should be smaller than this rate because working at home includes work after regular hours and employees may work at home partly while working at regular workplace.  

The ACS does not capture those who work at other places such as client offices, telecenters, coffee shops, and vehicles (Lister and Harnish 2011).  In the 1990s, neighborhood telecenters were promoted in California, under the sponsorship of the State of California Department of

Transportation (Caltrans) with funding from the Federal Highway Administration (Varma, Ho et al. 1998).  However, the pilot program was not promising and hence discontinued.  There are also 53 million freelancers in the USA, who account for 34% of the work force^[14].  Some of them work in coffee shops or libraries although the exact number is unknown. Many freelance as a second or third job. 

An employees’ decision to telecommute and hence the penetration of telecommuting is affected by employers’ attitudes toward teleworking.  Although telecommuting has the potential to increase productivity and save office-related costs, managers worry about the lack of team-building, commitment, and control (Baruch 2000).  In 2013, Hewlett-Packard required most of their employees to work at the office instead of from home in the name of building engagement, collaboration and innovation^[15].  Yahoo and Best Buy also implemented similar polices to scale down or abandon their telecommuting programs²⁶.   

3.2.3 Perspective

Based on the American Community Survey, an additional 3% of the work force may work at home by 2040.  Although the literature offers mixed evidence on the impact of telecommuting on travel, it is expected to help reduce commute travel during peak hours.  However, the ability to telecommute may motivate workers to relocate farther away from workplace in the long term.  This will increase commute distance on non-telecommuting days and may result in increases in distance traveled by automobile.  

Travel time savings from telecommuting will bring about additional non-work travel, which offsets the benefits of telecommuting.  Since non-work travel tends to occur closer to home than commute trips, telecommuting is expected to reduce VMT slightly (Choo, Mokhtarian et al. 2005). Because telecommuters’ activity space is closer to home than to work on telecommuting days (Saxena and Mokhtarian 1997), telecommuting will help ease traffic congestion on the freeways and around employment centers even if the number of trips (including both commute and non-work travel) is the same on telecommuting and non-telecommuting days.  

Although coffee shops have become a new workplace for many workers, they are not expected to materially influence vehicular travel, similar to neighborhood telecenters.   In particular, workers at neighborhood telecenters tend to have a higher number of return home trips and other non-work trips on telecommuting days and they also tend to shift from other modes to driving alone (Balepur, Varma et al. 1998).  The potential increases in engine cold starts will further reduce the environmental benefits of telecommuting. 

It is worth noting that ICT have made traditional “9-5” work schedule and fixed locations obsolete.  People can work at home, at restaurants, while travelling and at any time when it is convenient. 

People can work remotely as well as at primary office in a day. Therefore, not all the implications of teleworking relate to transportation.   

3.3      E-shopping

“Shopping online is about to explode. Retailers of all types are expanding product offerings, adding in-store pickup, free shipping and experimenting with social media. It’s getting harder to tell pure play Internet retailers from the bricks and mortar shops with online portals, and all of them are reinventing how we’ll shop online in the future” (Heller 2011).  In 2000, retail sales of Ecommerce^[16] in the USA totaled about $27.5 billion, according to US Census E-commerce Report.  In 2014, the sales have increased to $297.5 billion.  The 10-fold growth of E-commerce signifies its pervasive impacts on retail industries as well as transportation.  

3.3.1 The impacts of e-shopping on transportation 

Transportation planners are interested in the changes that e-shopping will bring to transportation system. In particular, planners tend to focus on the effects of e-shopping on individuals’ activitytravel patterns. The potential of online buying to substitute for traditional in-store shopping and reduce personal shopping travel has important implications for travel demand management and congestion mitigation.  According to the 2001 National Household Transportation Survey (NHTS), on average shopping travel accounted for 14.4% of annual VMT per household and 21.1% of annual vehicle trips per household (Hu and Reuscher 2004).  Therefore, the growth of online buying could have the potential to reduce traffic if it does replace physical shopping – substitution.  On the other hand, if e-shopping induces new shopping trips, it is likely to generate more personal travel to existing transportation systems – complementarity.  

What does the literature say on the impact of e-shopping on traditional store shopping?  Some descriptive studies offer mixed outcomes.  Cairns et al. (2004) report that online shopping reduces at least one vehicular trip for 80% of 538 U.K. Internet users polled by British Telecom.  However, Sim and Koi (2002) state that 88% of 175 Singapore online buyers do not observe any influence on their travel to stores.  Corpuz and Peachman (2003) show that 14% of respondents in Sydney would have not made the purchase if online shopping were not available and 19% would have made a special trip to stores.  That is, online shopping not only induces consumption but also replaces physical trips.  

Several multivariate analyses have been conducted to understand the relationships between online shopping and store shopping. Using the 2000 San Francisco Bay Area Travel Survey, Ferrell (2004, Ferrell 2005) employs two-step linear regression and structural equation models, respectively, to examine the impacts of teleshopping on shopping-related travel behavior.  In the two studies, he finds substitution, complementarity, and neutrality depending on the dependent variables of interest.  Farag and colleagues conducted a series of studies in the Netherlands and find that online shopping frequency is positively associated with store shopping frequency, even after controlling for shopping attitudes and/or demographics such as income (Farag, Schwanen et al. 2005, Farag, Krizek et al. 2006, Farag, Schwanen et al. 2007).   A complementary relationship is also found in Northern California (Circella and Mokhtarian 2009), Israel (Rotem-Mindali 2010), Norway (Hjorthol 2009), and the Twin Cities (Cao, Xu et al. 2012).  Freathy and Calderwood (2013) and Calderwood and Freathy (2014)  find both complementary and substitution effects, but they also state that online shopping has a limited impact on shopping travel in the Scottish Isles. Hiselius, Rosqvist et al. (2015) also conclude that there are few differences in trip frequency and travel distance between frequent online shoppers and other shoppers.  Although complementarity seems to dominate the literature, the impact of the Internet on shopping is more complex than it appears. By decomposing the shopping process of books and digital products, Cao (2012) concludes that the Internet facilitates a hybrid shopping process:  people use different shopping channels at different shopping stages.  For example, an individual may become aware of a product in a store, search for product information through a home computer, and then make a trip to the store to acquire it.  Using the 2009 National Household Travel Survey (NHTS) data, Zhou and Wang (2014) find that the relationships between online shopping and shopping trips are asymmetric: online shopping encourages store shopping but the latter reduces the former.  

A positive association between online shopping and store shopping is often reported in the literature.  This is not surprising because ICT gives individuals many new motivations to engage in physical travel (Dal Fiore, Mokhtarian et al. 2014).  In the context of shopping, Mokhtarian and Circella (2007) propose three mechanisms for such a positive association.  First, buying online induces purchases for other related products such as accessories, which may take place in stores (and conversely for shopping in stores).  This represents a direct causal influence.  Second, Internet purchasing eliminates travel time to and from stores and the time saved may be used for other shopping-related activities and travel.  This is an indirect causal influence.  Third, the association is spurious, resulting from factors antecedent to both online shopping and store shopping.  For example, women may buy both online and in stores more frequently than men, and affluent people and those who enjoy shopping may have a high demand for shopping through multiple channels.  Furthermore, online shopping enables consumers to fragment their shopping.  For example, instead of buying all products through a single trip, consumers may purchase the products through different channels in multiple episodes.  In this way the ability to shop online may induce additional shopping trips to stores.  

The relationships between online buying and store shopping are complicated and far from being settled knowledge (Rotem-Mindali and Weltevreden 2013).  Furthermore, the Internet allows users to search product information online.  This also increases store shopping trips (Farag, Schwanen et al. 2007, Cao, Douma et al. 2010).  Moreover, the rise of e-shopping means increasing delivery trips to end consumers, which leads to the growth in freight transport (See Chapter 8) (Anderson, Chatterjee et al. 2003).  

On the other hand, the relationships between online shopping and store shopping may differ by product type (Cao and Mokhtarian 2005).  Product attributes greatly affect the suitability for online shopping and hence the potential substitution effect (Peterson, Balasubramanian et al. 1997).  Peterson et al. (1997) argue that low-cost, frequently-purchased, and physical products are more suitable for traditional stores; low-cost, frequently-purchased, and informational products are more likely to be purchased online; and high-cost, infrequently purchased goods can be purchased through both traditional stores and Internet stores.  E-shopping is likely to have a detrimental impact on traditional retailers on digital products – music and books.  For example, shares of Best Buy decreased more than 30% during 2011; the decline in sales was due to fierce competition from online vendors as well as Wal-Mart and Target (La Monica 2011); Borders, the second-largest bookstore in the U.S. filed for Chapter 11 bankruptcy protection in February 2011.

3.3.2 Trend analysis

Applying the logistic growth curve to the E-commerce data from 2000 to 2015, three different scenarios with different saturation levels are presented in Figure 3.2.  The low-saturation scenario assumes that retail sales of E-commerce will eventually account for 10% of total retail sales.  The rationale for this scenario is that although e-commerce is growing, its growth rate is slowing down during the past several years.  This rate is optimistic because the share of E-commerce in total retail sales in China is already at 10.7% (CNNIC 2014).  The high-saturation scenario assumes that the saturation rate is at 50%.  E-shopping will not eliminate traditional shopping because not all products are suitable for e-shopping and many people consider shopping is an important leisure activity. Since “53 percent of consumers would prefer to see, feel and touch a product before buying”^[17], the saturation level of e-commerce might reach about 50% of total retail sales.  This rate

           

In the low-saturation scenario, the share of e-shopping will reach 90% of the saturation level (9%) in 2023. In the medium-saturation scenario, the share of e-shopping will reach 90% of the saturation level (27%) in 2037. In the high-saturation scenario, the share of e-shopping will reach 90% of the saturation level (45%) in 2042.  This implies that the peak impact of e-shopping on traditional shopping is likely to occur within the time frame of the next long-range transportation plan.   

3.3.3 Perspectives

Though shopping travel has declined over the past decade (as discussed above and in Chapter 8), there is to date no persuasive causal evidence that online shopping is responsible, and the degree to which it has the potential to significantly reduce individuals’ shopping trips to stores in the future remains uncertain.  Most studies concluded a complementarity effect of e-shopping on traditional store shopping (Andreev, Salomon et al. 2010).  Although a few activity-based studies found that e-shopping tends to reduce shopping travel and other leisure activities (Ferrell 2004, Ferrell 2005, Ding and Lu 2015), activity diaries conducted for one day are not long enough to capture the whole process of e-shopping, which can take days or even weeks to complete because of its temporal fragmentation.  

Because almost all existing studies are based on relatively low shares (less than 10%) of ecommerce in total retail sales, it is less certain about the relationships between e-shopping and store shopping when the share grows much higher.  However, since ICT tends to have a dominant complementary effect on transportation in the long run (Choo, Lee et al. 2007), it is not unreasonable to expect that the number of shopping trips will grow or at least stay the same, whereas the delivery trips and freight transport associated with e-shopping will definitely grow.  If the saturation rate of e-commerce is at 30% or higher, the delivery traffic on local streets will increase substantially.  The bypass of wholesalers and retailers in the network from manufacturers to consumers will significantly change the operation patterns of freight transportation (Anderson, Chatterjee et al. 2003). 

The impacts should be considered in the next regional transportation plans because e-shopping is expected to be close to saturation before or around 2040.    

3.4       Other Tele-activities and Travel

Andreev and colleagues (2010) reviewed about 100 studies on tele-activities.  The vast majority of the studies are on telecommuting and teleshopping.  Some studies focus on tele-banking, teleleisure, and so on.  Some studies found that ICT tend to increase travel for leisure activities (Andreev, Salomon et al. 2010).  Limited evidence suggests that telebanking and telemedicine tend to replace physical travel to corresponding service locations (Andreev, Salomon et al. 2010).  This may be because banking and seeing a doctor are more of mandatory activities.  On the other hand, the time saved by telebanking and telemedicine may be used for other indoor and/or outdoor activities.  

Telemedicine is particularly intriguing because of its potential to reduce travel during inconvenient occasions such as midnight, weekend, and while working.  On the other hand, telemedicine will not make traditional office visits obsolete.  Telemedicine is more suitable to minor health problems and is an alternative to some urgent care and primary care visits.  It is particularly viable for specialty service in rural areas where medical service is inadequate.  It was projected that global telemedicine market will be double from 2014 to 2020^[18].  The penetration of telemedicine has many barriers from the perspectives of doctors, patients, and insurance industries^[19].  Further, it greatly depends on government regulations and laws; it is limited or even prohibited in many states³¹. 

Further, online schooling/training/conferencing has become more and more popular.  However, studies on the relationships between these virtue activities and travel are scarce.  There is no rigorous inference regarding their impacts on travel.  For example, teleconferencing enables people to “meet” without travel.  On the other hand, it increases people’s social network and also increase motivations to meet other people in person (Button and Maggi 1995).  The net effect is unclear.   Social travel has declined in the past decade, however.

Bandwidth can be a barrier for some tele-activities such as telemedicine and tele-learning, particularly in rural areas where broadband service remains limited.  The inevitable improvement in service quality will encourage those in remote areas to engage in virtue activities.  This will improve their access to jobs and services and reduce their long-distance travel. 

3.5       Summary of ICT and travel

Since the introduction of ICT, researchers have examined their impact on transportation system.  Early studies have extensively explored the relationships between telecommuting and travel and estimated the aggregate impact of telecommuting on transportation system.  As e-shopping proliferated in the 2000s, geographers and transportation planners have investigated the influences of e-shopping on traditional store shopping and related travel behavior.  Overall, the studies on telecommuting and e-shopping dominate the literature on the connections between ICT and travel; a few studies started to explore online banking, telemedicine, and virtual leisure activities. 

The literature on telecommuting and e-shopping offers mixed evidence on the impacts of ICT on travel. For telecommuting, the key findings include the following:

 Telecommuting reduces commute travel during peak hours as well as non-peak hours;

 Telecommuting enables commuters to move farther away from their employment location and become auto-dependent;

 Telecommuting increases non-work travel, which takes place mostly close to home;

 Telecommuting reduces VMT slightly but it helps mitigate the growth of congestion on freeways; 

 Telecommuting is likely to influence the travel of other household members but concrete evidence is limited; 

 Working at informal sites such as coffee shops will not materially reduce VMT. 

For e-shopping, the literature shows that

 Online searching is positively associated with store shopping (people who buy online also buy in person more); 

 Studies are mixed on whether e-shopping reduces travel to stores and other leisure activities in the short term;

 E-shopping for now digital products (books, records, videos) has already changed retail patterns and shopping travel behavior; 

 Online buying increases delivery traffic and freight transportation;

 Existing studies are based on the small share of e-shopping in retail industries.  If its share is large enough to change the distribution of commercial land uses in the region, e-shopping will have a profound effect on shopping-related travel.   

From the transportation perspective, a goal behind the drive to promote ICT is to reduce travel.  Although ICT are often promoted as a virtual alternative to physical travel, transportation planners should be realistic about the relationships between ICT and travel:  although the short term effect of ICT on travel may be substitution, in the long term, travel demand has historically grown as ICT demand increases (Choo, Lee et al. 2007, Choo and Mokhtarian 2007, Andreev, Salomon et al. 2010). The extent to which that relationship holds in the future is an open question, as ICT becomes higher and higher quality. 

Based on an extensive analysis of the literature, Mokhtarian (2009) offers twelve reasons for the paradoxical results:

1.      “Not all activities have an ICT counterpart”. For example, activities such as gardening and repairing need workers to be at specific locations. 

2.      “Even when an ICT alternative exists in theory, it may not be practically feasible”.  For example, technology may not be available at certain time or certain areas. 

3.      “Even when feasible, ICT is not always a desirable substitute”. For example, some people consider shopping an important leisure activity and they do not want to entirely eliminate it. 

4.      “Travel carries a positive utility”. Sometimes, people travel for its own sake, such as the Sunday drive, jogging, or just getting out of the house. 

5.      “Not all uses of ICT constitute a replacement of travel”. For example, if teleconferencing were not available, people may not hold a meeting at all. That is, technology induces demand.

6.      “ICT saves time and/or money for other activities”. For example, if an individual saves travel time because of working at home, she may travel for other purposes within her travel time budget. 

7.      “ICT permits travel to be sold more cheaply”. For example, the Internet makes comparable shopping for air tickets easier and hence lowers the cost of travel for consumers. 

8.      “ICT increases the efficiency of the transportation system, making travel more attractive”. For example, the implementation of ramp metering improves traffic flow on the freeway and hence makes long-distance commuting more desirable than before. 

9.      “Personal ICT use can increase the productivity and/or enjoyment of travel time”. For example, watching iPad while travelling reduces the disutility of travel and hence the subjective value of travel time is not all wasteful (see Chapter 2). 

10.  “ICT directly stimulates additional travel”. For example, an instant message or use of mobile phone may motivate an individual to visit certain place and increase car use (Hjorthol 2008). 

11.  “ICT is an engine driving the increasing globalization of commerce”. The globalization of commerce promoted by ICT induces additional travel such as freight transport from overseas and business travel among different countries. 

12.  “ICT facilitates shifts to more decentralized and lower-density land use patterns”. For example, the ability to telecommute may enable individuals to move to a distant suburban neighborhood and suburbanites tend to travel more than urban residents.  Therefore, ICT indirectly increases travel (albeit off-peak). 

On the other hand, Mokhtarian offers four scenarios that ICT has the potential to reduce travel. 

ICT sometimes directly substitutes for travel.  For example, telecommuting slightly reduce VMT (Choo, Mokhtarian et al. 2005). “ICT consumes time (and/or money) that might otherwise have been spent traveling”.  For example, online shopping reduces out-of-home leisure activities (Ding and Lu 2015).  “When travel becomes more costly, difficult, or dangerous, ICT substitution will increase”.  For example, when there is a weather event, like a blizzard or hurricane, telecommuting becomes attractive.  Finally, ICT makes car-sharing more convenient.  Without mobile applications, the amount of shared ride will reduce significantly.  

Overall, although the promotion of ICT aims to reduce travel, it simultaneously induces new travel.

The net effect of further proliferation of these technologies remains ambiguous.                                                   

Chapter 4: Mobility-as-a-Service

For physical (rather than virtual) objects, one person's use prevents someone else's use. Many physical objects are not fully utilized by their owners. Cars, for instance, typically remain unused 22 or 23 out of 24 hours in the day. The basic idea of collaborative consumption is that things can be shared rather than individually owned (e.g., one rents hotel rooms and cars rather than buying condos and vehicles when on travel; it is now common to rent music, videos, and books), increasing economic efficiency.

Previously cars, taxis, and hotel rooms were rented from companies which could achieve large economies of scale, but in a way that was inconvenient for customers. Now it is possible to rent couches and cars and rides from individuals with excess capacity at the push of a button through an app. The degree to which economies of scale trump the network effects of distributed suppliers awaits to be seen. We roll out three dimensions of sharing—cars, rides, bikes—that have emerging implications for transport.  

4.1       Sharing Cars

"Carsharing" is a marketing term for modern car rental services. People are "sharing" the car just as they share a hotel room — by paying a third party and using it at separate times.^[20] In most US carsharing services, the ownership is by a private for-profit company (a few are non-profits); the service is not owned by its members. They employ mobile information technologies to avoid the repetitive contract negotiations that were once common when one wanted to rent a multi-thousand dollar vehicle from a multi-national corporation. 

There are notable differences between traditional and modern car rental. First, the car is reserved via a website (or more recently, a smart-phone app) and unlocked via a special member card or the phone itself, no real-time intervening labor is required between you and the car. Second, the newest of these services allows you to pick up and drop off your car at any legal on-street parking space, no longer only at special stations or locations (this model is used by Car2Go in Minneapolis and Saint Paul).^[21] Third, the rentals can be by the number of minutes, rather than days, so cars may be less expensive to rent for a short trip. 

In the US the first break-out company in this sector was Zipcar,^[22] which adopted at the station-based model widely used in Europe^[23], requiring cars to be returned to their rental station. The problems with station-based carsharing vs. the Car2Go models are several. First the stations may be inconvenient. Second the user has to know exactly how long is the trip, since overage charges are significant (some $50). From the station-based carsharing company’s perspective, with such a thin fleet of vehicles, the overage charge is essential to guarantee the car will be available for the next renter, but to avoid being late, the renter had to reserve the car for a longer time period than actual used. Third, the technology is a bit wonky, leading to risk of having no car available at all. 

Zipcar went public in 2011 with 8,000 cars, 500,000 members and $186 million in revenue. Never profitable, it was acquired by Avis in 2013 at about one-third its 2011 market capitalization. 

Car2Go (and others) all strive for various improvements on the same theme, most notably the ability to park anywhere rather than at a station.³⁶ New economic models include RelayRides, which allow individuals to rent their own car, and is perhaps best suited to airports where cars are otherwise parked for a long time. 

Like most automakers, BMW is entering the carsharing market with DriveNow. Ian Robertson with the company says “As a mobility provider, the BMW Group is not simply an automobile manufacturer”.³⁷ BMW also combines carsharing with leases, so for instance, you can lease a small EV during the week and get access to a larger vehicle on the weekend.

The importance of carsharing is not as a replacement for rental cars, which are still standard in their traditional market of airports and auto replacement during servicing — though that may change as well. Carsharing also is not cost-effective as a replacement for daily commuting trips. However if you walk, bike, or take transit to work, it might be good to replace owning a car, or second car, for the occasional (say weekly) trips that are too far to practically walk or bike, and too inconvenient to use a transit system that serves downtown well and little else. In crowded urban areas where paying for parking at your home is a real financial burden, carsharing is more promising than most of America where parking is practically free.

                                                 

³⁶     Car2Go, a newer service, works better for users. Some 535 cars are deployed throughout Minneapolis and St. Paul, so a user can reserve a car (no more than 30 minutes in advance) within walking distance. The app will show directions to the car. 

Once you enter the car, find the key, turn the ignition, fill out a brief survey on car quality, and off you go. Unlike Zipcar, Car2Go bills by the minute. When you are done, you check out from the car.

Car2Go uses Smart Fortwo vehicles, which are the smallest and least expensive commercially available road-worthy vehicles in the US.  The Smart Fortwo, despite being a unit of Daimler, best known for the Mercedes Benz, is not a luxury vehicle.

³⁷     Vijayenthiran, Viknesh (2011-03-24) “BMW Is No Longer Just An Automaker, It’s A Mobility Provider.” Motor Authority http://www.motorauthority.com/news/1057282_bmw-is-no-longer-just-an-automaker-its-a-mobility-provider

Figure 4.1: North American Carsharing Growth

The car-shedding question remains: How many households will surrender a second (or first) car for the occasional trip?^[24] Is the market thick enough that the likelihood of finding a car nearby is high enough that it is reliable enough to use? With Car2Go there is no guarantee there will be a car within walking distance. While the service tries to rebalance their fleet if it gets out of whack (for instance, all the cars are in St. Paul and all the demand is in Minneapolis), this is not free for them, nor does it occur instantaneously. This is where other services (taxi, transport network companies, transit) come in as backups. This is also where autonomous vehicles can be important.

Nevertheless, people don't want to think about every transaction, and if they are charged per use, obviously would use less, and will be less happy, and more determined to get a car of their own to avoid transaction costs. Cars have costs of their own, but they are less frequent and less obvious. If the charges are invisible though, people may not think about them.  Just as information technology went from terminals and mainframes to personal computers, and internet cafes to internet at home, transport technology went from trains and transit to private transport once we could afford it. The cost savings will have to be considerable for most people to want to go back. But habits are easier to form in the young. An urban college student who joins Car2Go may keep it after they graduate if they remain city-bound.

The best market for Car2Go may be the urban hipster: with enough money to afford, enough transit to get to work and back with minimum hassle, enough childlessness to have a simple schedule, enough desire to signal greenness to avoid owning a car, but enough sense and desire for dates in the country or trips to Ikea to recognize the occasional need. 

Figure 8.1 shows trends on carsharing in North America. It is not clear where market saturation is, and whether the dip in 2015 is just a data issue or indicative that perhaps ridesharing is stealing some carsharing thunder. Notably carsharing company Shift shuttered in Las Vegas in mid-2015.^[25] 

4.2      Sharing Rides

Just as carsharing refers to modern car rental services, "ridesharing" is a marketing term for modern taxi services, and providers like to think of themselves as “transportation network companies.”^[26] You might have thought ridesharing was the same as car-pooling. And it is, if you think of modern "ridesharing" drivers as your friends giving you a lift (or in the name of one company a Lyft), not for money, but for a voluntary ‘donation’. Whether this attempt to skirt the rules and regulations of taxis succeeds is a battle to be fought out in thousands of local markets globally. In markets where an agreement has been reached, the ‘donation’ results in an actual charge and the process—enabled by smartphones—is taking off.⁴¹ 

The car you get with Lyft (or UberX) is the driver's personal car, not a fleet vehicle.

Lyft is in many ways simply an app with a back-end (rather, "cloud-based") dispatch service. They claim to be a "transport network company whose mobile-phone application facilitates peer-to-peer ridesharing by enabling passengers who need a ride to request one from drivers who have a car." They insist the drivers are independent (as are the riders). The difference between this and a taxi dispatcher is thin. An internet definition of a taxi is "a car licensed to transport passengers in return for payment of a fare, usually fitted with a taximeter." So for taxicabs, the arrangement between the rider and the passenger is mediated by the government (which licenses the vehicles). 

Are Lyft drivers licensed to transport passengers for payment? This is a major point of contention. They are licensed drivers, and any licensed driver (above a certain age and level of experience) is eligible to carry passengers. The cars are private cars (at least sometimes) though that is little different than how taxis operate in other parts of the world. Many Singaporean taxi drivers will take fares when going between where they are going anyway, but otherwise treat the taxis as a personal vehicle.

Lyft now offers jitney (shared taxi, dollar van, informal transport) type services, dubbed Lyft Line in selected markets. (Uber has the similar UberPool)  These serve either one pickup going to multiple destinations, or multiple pickups going to one destination, or multiple origins to multiple destinations, and compete with both taxi and public transit. (Though it would not be exactly fixed routes, one could imagine regular runs with a known coterie of passengers). This is at a lower rate. While these services are at the time of this writing only in San Francisco and New York, Lyft now claims that Lyft Line comprises 50% of Lyft's rides in San Francisco and 30% in New York.⁴² (Not all of Lyft Line customers wind up in a shared ride, they just indicate a willingness to share in exchange for a lower fare, and get the lower fare regardless of whether another passenger can be found). The ease of making ride requests and payments is what drives many customers to choose Uber or Lyft over traditional taxis.

                                                 

⁴¹ How does Lyft (or any other Transportation Network Company) work? From the perspective of the user (1 and 2 are one-time) the following is the sequence:

1.        Download the App.

2.        Enter the required info

3.        Open the App and summon a Ride.

4.        Get in the Lyft vehicle when it arrives (the driver will usually call to confirm pickup location/time), it has a small mustache in the window.

5.        Tell the Driver the Destination — You can enter this in the app as well, though it doesn't affect who picks you up (yet) apparently. This is an opportunity for efficiency, as drivers, for instance, may be happy to go towards their home near the end of a shift, but not away. On the other hand, that might lead to too few drivers "bidding" on prospective customers.  6. Ride

7.        Get out of the Car.

8.        Check App to rate the driver. Payment is automatic unless you want to change your payment.

⁴² Bregman,                 Susan     (2015-04-23)        Uber       and         Lyft        claim      carpooling             success.               TheTransitWire.com. http://www.thetransitwire.com/2015/04/23/uber-lyft-claim-carpooling-success/

Differentiating status and class is another important element. Users are hip enough and wealthy enough to use the new technology and not have to sit where others from other classes have sat before. As these services become widespread, humans will undoubtedly develop new forms of elitism.

Venture capitalists believe these will be very successful companies. Uber has been valued at over $50 Billion. There are however many competitors (including of course Lyft, as well as BlaBlaCar, Didi Kuaidi,^[27] Gett, Curb, Hailo, Blacklane, Sidecar, Zimride, iHail, and Flywheel, among others) and the stickiness of riders and drivers to any particular company is weak and their limited advantages to larger services over smaller ones.^[28]  The competition from the new entrants has driven taxi companies to step up their game, a number of the services listed are better interfaces to traditional taxi. Drivers are already simultaneously on multiple networks, so the expected pickup time doesn’t vary much from one app to another. Waze, a subsidiary of Google, is testing a true peer-to-peer, realtime, no payment ride-sharing service.  Shuddle, KangaDo, and HopSkipDrive are trying to be the “Uber for kids”.^[29]  SheTaxis aims to be an “Uber for Women.”^[30] 

4.3       Sharing Bikes

A recent popular internet meme^[31] noted that in Europe, bicycles were outselling cars.^[32]  The National Bicycle Dealers Association (NBDA) reports^[33] annual bike sales on the order of 18.7 million for 2012. In contrast, US car sales are on the order of 8 million, along with another 8 million light trucks. So where are all these bikes, why are they not seen more on roads everyday? 

Many of the bikes sold annually are for children (5.7 million of the 18.7 million are below 20 inches wheel (50 cm)), but even so, 13 million are 20 inch and above wheel size, and 13 million is still much bigger than 8 million cars (and near 16 million light vehicles, note also that many light vehicles are not for personal use). Even just inspecting specialty bike shops, which sell at the higher end, that's nearly 3.1 million bikes per year, which while less than cars, is still a pretty big number.

Yet, the volume of trips by bike and certainly miles by bike are much lower than by car and are not poised to overtake in the US. We don't even see 3.1 million bike commutes daily in the US (The American Community Survey reports (undoubtedly an under-report, ^[34] but still a small share) 865,000), so these are more likely for recreational than utilitarian purposes.   

Another reason for this statistic is that bikes don't last as long as cars.^[35]

So lack of bikes does not seem to be a problem, but bikes where you want them may be. If I didn't take my bike to work, I can't use my bike for a lunchtime ride at work. If I am a tourist, I probably didn't bring a bike with me. Wouldn't it be great to just get a bike when and where I want it, and abandon it there at the end of the trip. Well, it is not quite modern carsharing, but widely dispersed bike stations make this closer to possible. As shown in Figure 8.2, bike sharing systems have grown in many cities worldwide.

Bikesharing, the modern version of bike rentals, (just as carsharing is the modern version of car rental) has both a membership and per-use model. A member signs up online, gets an electronic key in the mail, and can visit a bikeshare station in their system and simply insert a key in the slot next to the bike they want, and pull it out, and remove the bike. They then have, say, 60 minutes to use the bike before it needs to be returned to a station (any station, not just the one it was borrowed from). A one-time user has to insert a credit card. These measures ensure the bicycles are returned and not found in the bottom of a nearby river, the sad outcome of early free bikesharing schemes. The bikes, while functional, are unlikely to be a model a regular bicyclist would purchase. They are especially heavy, and only 3-speed, so the risk of theft is relatively low.

Bikesharing is about transport as much as active recreation, and a way of connecting these two things.

           

 

Figure 4.2: Growth of Bike Sharing Systems Globally

The number of systems may be leveling off. Maturation in the number of systems is a good thing in many respects. A next step, however, is to stop adding new systems in favor of increasing the services of existing systems, inter-connecting and inter-operating (maybe even consolidating some). A related aim is broadening the subscriptions globally, so that memberships can be used on any system in the world.^[36] 

Nice Ride Minnesota, the largest bikesharing system in the state, has shown continuous growth from 2010 through 2015, as shown in Figures 4.3 and 4.4.^[37] This is complemented by a significant increase in those years in bicycle-dedicated infrastructure, including separated bike-lanes. More bike traffic is expected to have a safety-in-numbers property that crash rates per bicyclist will decline with an increase in the number of bicycles. 

Bikesharing can function as an extender of transit service, as people take transit, transfer to a bikeshare to reach a final destination (or for recreation), and to return to the transit stop. This requires a station at the destination end, or the destination to be short duration. Future bikesharing need not be be station-based (with GPS and smart-phone apps), though such as system has yet to be deployed in Minnesota. So there are continued opportunities for growth with the current technological arrangements. 

  

Figure 4.3: NiceRide’s Service Area continues to grow (source NiceRide MN, 2015)

4.4      The Future of Sharing: Cloud Commuting

For communities where densities are relatively low, incomes high, and thus taxis scarce, the most reliable strategy for timely point-to-point transport is for people to maintain personal transport close at hand. Cars and bikes, which they own, are parked at their homes, workplaces, or other destinations. But with more widespread use of information technologies, ownership and possession are no longer necessary prerequisites for on-demand mobility. Widely called the "sharing economy" or "collaborative consumption", its manifestations in transport: "carsharing" and "ridesharing" are viable if not widespread.  Couple these technologies with autonomous vehicles discussed Chapter 1, and one arrives at "cloud commuting" — the convergence of ridesharing, carsharing, and autonomous vehicles.^[38]

Figure 4.4: NiceRide’s Annual Revenue continues to grow (source NiceRide MN, 2015)

More formally, this range of options can be termed Mobility-as-a-Service (MaaS). While nascent today, clearly big players are placing big bets that this will be a big change in how people travel. It is this which explains Uber’s high market valuation.   A vehicle from a giant pool of autonomous cars operated by organizations based "in the cloud" would be dispatched to a customer on-demand and in short order, and then would deliver the customer to her destination (be it work or otherwise) before moving on to the next customer. Even more efficiently, it might pick-up or drop-off some additional passengers along the way and may offer customer specific features.^{[39] [40]}

We quickly run down implications as MaaS emerges.  

Smaller, more modern fleet. The customer benefits by not tying up her capital in vehicles, nor having to worry about maintaining or fueling vehicles. The fleet is used more efficiently, each vehicle would operate at least 2 times (and as much as 10 times) more distance per year than current vehicles, so the fleet would turnover faster and stay more modern. 

Fewer vehicles overall would be needed at a given time. It is likely customers would need to pay for this service either as a subscription or a per-use basis. Advertising might offset some costs, but probably not cover them. However retail stores (if they survive) might subsidize transport, as might employers, as benefits for customers or staff.

Coverage, logistics. Like traditional fixed-route transit, MaaS will function better in urban areas than rural areas. Response time will be shorter (potentially faster than getting a parked car); size and variety of the vehicle pool will be greater; parking in high value areas becomes less troublesome. MaaS will also fit better for nonwork rather than work trips, as the regularity of work increases the value of either vehicle ownership or regularly using micro- or macro-transit versus renting by the trip. 

Autonomy solves the localness problem facing existing carsharing services, since the cars come to you. Like current bike sharing systems, there would need to be load balancing features, so the cars were pre-dispatched to areas of anticipated demand, and maybe coordinated carpooling at peak times.

Costs. Automation also structurally transforms the labor costs of ridesharing services.^[41] It allows a variety of vehicles to serve customers, rather than a single, literally least common denominator model. An interesting aspect of this from the perspective of travel demand is that with MaaS, people will probably pay by the trip, either directly, or through choosing the right plan of service roughly proportional to use. While the average cost of car ownership, now a quite significant share of household expenses, goes to zero for those who join this system, the out-of-pocket marginal cost per trip rises quite significantly. The implication is fewer trips for people give up on vehicle ownership. People paying by the minute or the mile will want to reduce trip distances.

Electrification. Autonomous and shared vehicles will interact with electrification discussed in Chapter 6. A rental service of self-driving autonomous vehicles, that are ordered on-demand may provide you a fully charged electric vehicle for your trip. Much like the Pony Express, which swapped horses rather than requiring riders to wait for their own horse to rest, the service may provide a replacement vehicle mid-trip rather than requiring you to wait around to charge your vehicle.^[42]  There is no requirement that cloud-based, self-driving vehicles be electric, but as cars get smarter they should be able to charge themselves, alleviating some of the concerns associated with EVs.

Street Design. Streets designs will need to accommodate pick-up and drop-off as a major feature, so curb space will need to be re-arranged so people know where to meet their car (and vice versa), so they don’t get into the wrong white Prius.  While we lose the need for parking, we might think of channelizing roads more like airports or multi-way boulevards than the monolithic pavement they are today.

4.5      The Future of Travel Demand and Where We Live: The Up Scenario

In Chapter 1 on Autonomous Vehicles, we described the “Out Scenario”. Privately held autonomous vehicles which are faster and reduce the cognitive burden on drivers will increase suburbanization. In contrast, MaaS eliminates the fixed cost of transportation, and exchanges it for a higher per cost trip. Logically, if the time or money cost per trip rises, there should be fewer and shorter trips, or reduced demand.

The share of ownership versus MaaS is thus an important predictor of travel demand in the coming years.

Up: Less vehicle ownership with increased use of MaaS in cities, raising the value of cities. Driverless cars which can be summoned on demand allow people to avoid vehicle ownership altogether. This will reduce vehicle travel, as people will pay more to rent by the minute than they do when they own. Since total expenditures on transport are saved, additional funds are available to pay for rent in cities, and more trips are by walk, bike, and transit. People who seek the set of urban amenities (entertainment, restaurants, a larger dating pool) will find these amenities increasing in response to the population. The greater value in cities with the new more convenient technology leads to more and taller development. (Hence the use of the word “Up”.)

A direct knock-on effect of the “Up” Scenario is that it will transform the need for parking. It will also means vehicular dead-heading (cars driving without any occupants to their next call), a phenomenon we now have with transit and rail, though even those vehicles usually have drivers.

At the more local, urban level, the Mobility-as-a-Service model implies spaces now devoted to cars can be repurposed—everything from street space to buildings. Garages turn to accessory dwelling units. Gas stations and parking lots to anything with a "higher and better use". Autonomous vehicles can drop off their passenger at the front door, and then park themselves in far less space than drivers currently require (or move on to their next passenger), and that space need not be so close to the most valuable urban areas. On-street parking is not needed at all, one more aspect of roadspace reconfiguration.

As David King says “Redeveloping surface parking may be the single biggest challenge facing US cities, but is also a rare opportunity for cities and many suburbs to rebuild themselves fairly easily and quickly.”

           

4.6       Discussion

While we might think of “Up” and “Out” (described in Chapter 1) as contrasting scenarios, they are not exclusive. More people living in the suburbs or exurbs does not mean fewer people live in cities, so long as there are more people overall.  We expect more growth will be either central or exurban, and less in existing low-density suburbs which cannot effective offer MaaS nor fully exploit the wide open spaces of the exurbs.

The model of vehicle ownership is being challenged with alternative ways of providing transportation services enabled by mobile telecommunications. Broadly these Mobility-as-a-service (MaaS) technologies can up-end the current model of transportation ownership. Enabled by technologies such as social networks with known identities, real-time ride matching, excess seat capacity in passenger vehicle fleet, to be successful requires critical mass. The full extent of the markets for these services is not yet clear. There are implications for transit, which might perceive MaaS as poaching of customers. 

We posit that paying for trips by marginal rather than average cost will reduce number of trips and the availability of such services will reduce number owned vehicles. Social relationships will also change with these new technologies.  “Sharing” models include: Car Sharing/Rental (e.g. Car2Go, ZipCar, HourCar, Enterprise Car Sharing), Peer-to-Peer Car Sharing (e.g. Relay Rides), and Bike Sharing/Rental (e.g. NiceRide). “Ridesharing” Transportation Network Companies include services like Lyft and UberX, but are also associated with the changing nature of taxi services. Fully dynamic ridesharing (peer-to-peer transportation) is the modern phone-driven analog of hitchhiking.

Sharing—be it cars, bikes, boats or information—has strong network effects driven by convenience (a characteristic the time-starved seem particularly mindful of). But, macro versus micro transit discussions in the following chapter bring up matters of economies of scale versus economies of scope. There’s a role for both.

For example, one is more likely to use carsharing if more neighbors use it, since that makes it more likely there will be a car in front of one's house, workplace, or wherever, when it is desired. Reducing vehicle access time from 10 minutes to 5 minutes, or 5 minutes to 2 minutes is significant, especially when most trips are only 20 minutes long. As with any social network, it is not clear in advance which if any will take off. As with many networks, there needs to be a large up-front capital investment. But unlike transit systems, carsharing is dealing mostly with mobile capital. If the program doesn't work in place A, cars can be redeployed to place B, or at worst, sold in a used car lot. Further the programs are privately funded, which is more suited to innovation and adaptation, and accepting of failure, than publicly funded transit agencies.

The Car2Go model has not yet put in enough capital, nor has enough demand, so there is a car waiting on every block. To do so is no small step, and may require automation.^[43]

The economics and environmental benefits ^[44] of renting rather than owning are clear, but the sociology and the the role of regulation^[45] remain unclear.  People willingly use hotel rooms, or bikes, or library books that have been used by others before, but not, typically, cars. How do cars get transformed from an owned good to a rented service? In part this is generational. If you have never owned a car, new habits can be formed. But that type of change is very slow, perhaps as slow as generational shift. Early adopters and the carless may be quick to join. Some/many/most Americans use their cars often enough, in places remote enough, or customize their cars sufficiently that MaaS will not be advantageous in most circumstances. The question is: What is the winning fraction?

We are hesitant to give an answer. 

Loosely, environments which are currently dominated by multi-family living (apartments) are more likely to be places where Mobility-as-a-Service succeeds. Densities are higher, enabling various types of shared services to have shorter response times, and making land values higher so car storage is more expensive.

Nationally 35% of the US population are renters and 14% are apartment dwellers. In Minnesota only 11% of the population are apartment dwellers.^[46] To be clear today, not all apartment dwellers are transit riders or walk to work (about 20% do), and not all single-family dwellers use the automobile to get to work (though 96% do). Any change from a vehicle ownership-dominated to a MaaS regime will take decades, and likely be slower than the introduction of automation in the first place.

           

Chapter 5: Quantified Self and Quantified Networks

5.1       Introduction

There are now sensors measuring, tracking, and reporting travelers’ location, pollution, noise, travel cost [e.g. in-vehicle taxi meters], health, calories, and so on. As people are more interested in self-tracking their daily behaviors (i.e., the increasingly popular Quantified Self movement), transportation agencies have a new wealth of data that can monitor the movement of people and vehicles across networks. These will enable advanced traffic, air quality, and new security monitoring systems, among other uses. How can agencies exploit these data? How will the “quantified self” and “quantified networks” interact?

To explore answers to these questions, I first review existing smartphone-based mobility sensing and quantification apps. Much of the Quantified Self Movement is associated with the advance of the smartphone technology. Innovative mobility apps have the power to transform the relationship between transportation networks and travelers. The Future of Transportation series produced by CityLab (formerly The Atlantic Cities) has concluded that the smart phone is the most important transportation innovation of the decade.  There has been an explosion of mobility apps and interfaces (e.g., RideScout, Google Maps, OMG Transit, Waze, and Uber) that help people to make more informed mode and route decisions. These apps offer functionalities ranging from trip planning and navigation, to locating an approaching bus or for-hire vehicle. Of these apps, this report focuses on apps that are specifically designed for travel data collection (i.e., sensing and quantification). Besides reviewing existing smartphone-based mobility sensing and quantification apps, this report discusses the potential application of these apps in travel behavior intervention and mobility management, followed by discussing potential areas for future research when it comes to possible integrations between Quantified Self and Quantified Networks technologies. 

5.2       Smartphone-Based Mobility Sensing and Quantification Apps

Deriving travel information from mobility sensing data (especially location sensing data) involves significant processing due to the large amounts of data produced by GPS units/loggers or smartphones over time.  Based on the current literature (Flamm & Kaufmann, 2007; Gong et al., 2014; Schuessler & Axhausen, 2009; Tsui & Shalaby, 2006; Wan & Lin, 2013), Figure 5.1 overviews the steps when using GPS sensing data (either smartphone-based or non-smartphone based) to identify and quantify travel characteristics. In general, the selection of data filtration techniques depends greatly on the detail of the analysis to be conducted and the quality of GPS data being used. Speed, distance, and time were the most commonly used data filters. For activity/ trip identification, the use of speed, time, and spatial density were amongst the most common methods used. While studies used different methods for mode detection, only Zheng et al. (2008) compared the detection capabilities of different models and concluded that decision tree models outperformed others. Trip purpose detection was found to rely heavily on the availability of accurate geographic information (land use or points of interest data), the quality of which varies from place to place.  

Figure 5.1: Major Quantification Steps for Analyzing Mobility Sensing Data

The steps illustrated in Figure 5.1 formed the foundation for data processing algorithms in the smartphone mobility sensing and quantification apps listed below: 

        The earliest smart phone app for travel data collection is the MoALS system developed by Itsubo and Hato in 2006, requiring users to input the start time, destination and mode of trips; it records GPS data in the background (Itsubo and Hato, 2006). The data collected by the app is transmitted to a server and can be displayed on a website. Users are also required to login to the website every day and confirm the data collected by the app.

        TRAC-IT is a smartphone app developed by the University of South Florida to better understand household travel behaviors and provide feedbacks to users (Winters et al, 2008). TRAC-IT has two modes to collect travel data: It either logs GPS data continuously in the background or allows users to initiate and terminate data logging for a trip. Meanwhile, the app asks users to register trips and provide information on attributes that cannot be extracted directly from GPS data. For example, the app asks users questions about trip purposes and trip modes. After trip-related information is collected, it transmits the collected data to a remote server for data processing and retrieves travel advisory feedbacks from the server.

        CycleTracks is an app developed by the San Francisco County Transportation Authority (SFCTA) to collect bicycle use data in the area. It utilizes the GPS sensor in smartphones to record information on users' bicycle routes and times, and displays their rides on maps. The app allows users to register bicycle trips, initiate GPS tracking, and specify the trip purpose of each bicycle trip (SFCTA, 2015).

        Future Mobility Survey (FMS) is a smartphone based travel survey mainly deployed in Singapore as a part of the nationwide Singaporean Household Interview Travel Survey (HITS). To minimize user’s burden, the app is only used to collect spatiotemporal information of travels while providing a web interface to collect travel attribute data such as trip purposes, trip modes and trip satisfaction. In addition, the web interface is also used as a portal for users to input demographic information, review travel data and provide feedbacks (Cottrill et al., 2013).

     Moves is a commercial app developed by ProtoGeo to collect all-day non-motorized travel (including walking, running and cycling) data on smartphones. It logs GPS and accelerometer data continuously, and automatically identifies trips and non-motorized modes used in any of the trips. The app also allows users to tag activity locations using three location types (home, work, or school) or place names available from Foursquare API (https://developer.foursquare.com/). In addition, Moves has a pedometer function that counts the daily number of steps that users take and calculates daily calories burned (ProtoGeo, 2013).

     ATLAS, standing for “Advanced Travel Logging Application for Smartphones”, is an application designed specifically for collecting travel data by the University of Queensland (Safi et al., 2013). Its goal is to provide a user-friendly and convenient interface for users to record travel data with minimum burden and maximum accuracy. The app provides functions to record location data from the GPS sensor and trip attribute data such as purposes and modes. It requires users to initiate and terminate the tracking of each trip. Besides, it also collects basic socio-demographic information such as age, gender and car ownership on the smartphone. In a later study, ATLAS was utilized in a Smartphone-based Individual Travel Survey System (SITSS), which was deployed as a part of the national household travel survey of New Zealand (Safi et al., 2015). In SITSS, it was developed to be a two-step data collection approach during which the app first logs GPS data continuously and then invites participants to record their travels at the end of each day. The app automatically detects movements longer than 400m and starts the travel recording; it stops recording automatically when the device becomes stationary for longer than 360 seconds.

     CONNECT is an apps developed by Ghent University to collect travel data. It provides functions for users to register trips, initiate GPS and accelerometer tracking, and specify trip characteristics. It also allows automatic triggering of specific surveys based upon user input. For example, questions related to bicycle trips pop up when users register such trips (Vlassenroot et al., 2014).

     Quantified Traveler (QT) developed by the University of California, Berkeley is aiming to record travel data and provide travel suggestions that aim to trigger more sustainable travel behaviors while satisfying travelers’ preferences (Jariyasunant, et al., 2015). The QT logs GPS and accelerometer data continuously in the background, transmits location and accelerometer data to a server in the cloud for post processing, and periodically requests the user to access feedbacks through a website that displays post-processing results from the server, including travel cost and calculated footprints resulted from a user’s travel: the calories burnt and CO2 spent traveling.

     SmarTrAC is an app developed by the University of Minnesota based on a previous app UbiActive (Fan et al., 2012). It logs GPS and accelerometer data in the background, and automatically identifies trips, as well as travel modes and trip purposes. The app provides users immediate read and write access to results from real-time detection of travel modes and activity types. The app also allows users to add additional details about their activities and trips, such as companion information and their emotional experience. SmarTrAC distinguishes itself from other apps in three aspects: first, real-time and highly-accurate detection of travel mode (an overall accuracy of 96% in classifying motorized vs. non-motorized trip segments and an overall accuracy of 86% in classifying travel modes across six mode options including car, bus, rail, wait, bike, and walk) and trip purposes (an accuracy larger than 95% in detecting trip purposes when trips taking place at previously identified location); second, on-the-fly visualization and annotation; and third, continuous self-improvement of its detection modules (Fan et al., 2015). 5.3        Quantified Self and the Potential of Travel Behavior Intervention

Research has shown that generic mass media intervention can be used in raising the awareness of the health benefits associated with sustainable transportation (e.g., the U.S. Walk to School programs) and in increasing walking and biking mode share (Ward et al., 2007). However, critics question the effectiveness of generic information in bringing about sustained behavioral change. It is suggested that interventions that are closely tailored to individual needs may include less redundant information and are more likely to be read, saved, remembered, and discussed (Smeets et al., 2008). With increasingly quantified travel behavior, many of the travel quantification apps can be designed as smartphone-based behavior intervention tool for promoting travel mode shifts from driving to more sustainable modes.  For example, building upon existing apps’ data collection capabilities, intervention-driven tools can be designed to provide customized messages and action plans to the user after detection of each driving trip. The tool can incorporate a combination of three strategies to promote mode shifts:

          Awareness: Messages describing environmental impacts associated with each driving trip (e.g., carbon emissions) could be displayed to the user. 

          Motivation: Messages describing personal benefits of a mode shift (e.g., cost savings and health benefits) could be displayed to the user to reinforce positive aspects of transit and nonmotorized travel. 

          Action: Implementable mode-shift plans could be provided to the user through the phone. For example, if the application detects a driving trip made from home to a grocery store, the tool could utilize maps of bike rental facilities and public transit services to provide information on how to travel to the destination by alternative transportation modes, including information on where and how to rent a bike or board a bus/train as well as information on the best walking/biking/transit routes.

There is no shortage of health-oriented apps (e.g., Moves and Apple HealthKit) that could promote travel-related physical activity, such as walking and biking. Yet, neither set of apps can effectively and systematically promote travel mode shifts from driving to more sustainable modes—they were simply not designed for that purpose. Nonetheless, the success of health-oriented apps provides important insights into the potential of travel quantification apps in behavior intervention. Several previous studies have evaluated the use of mobile phones to support healthcare and public health interventions (Kailas et al., 2010; Boulos et al., 2011; Dennison et al., 2013). There is consensus that smartphones have the following advantages for delivering of health-promoting behavioral interventions (Dennison et al., 2013):

        Portable devices offer the opportunity to bring behavioral interventions into important real life contexts where people make decisions about their health and encounter barriers to behavior change. 

        Smartphone apps may provide cheaper, more convenient, or less stigmatizing interventions that are unavailable elsewhere.

        The connectedness of smartphones facilitates the sharing of behavioral and health data with health professionals or peers.

        Smartphones enables continuous and automated tracking of health-related behaviors and timely, tailored interventions for specific contexts.

These same advantages could apply to smartphone apps that aims to promote travel sustainability. The 2015 Urban Mobility Scorecard shows that urban areas of all sizes are experiencing increased congestions (Schrank et al., 2015). The Twin Cities region has worse congestion than ever before. As of 2014, the region’s congestion problem resulted in 99.7 million person-hours of annual travel delay and 38.5 million gallons of annual excess fuel consumption. The smartphone-based quantified-self technologies offer opportunities to develop efficient, low-cost, and innovative approaches to collecting detailed travel behavior data as well as promoting sustainable travel behavior.

5.4       Discussion on Quantified Self and Quantified Networks

Compared to mobility-related Quantified Self technologies, roadway-based vehicle detection and surveillance technologies are relatively mature technologies that have been used to support traffic management and traveler information services. The roadway based sensors have been used across transportation networks to provide enhanced speed monitoring, traffic counting, presence detection, headway measurement, vehicle classification, and weigh-in-motion data (i.e., to provide quantified network data). Mimbela and Klein (2007) categorized roadway-based sensors into two major groups:

 In-roadway sensors: Sensors are either embedded in the pavement/subgrade of a roadway or tapped/attached to the surface of the roadway. Examples include inductive loop detectors, weigh-in-motion sensors, magnetometers, pneumatic road tubes, etc. 

 Over-roadway sensors: Sensors are mounted above the surface of the roadway either above the roadway itself or alongside the roadway (offset from the nearest traffic lane by some distance). Examples include video image processors, microwave radar sensors, ultrasonic sensors, acoustic sensors, iBeacons/Bluetooth transmitters, etc. 

To date, roadway-based sensor technologies have only limited interactions with smartphone-based sensor technologies. Yet, the potential for the two types of technologies to interact is strong: 

 Roadway sensors can provide time-dynamic data on transportation networks that can help to improve data processing modules to derive more accurate travel behavior information from smartphone sensor data. For example, when it comes to identifying activities vs. trips using smartphone sensor data, existing technology heavily depends on speed. As a result, trips in congested roadways or encountering an accident are likely to be identified as activities. Roadway sensors that can monitor traffic accidents and traffic congestion could be used to improve trip identification.  In addition, new sensors that employ iBeacons/Bluetooth technology can provide more accurate location and proximity data than smartphone built-in location sensors. Such highly accurate location identification technology can help disadvantaged people such as the visually impaired to navigate transportation systems. 

 Smartphone sensors can capture dimension of data that current roadway sensors have difficulties in capturing such as acceleration patterns. Combining smartphone sensor data and roadway sensor data will allow transportation agencies to better understand traffic patterns and generate better data to support traffic management.

The following issues merit discussion when it comes to possible integrations between quantified self and quantified networks technologies: 

 Smartphone sensors are sensitive to environmental conditions such as heavy rain/snow, urban canyon, extreme hot or cold temperature, etc. Although in-roadway sensors have relatively consistent performance across environmental conditions, many of the overroadway sensors are sensitive to various environmental conditions. How the smartphone sensors and roadway sensors can be integrated to provide robust data across extreme environmental conditions merits future research. 

 Most of the roadway sensors focus on detecting motor vehicles. Although there has been substantial progress made in roadway sensing technologies for detecting pedestrian and bicycle traffic, such technologies are much less mature. Smartphone technologies are especially good at detecting non-motorized modes, yet relatively weak to distinguish between private car and public transit especially in urban traffic without using public transit infrastructure data. How the two sets of technologies be integrated to generate multi-modal data merits future research. 

 One fundamental concern about smartphone apps for mobility data capture is the battery power consumption. Since the battery power depletes rapidly from frequent and continuous operation of multiple sensors, especially the GPS receiver, the inconvenience caused by frequent battery recharge often drives away users and offsets the potential benefits of using those apps. When properly integrated, roadway sensing technologies (e.g., transiting location/speed data to smartphones so that smartphones need not to use GPS receiver to capture location/speed data) may offer opportunities for smartphone apps to preserve battery power. 

 The usage of some roadway sensors (especially in-roadway sensors) tend to be expensive as their installation and maintenance often require pavement cut and/or lane closure. With sufficient user base, smartphone data may have the potential to eliminate the needs for more expensive roadway sensors. Companies such as AirSage have explored ways to aggregate signaling data from cellular networks to provide real-time speed and travel times for major roads. How smartphone data may replace/substitute part of the roadway sensor data merits future research. 

 Privacy will be a major concern for possible integration between Quantified Self and Quantified Networks technology since few people would want to be tracked in their daily life even in cases when the collected data is used for research. Who will own the data? Who can use the data? What type of smartphone and roadway data can be integrated for future analysis?  Data security and privacy protection issues are critical issues meriting future research. 

           

Chapter 6: Electrification and Alternative Fuels

6.1       Introduction

The current mix of fuels to power on-road vehicles is diversifying at an increasing rate. Desires to reduce greenhouse gas emissions, noxious emissions, fuel use, foreign oil consumption and cost of driving have led to a new era of power use for transportation. While gasoline remains the dominant fuel for light duty vehicles and diesel the dominant fuel for heavy duty vehicles in Minnesota and the US, an increasing percentage of vehicle power is being derived from “alternative” transportation fuels, such as biofuels, natural gas and electricity. The largest use of alternative fuels in the State of Minnesota is a result of legislative mandates that require fuel retailers to blend quantities of biofuels with traditional fossil-derived transportation fuels. The “drop-in” replacement fuels do not require a specialized fleet of vehicles to use these fuels and thus do not drastically affect the type of vehicles on the roadways. Other alternative fuels, such as electricity or natural gas, require specialized powertrains that differ from conventional gasoline and diesel-powered vehicles. This report refers to both drop-in replacement fuels and fuels requiring specific drive trains as alternative fuels.

Issues with alternative fuel use that are relevant for the Minnesota Department of

Transportation include,

1.      Loss of tax revenue for roads as result of switching to fuels that do not have highway tax

2.      Change of vehicle mass as a result of alternative vehicle power train

3.      Change in vehicle emissions and resulting air quality (mobile emissions sources vs.

point sources)

4.      Change in vehicle activity and ownership costs

5.      Refueling and charging infrastructure

6.      Travel times and robustness

7.      Increased difference in weight between heavy duty and light duty vehicles as a result of power train and emphasis on fuel efficiency

Important drivers of the impacts of alternative fuel use are the degree to which alternative fuels are being adopted in on-road vehicles in the Minnesota transportation fleet, whereby the majority of fuel shifting has resulted from increased use of ethanol in light duty vehicles (displacing gasoline), increased use of biodiesel in heavy goods vehicles (displacing diesel), increased use of electricity as a primary power source, and increased use of natural gas (compressed and liquefied) as a fuel. 

This report discusses the current and likely future trends of alternative fuel use in the State of

Minnesota and the subsequent impact of alternative fuel use on issues relevant to the Minnesota Department of Transportation for each of the major alternative fuels being adopted within Minnesota.

6.2    Alternative Fuel Vehicle Trends

The adoption of alternative fuels differs greatly by the type and availability of alternative fuel. The following sections discuss trends of the major fuels that are achieving penetration in the Minnesota vehicle fleet.

6.3       Drop-in Biofuels

Drop-in replacement biofuels, such as ethanol and biodiesel have grown rapidly since their first adoption in the early 1990s. The national consumption of ethanol and biodiesel has undergone near exponential growth since the year 2000 as a result of a mix of mandated blend levels from states and federal government, as well as incentives. 

The stated aims by the Minnesota Department of Agriculture for ethanol and biodiesel production and use within the State of Minnesota are to provide a new market for agricultural products, displace fossil fuels and help meet the US EPA standards for carbon monoxide within the Minneapolis-St. Paul metropolitan area. To achieve these aims, the State of Minnesota has adopted a series of legislative acts to encourage ethanol production and consumption beginning in 1980 with legislation that offered a 4 cent per gallon pump tax credit for 10% ethanol blends. By 1986, forty percent of the state’s gasoline was blended with 10% ethanol as concerns over other oxygenates (MTBE) increased. Further legislation reduced the pump tax credit to 2 cents and initiated a 20 cent per gallon incentive payment for ethanol produced in the state. In 1992, a minimum 2.7% oxygen content requirement was established for gasoline. It was made effective for the entire year (as opposed to just summer months when air quality is worse) in the Minneapolis St. Paul metropolitan area in 1995 and then statewide in 1997. 

In 1994, the pump tax credit was phased out while oxygen requirements for fuel statewide were phased in. In 1995, a statutory goal to develop 220 million gallons of Minnesota ethanol production was established, which was further expanded to 240 million gallons in 1998. By 2000 all other oxygenates (MTBE) were effectively eliminated, leaving ethanol as the only oxygenate allowed in attempt to reduce groundwater contamination while controlling unburned hydrocarbon and carbon monoxide production. Producer payments were reduced to 13 cents per gallon for fiscal years 2004 through 2007. The ethanol production goal was to increase to 480 million gallons by 2008 and the 2.7% oxygenate requirement (equivalent to a 7.4% volume of ethanol) for gasoline was replaced by 10% ethanol requirement (3.65 % oxygenate). 

In 2005, law required a 20% ethanol content or a maximum of what the EPA allowed (known as the blend wall) in all gasoline by 2013, which had been delayed to start by August 30, 2015 were the EPA to have approved higher blend levels (E15 or E20). The EPA did extend the blend level to E15 for 2001 and newer vehicles, but Minnesota has not mandated E15 statewide due to the restriction on older vehicles. There are currently 36 E15 retail locations in Minnesota in addition to the 263 station that sell E30 or E85 that are available exclusively for flex-fuel vehicles^[47]. The pressure to increase the fraction of ethanol within gasoline has diminished as the US has increased domestic oil production, lessening the demand for foreign oil [1]. An ongoing debate continues as to whether ethanol blend levels should be raised, as concerns over the energy and environmental implications of corn-based ethanol production are being investigated [2-4]. 

As a result of these incentives nearly 20% of Minnesota’s corn crop is converted to ethanol in Minnesota’s 21 refineries with a nameplate production capacity of over 1 billion gallons (76 trillion BTUs) of ethanol each year, which allows 10% of Minnesota gasoline to be replaced by ethanol. The shipments of agricultural commodities throughout the state are affected by these programs, and resulting road use in rural areas differs as more agricultural products are trucked short distances from nearby farms to local refineries, rather than placed on rail and shipped out of state to foreign markets. 

Similar legislation has affected the quantity of biodiesel produced and consumed within the State of Minnesota. In 2005 Minnesota law required that a minimum of 2% biodiesel be blended in all diesel fuel sold in the state for on-road purposes (thus excluding locomotives and mining equipment). In 2009 the required percentage of biodiesel in Minnesota increased to 5% and in 2014 increased again to 10% biodiesel. Currently the 10% biodiesel mandate is achieved by a 15% biodiesel blend level in summer months (April through September) and then a 5% blend level in winter months. Beginning in 2018 the summer blend level will increase to 20% and the winter level will remain at 5%. As a result of the mandates, the State of Minnesota produces nearly all of its biodiesel within the state, which accounts for nearly 13% of the state soybean crop. [5]

Minnesota leads the nation in terms of biofuel use, but national policy has also increased domestic biofuel use. The primary drivers for biofuel use at the national level have been legislation enacted that requires the amount of biofuels produced and blended to achieve specific targets by given dates. The Renewable Fuel Standard (RFS) was first enacted in 2005 and required that 7.5 billion gallons of renewable fuels (primarily achieved with corn ethanol) be blended into gasoline by 2012. The Energy Independence and Security Act (EISA) of 2007 expanded the RFS program, now called RFS2, to blend higher levels of fuel increasing from 9 billion gallons in 2008 to 36 billion gallons by 2022 [6]. The EISA also distinguished renewable fuels by their lifecycle greenhouse gas (GHG) impact, as a primary driver for the increased production of renewable fuels was to lower the GHG emissions associated with fuel use. RFS2 required that both renewable diesel and gasoline alternatives achieve specific GHG reductions relative to the displaced fuel. Since the adoption of the standard, the EPA has made multiple modifications to the program as production of ethanol from non-corn sources, called advanced biofuels and cellulosic biofuels, have not been achieved thus forcing the EPA to reduce the required amount of advanced and cellulosic biofuel use.

Historically, domestic ethanol production has been protected by tariffs on foreign ethanol imports. A 54 cent per gallon import tariff effectively prohibited importation of Brazilian sugarcane ethanol, which is more efficient to refine than corn. The foreign tariff and 45 cent per gallon tax credit to blenders were eliminated in 2012 allowing open access to the US market for foreign biofuel producers.

Figure 6.1 shows the quantity of ethanol and biodiesel consumed within the US from 1980 to

2014. Over the same period total fossil fuel consumption increased from 19,000 to 24,826 Trillion BTUs [7], resulting in a total biofuel consumption of 5.6% for transportation activities in 2014. As seen the consumption of biofuels increased dramatically in 2000 as a result of legislation nationally and in individual states to encourage biofuel production. The levels of ethanol consumption have plateaued since 2010 as a result of an inability to meet RFS2 targets with low GHG ethanol, and increased concerns over the environmental performance of corn-based ethanol.  Biodiesel consumption has continued to increase since 2010 due to greater demand and less concern of environmental performance. The lifecycle GHG emissions associated from biodiesel are less than half (≈76% reduction) of those associated with diesel fuel use even when accounting for land use changes [8]. As a result the total biodiesel fuel content continues to increase but will likely plateau in the future due to cold weather blending limitations and lower yields of biodiesel per acre when compared to corn or cellulosic ethanol, see Table 6.1. 

Table 6.1: Ethanol and biodiesel yield per acre from selected crops. [9]

           

Standard 2^nd Implementation [10] and revised by the EPA in 2014 [11].

National levels of biofuel consumption are likely to plateau near the current levels of consumption, as shown in Figure 6.2. The Energy Information Administration historic [7] and projected [12] levels of biofuel consumption within the transportation and liquid fuels sector indicate that the US is near the peak of biofuel consumption at 6.3% of total fuel use. The production of biofuels is projected to increase, but will remain stable relative to the consumption of total fuel use, e.g. both total fuel use and biofuel production will increase proportionally. Dynamics are likely to occur within the mix of biofuels produced and consumed. As technology develops and RFS2 mandates incentivize greater production of fuels from non-food sources, such as cellulose and crop residues, there may be larger sources of advanced and cellulosic fuels available. Long term pressures for reduced greenhouse gas emissions and sustainable agricultural practices may present a competition in supply of biomass products, as the electrical sector seeks to decarbonize coal-fired electric power production with increasing fraction of biomass co-firing. The relative levels of biomass that are dedicated to electricity versus liquid transportation fuel production are likely to be a result of government incentives, suitable low carbon substitutes and the evolution of the transportation fleet from liquid-fueled vehicles to vehicles powered by electricity and other energy carriers.

Figure 6.2: Historic and projected biofuel consumption of total fuel use where blue dots represent historic data [7] and black dots represent projections by the Energy Information

Administration [12] and the black line represents predicted percentage of biofuel use as predicted by S-Curve (S(t) = K/[1+exp(-b(t-t0)], K=0.063, b=0.38, t₀=2008, R²=0.9904)

6.4       Electric Vehicles

Vehicles powered by electricity offer an alternative to liquid fueled vehicles and have received increased attention over the last decade. The primary drivers of electric vehicle technology are the desire to reduce noxious emissions in urban areas and greenhouse gas emissions globally. As shown in Figure 6.3, the sales of electric drives has increased steadily since 2000 with hybridized electric vehicles (HEVs) reaching roughly 6% of the new car vehicle sales market. While HEVs derive all of their power from liquid fuels, fully electric vehicles (EVs) and plug-in hybrid electric vehicles (PHEVs) are able to derive at least a portion of their electricity from the power grid, thus displacing liquid transportation fuel use. PHEVs and EVs are relatively newer technologies within the market and are still undergoing development by many manufacturers with most offerings on the market representing the initial model available from the manufacturer. The split in sales between EVs and PHEVs is roughly equal with 152,000 EVs and 164,000 PHEVs sold in the US market from 2011 to April 2015 [13]. Globally the number of vehicles with electric drives within the fleet (excluding non-plug in hybrids) has been doubling every year for the past three years, and could reach 1 million vehicles globally if trends continue by 2016^[48]. 

Figure 6.3: Total US new car sales from 1990 to 2015 [14], as well as percentage of hybrid electric vehicles (HEV) and combined electric vehicle and plug-in hybrid electric vehicle sales [13].

In Minnesota the trends for HEV, EV and PHEV sales have lagged those of the national economy, as the share of Minnesota vehicle sales with hybrid drives has plateaued at 2 to 3.5% since 2009, as shown in Figure 6.4. The adoption of electric vehicle drives has been delayed as a result of higher prices for vehicles, cold winter-time temperatures which lowers electric vehicle range and a lack of state incentives that encourage purchase of electric drive vehicles. Also, corporate average fuel economy (CAFE) standards have increased the fuel efficiency of non-hybrid vehicles lowering the gains of hybrid drives relative to traditional vehicles. The low fuel prices of 2014 and 2015 are likely to keep the sales of hybrid drive vehicles low. Long-term increases in sales of electric drive vehicles are likely to be driven by national incentives and regulation seeking to reduce GHG emissions.

Figure 6.4: New car sales and percentage of vehicles with electric drives (HEVs, PHEVs and EVs) in Minnesota from 2008 to 2014 [15].

6.4.1 Powering Electric Vehicles 

The primary inhibitor to electric vehicle penetration is the size and cost of the battery required to achieve a range that is acceptable to consumers. The industry remains divided on how to provide a vehicle with the typical 200 to 300 mile range that is expected by consumers, with nearly half of the manufacturers providing PHEVs which have a small battery capacity able to deliver a portion of the drive in all electric mode (typically, 20-50 miles) and others providing full EVs with larger battery packs able to deliver a total range of greater than 100 miles. As the electrical storage capacity of vehicles increases, the weight and cost of the vehicle rise proportionally. However, by converting to full EV rather than PHEV, the manufacturer is able to eliminate duplicate engine and motor systems lowering manufacturing costs. 

The optimal size of full electric capacity was studied for PHEVs and EVs and it was found that for the average US commuting distance a PHEV with an all-electric range of 20 miles optimizes vehicle cost, gasoline consumption and greenhouse gas emissions [16]. As prices fall and the weight of batteries decrease (energy densities rise), increased electrical capacity is optimal and more cost effective. A study of consumers’ willingness to pay for EV attributes found that consumers require pay back periods of 5 years to offset the initial EV cost premium. Battery prices of $300-350 per kWh are required to achieve a 5 year payback with the current $7,500 federal tax credit [17]. While $300 per kWh battery costs are below the current industry average, a recent study found that the that EV battery costs are declining by 14% annually, from above $1,000 per kWh nearly a decade ago to around $410 per kWh in 2014, and that the cost of battery packs used by market-leading EV manufacturers are even lower, at $300 per kWh [18]. 

As PHEVs and EVs become more prevalent, the number of charging stations must increase to allow for reliable charging throughout the transportation network. Currently 10,409 public charging stations are available within the US (in comparison there are 121,000 gas stations) of which 192 are located in Minnesota [19]. For full coverage the State of Minnesota would need to have charging stations roughly every 100 miles, along with at-home charging. While the

Minneapolis-St. Paul metropolitan area has a high density of stations, outstate areas require higher density to allow full access to all state locations with EVs. 

6.4.2 Electric Vehicle Penetration

The electric vehicle penetration rates for the light duty vehicle fleet were estimated by fitting Scurves to the adoption rate of HEVs and combined PHEV and EVs in the US fleet. The historic HEV trends provided sufficient data to fit an S-curve to and it was assumed that EV and PHEV adoption would follow a similar adoption rate delay by 10 years. As shown in Figure 6.5, the adoption of HEVs in the US fleet is expected to grow until it reached 30% market share by 2035 and then decrease thereafter as PHEV and EVs begin to eclipse HEV sales in 2037. The growth of PHEV and EV sales nationally are expected to increase from their current levels in 2014 to achieve a 24% new car sales penetration in 2035 and 68% penetration in 2050. The penetration of PHEV and EV sales in the State of Minnesota is expected to lag the penetration rate of sales nationally by four years, as is the case for current HEV adoption in Minnesota relative to the US. Therefore, the penetration of PHEV and EVs for new cars sold in Minnesota is 16% in 2035 and 56% in 2050. The modeled rate of growth corresponds to a near 20% annual increase in PHEV and EV sales until 2030 before leveling off, which agrees with projections of EV sales by the International Energy Agency. [24] 

The penetration of HEV, PHEV and EVs into the stock of all vehicles within the State of Minnesota will lag the penetration of electric drive vehicles into new car sales. The average age of light duty vehicle in the US is 11.5 years old [25] and thus the time required for the total vehicle stock to reach the levels shown in Figure 6.5 will lag the makeup of new car sales by more than 10 years.

The adoption of electric drive technology within the heavy duty vehicle fleet will be significantly more limited. Specialized routes that require stop and start vehicle operation, such as those traveled by buses and garbage trucks, are likely to adopt a degree of electric hybridization. A recent study of bus fleets has shown that hybrid-electric bus operation can reduce fuel consumption and help to achieve noxious and GHG emissions reductions within urban environments, but costs for hybrid conversion must be reduced for the emissions savings to be cost effective relative to traditional diesel technologies. [26]  

Figure 6.5: Historic [13] and projected HEV (blue circles) and combined PHEV and EV (red circles) adoption rate as a percentage of total new car sales for the US and MN. The dotted,

solid and dashed black lines represent the predicted percentage of US HEVs, US PHEV and

EVs, and Minnesota PHEV and EV as determined by S-Curve fits (S(t) = K/[1+exp(-b(t-t0)], K=0.95, b=0.13, t_0,US=2043, t_0,MN=2043, R²=0.79).

6.4.3 Dynamic Wireless Power Transfer

Due to the limitations of vehicle batteries (weight and charging), recent efforts have sought to identify the potential for on-road charging via dynamic wireless power transfer (DWPT). DWPT approaches embed charging coils within the roadways to transfer power to vehicles while in use. The UK government recently released the most comprehensive feasibility study to date on DWPT, which included a large survey of stakeholders, identified technology requirements, examined costs and impacts and outlined future on-road testing of the system. [20]

Respondents to the surveys indicated that industrial stakeholders and private consumers are more likely to purchase an EV if DWPT was available on highways. Private consumers also expressed an interest in deployment of DWPT capacity on A roads (major highways). The adoption of vehicles with DWPT charging capability is likely to require an 18 month to three year return on investment and would require that the DWPT system be user-friendly, practical and simple and reduce CO₂ emissions. [20]

DWTP vehicles are likely to be more expensive than EVs, which are already more expensive than diesel and gasoline vehicles. Early adopters of the DWTP technology are likely to include light weight delivery vehicles (< 32 tons). [20]

The DWPT technologies evaluated (17 total) were found to be between a Technology

Readiness Level (TRL) 4 and 8 with a manufacturing readiness level (MRL) between 3 and 7. No DWPT systems are currently available on the open market, although several are undergoing experimental trials. It was suggested that DWPT technologies could support autonomous vehicle functionality which would add to the benefits of installation. Installation of the DWTP system could coincide with load sensors and other maintenance. [20]

EV batteries and required DWTP charging systems depend on the vehicle and drive cycle with larger, heavier vehicles requiring substantially higher power transfer. Cars, vans and SUVs could viably be used in fully electric mode, with, DWPT increasing range and/or reducing required battery capacity. Heavy duty trucks, however, have a larger power requirement than what would be practical for DWTP to provide thus limiting heavy duty trucks to operation in hybrid mode. [20]

Three different DWTP installation practices are considered: trench-based construction (where a trench is excavated in the roadway for installation of the DWPT primary coils), full lane reconstruction (where the full depth of bound layers are removed, the primary coils installed and the whole lane resurfaced), and full lane prefabricated construction (where the full depth of bound layers are removed and replaced by pre-fabricated full lane width sections containing the complete in-road system). The trench-based and full lane reconstruction were identified as cost-effective options, whereas full lane prefabrication was too new to confirm as a cost-effective option.[20]

Off road trials were suggested with different DWPT road construction methods and different

DWPT system manufacturers to investigate the potential long term impacts on road degradation. Test track trials are planned that investigate the system charging performance, reliability and safety that will be conducted. [20]

South Korea is currently trialing DWPT charging along a 15 mile stretch of road in an 8 km (5 mile) stretch of highway in the Gumi. So far during testing, engineers have recorded an 85% transmission efficiency with the cables and coils. [21]

An alternative to charging batteries is combining fast charging systems with super (or ultra) capacitor energy storage. Supercapacitors have a high power density that allows them to charge and discharge faster than typical batteries. The high power density of capacitors is offset by the energy density which is an order of magnitude or more lower than conventional batteries, see Figure 6.6. The higher power density lends itself well to intermittent charging systems whereby regularly spaced charging on specific routes can satisfy driving demands even for large public transit system. Demonstrator projects are being trailed in the China for above-ground streetcars that are able to be installed without overhead wires and have lower capital and operating costs than continuous coils [22]. The multiple charging and storage technologies highlight the opportunities for optimization of vehicle storage and charging systems.  In the Minnesota context, Arterial BRT would be an opportunity to consider for DWPT.

Figure 6.6: Energy and power density of energy storage technologies that can be supplied to vehicles. [23]

6.5      Natural Gas

While the light duty vehicle fleet is likely to increase the adoption of electric drivetrains, the power requirements for heavy duty vehicles are prohibitive for current and foreseen battery technology. Therefore, fleet managers are seeking other power sources to enable lower operational cost, reduced noxious and lower GHG emissions. An emerging fuel alternative is natural gas which has become attractive in recent years as a result of the low price of natural gas relative to diesel fuel. 

Natural gas offers several advantages to diesel fuel. It has lower CO₂ emissions per unit energy (50.3 gCO₂/MJ) than diesel fuel (69.4 gCO₂/MJ), produces less particulate matter and NO_x emissions than diesel fuels, and has been 20-40% cheaper than diesel fuel on an equivalent energy basis since 2005 [27]. Conversely, natural gas has a lower energy density per unit volume (9.3 MJ/L) than diesel (35.8 MJ/L), typically has lower combustion efficiency, and when used in internal combustion engines can emit significant quantities of methane resulting in higher GHG emissions than diesel engines. [26]

Despite the disadvantages, the benefits of natural gas as a transportation fuel have resulted in an increased adoption rate of the energy source for heavy duty vehicles. As shown in Figure 6.7, the EIA predicts that natural gas will be consumed at increasing rates for heavy duty vehicle use. By 2040 the percentage of natural gas use is expected to reach nearly 7% of total energy use for the heavy duty vehicle sector.

Figure 6.7: Natural gas portion of heavy duty fuel use projection from 2012 to 2040 [28].

In order to enable natural gas heavy duty vehicles a sufficient refueling infrastructure must be in place to provide full deployment. Cities such as Los Angeles have recently deployed compressed natural gas (CNG) refueling stations throughout the metropolitan area, which has allowed them to adopt a bus fleet almost entirely powered by natural gas [29]. The benefit of using natural gas fuel outweighs the additional cost of CNG vehicles (≈10%) when the price of natural gas is 40–50% below diesel prices. This price differential maintains the payback period at 3–4 years or less [30].

6.6       Other Fuel Alternatives

Many other fuels have been offered as an alternative to traditional gasoline and diesel fuels. Additional liquid hydrocarbons, such as dimethyl ether, methanol and butanol, may be combusted in internal combustion engines similar to gasoline and diesel

6.7      Fuel Tax Revenue

Fuel tax revenues are likely to decrease from duties imposed on gasoline and diesel fuel use in the future. The primary driver for decreased revenues in the short and medium term is the increase in vehicle fuel efficiencies resulting from CAFE regulations. The Congressional Budget Office estimates that fuel tax revenues could fall by as much as 21% by 2040 relative to 2012 levels. [31]  

Alternative fuels are taxed at the pump in a similar manner to gasoline and diesel, however different tax rates apply in Minnesota. The Minnesota Department of Revenue taxes E85 at the pump at a rate of $0.2025 per gallon, pure biodiesel (B100) is taxed at $0.285, liquefied natural gas is taxed at $0.171 per gallon, and compressed natural gas is taxed at the rate of $2.474 per thousand cubic feet. Gasoline and diesel are both taxed at the rate of $0.285 per gallon. [32] Currently, the State of Minnesota does not tax electricity for vehicles, while five other states (CO, NE, NC, VA and WA) all have enacted some form of tax on electricity used for transportation. As electric vehicles begin to makeup a larger portion of the fleet taxes on electricity used to power vehicles will likely be adopted within all states including Minnesota (See Chapter 7). Ironically, given Minnesota’s Motor Vehicle Sales Tax, current price differentials between electric and conventional vehicles lead to greater sales tax revenue that currently offset the fuel tax revenue losses from electric vehicle charging. The differential may decrease as battery and electric vehicle prices come down.

6.8       Vehicle Mass 

Alternative fuels offer the opportunity to alter the drivetrain and thus impact vehicle characteristics such as acceleration and mass. Vehicle mass in turn impacts both the road network and fuel consumption of the transportation fleet. A focus on increasing fuel efficiency of the vehicle fleet will have the opposite effect on light duty and heavy duty vehicles, as the relationship between fuel consumption and mass encourages mass reduction in light duty vehicles and increased delivery weight of heavy goods vehicles. 

Recent work by Martin [33] has demonstrated that the mass of vehicles has remained relatively constant in recent years, but gains in fuel efficiency are possible as a result of vehicle mass reduction. For light duty vehicles a Ricardo analysis found a 0.5-1.5% improvement in fuel economy was shown possible for every 100 lb. (45 kg) decrease in vehicle weight for a range of vehicle classes and engines [34]. An analysis by Heywood et al. has predicted that it is possible for the light duty fleet to reduce vehicle mass by 20% for like vehicles [34]. While the reduction in vehicle weight is achievable within individual vehicles or vehicle segments, the total weight of the vehicle fleet is ultimately determined by consumer choice amongst vehicle segments, where trends have shown an increase in the purchase of larger vehicle classes. The available US light duty vehicle market has seen an increase in SUVs since their introduction in the 1980s, now representing 37% of all vehicles sold in 2015, which tend to be larger and heavier [35]. Consumer vehicle choice is influenced by a mix of vehicle attributes of which mass is influenced by preferences for increased acceleration and fuel efficiency. As shown in Figure 6.8, the mass of vehicles has remained nearly constant for the last decade while fuel economy has improved and acceleration has increased as a result of engine improvements. 

Significant reductions in vehicle mass would be possible as a result of more dramatic shifts in the vehicle ownership model. Vehicle sharing programs such as Car2Go increase the possibility of deploying large number of small vehicles within the transportation network that are specifically suited to transporting individuals, and thus, negate the need for larger vehicles that are often purchased in order to serve a variety of vehicle tasks (See Chapter 4). The changes in the ownership model are likely to coincide with other shifts in the light duty vehicle sector, such as adoption of autonomous vehicles (Chapter 1) and vehicle electrification.  People may also be more willing to use light and narrow one-person autonomous vehicles that were considered safe and enclosed from the elements (unlike today’s motorcycles).

Figure 6.8: Change in adjusted fuel economy, weight and horsepower from model year 1975 to 2014. [35]

Within the heavy duty vehicle fleet, the drive for fuel efficiency encourages fleet owners to carry larger loads which reduce the fuel consumption per ton-mile of delivered good. Heavy duty vehicles have significantly higher fuel consumption per ton-mile (or per tonne-km) when operated with partial loads. As shown in Figure 6.9, the energy index (kJ/tonne-km) is highest for small trucks with low load factors (the load factor is the fraction of load the vehicle is carrying relative to its full mass load capacity). There is a ≈45% energy index penalty for half loaded vehicles. The energy index reduces as the delivery vehicle size increases from Rigid to A-Double.    

An analysis of worldwide truck fleets indicates that Australia's truck freight transport is the least energy intensive, using about a quarter less energy per ton-mile than truck freight transport in the United States. This is in part a consequence of a high share of three-unit long haul trucks responsible for transporting a majority of freight through Australia’s interior. [36]

Figure 6.9: Energy index of heavy duty trucks for varying load factors. (Courtesy Prof. David Cebon, University of Cambridge).

6.9       Conclusion and Discussion 

As alternative fuels are adopted within Minnesota there are likely to be a number of impacts to the state’s transportation sector. It is likely that biofuel consumption is near saturation and future shifts will likely be between biofuels, as efforts focus on advanced and cellulosic biofuel production and consumption. These shifts may alter the routes of heavy goods vehicles in the state, as refineries shift from corn and soy feed stocks to cellulose derived from agricultural wastes and forest byproducts. In all cases the increased demand on Minnesota roadways is likely to be minimal as refineries will seek to locate their facilities as close as possible to biomass sources to lower production costs.

Electrification of the vehicle fleet is likely to occur within the light duty vehicle sector. The increased number of electric vehicles will require increased deployment of charging infrastructure. In order to encourage adoption of electric power trains charging stations must likely be supplied or subsidized by government agencies to develop a sufficient network of stations. Long term potential exists for deployment of an on-road charging system which would drastically improve vehicle efficiencies by reducing the weight of batteries required to be carried within vehicles. The installation and maintenance costs for on-road charging are not well known and are an area of ongoing study.

Natural gas vehicles are likely to become an increasing share of heavy duty vehicle fleet so long as the price of natural gas remains 40-50% lower than diesel fuel on an equivalent energy basis. A larger natural gas refueling infrastructure will need to be developed in order to serve an increased number of natural gas vehicles. It is likely that the refueling infrastructure will be developed by private organizations that manage fleets of vehicles with distinct end points. Larger deployment of natural gas vehicles would be enabled by state incentives for natural gas refueling infrastructure. Efforts must ensure that natural gas vehicles and refueling infrastructure do not emit significant quantities of methane, as the high global warming potential (34 times greater than CO₂) can negate any GHG benefits relative to diesel fuel.

Fuel tax revenues are likely to decline in future years as a result of improved vehicle efficiency, rather than a switch to different fuels. Despite the drive for fuel efficiency and incentivize electric vehicle use, a method for producing revenues from electrically-powered vehicles will need to implemented by the State of Minnesota to maintain a long-term funding mechanism for the transportation road network.

Finally emphasis on fuel efficiency in the light duty and heavy duty vehicle fleet is likely going to drive the weight of the vehicle segments in opposite directions. Light duty vehicles are likely to get lighter, especially as different ownership models allow for dedicated light duty vehicle fleets that focus on fuel efficiency for personal mobility. Heavy-duty vehicle fleet operators are likely to lobby for increased vehicle weight limits on Minnesota roadways in order to reduce the energy intensity of goods deliveries. The growing disparity in weight between the two vehicle classes may necessitate increased safety measures to reduce the severity of crashes between the disparate vehicle classes.

           

Chapter 7: Road Pricing

7.1       Introduction

Today, nationally user fees are not the primary source of roadway funds. This is especially true for local roads, as most localities fail to have a user fee the way states do, and instead rely primarily on property taxes (with some transfer of funds from the state, which may or may not be user-fee based). General revenue sources spread costs across non-users as well as users. They also send no signal about the appropriate amount of roads that should be built or how scarce road space should be allocated.  Like everything this share is disputed and depends on accounting (For instance, are Motor Vehicle Sales Taxes a user fee? Not if there is a general sales tax and motor vehicles are exempted from it, yes if they are in addition to it. Are gasoline taxes a user fee? Yes, except if gasoline is exempted from standard sales taxes that applies to most goods, in which case there is a hidden cross-subsidy. These vary by state.)^[49]

While the gas tax is better than the alternative of general revenue, or not paying for roads at all, it doesn't address some important problems. Today’s gas tax does not:

1.    Account for cost inflation in the road sector. 

2.    Account for rising fuel efficiency. 

3.    Pay for the full cost of building and maintaining local roads.

4.    Pay for air pollution.

5.    Pay for the full cost of crashes, which are borne individually through worsened health and life outcomes, and socially through the health care system. 

6.    Raise revenue from vehicles that do not use gasoline for fuel.

7.    Recover full costs of pavement damage from heavy vehicles.⁶⁶

8.    Address congestion, which requires time of day differentiation.  Traffic congestion is a problem. It is not getting measurably worse over the past decade, but it is not getting obviously better. Even if traffic reduces in the aggregate, it won't disappear to zero in the next decade. 

In principle, the first two points can be addressed with regular rate adjustments or indexing of the gas tax.  In practice, indexing has not been popular in the US. Massachusetts voters recently overturned a legislative plan to implement gas tax indexing there,^[50] and Wisconsin repealed their indexing.^[51] The public is uncomfortable writing a blank check to government, even if the revenue is constitutionally dedicated to transportation in general or roads in particular. The Federal gas tax was last raised in 1993, and in Minnesota in 2008 (and before that 1998), so rate adjustments from legislative bodies are infrequent. Other states have raised gas taxes in recent years.^[52] The popularity of gas tax increases (and other sources of funding) among the citizenry varies across polls, and depends precisely on the way the question is framed.^[53]

The third point can be addressed with a rate increase and adjustment to formulas. There is nothing in principle preventing a higher gas tax with funds returned to localities, in exchange for a reduction in local tax rates. That it has not happened is an indicator it faces political obstacles.

The fourth and fifth point can be addressed with a pollution or carbon tax^[54] and distance-based insurance (pay-at-the-pump)^[55] respectively, though the exact rates for these externalities will remain subject to controversy.

The sixth point cannot be addressed with a gas tax, and is a rising problem, but not yet at the point of crisis. This is discussed further below.

The seventh point could be addressed with a weight-distance tax, as in Oregon.^[56]

The eighth point is the most salient immediate justification for moving away from a gas tax. 

Congestion (queueing) occurs when demand exceeds supply for a period of time at a location. This results in delay (higher than free flow travel times) for travelers, which is an economic loss.  It turns out that many transportation systems management strategies are effective at the edge of congestion. For instance ramp metering, the traffic light at the end of the freeway on-ramp that tells a driver whether she can enter, is most effective by keeping traffic just below the critical point at which congestion sets in. If traffic is far below that critical point, there is no danger of significantly higher congestion in letting an additional vehicle on to the roadway. If traffic is well past that point, there is reduced value in not letting an additional vehicle on, traffic will be stop-and-go in any case, though logically congestion dissipates faster when fewer new vehicles are added to the road. It is near that critical point where it matters most.

It is often noted that if just 10 percent of cars were removed during rush hour, there would be little or no delay. This is true in a way. The problem would be the response of traffic. If there were no delay or other penalty, more cars would try to travel at that time, so getting rid of one slice of vehicles will induce another slice of vehicles to travel.^[57] 

Overall though, despite induced demand, in general higher prices deter demand,^[58] as marginal travelers shift routes, time of day, mode, destination, or forego the trip altogether to avoid the higher price. After paying for toll collection costs and the capital and operating costs of the toll facility in question, the revenue from a publicly owned toll facility can offset other transportation taxes, subsidize transportation investment, be returned to taxpayers, or be invested in something else. What is done with the money is primarily a political decision.

A transition from gas tax and other sources of revenues to a vehicle weight-distance tax with prices varying by time-of-day is a plausible path by which dynamic pricing can become mainstream. Such a system might be phased in. 

Economists have long suggested using a price mechanism to allocate scarce road space. There are a variety of approaches in practice.^[59] Several of these are discussed in the sections below.

7.2      Toll Roads

“Fuel taxes paid at the pump [are] relatively invisible to the public, simple to collect, difficult to evade, and collection and administration [are] inexpensive. In contrast, MBUF [is] not easy to understand (especially the complexities and distaste of collecting mileage and location information) and not as easy to collect.”^[60] Swenson et al. develop a scenario with toll roads as the path toward the same end state, positing 25,000 miles of toll roads in 2050 (compared with 6000 miles in 2015).

US National Highway System

Toll roads and bridges, long used as a simple infrastructure funding mechanism (often for bond repayment) with fixed rates, have begun to implement prices that vary by time of day, so that in addition to raising funds for a particular facility, they also manage traffic on that same facility by charging more in the peak and less in the off-peak. 

Pricing is used in many sectors (including transportation: notably all freight transportation, passenger aviation, and public transit), it has been resisted on roads where it might do a great deal of good. (Despite the various mechanisms, toll revenue is less than 1 percent of state transportation funds in Minnesota, though it varies by state, and is over 46 percent in Delaware, Nationally it is less than 6.9 percent of state transportation revenue).^[61] 

This resistance stemmed originally from the inefficiency of toll collection with toll booths, which resulted in delay rather than relieving it (and were never billed as congestion pricing, but were simply a revenue mechanism). With the advent of the Electronic Toll Collection in the 1990s, resistance moved more toward questions of “double taxation”, “equity” and “privacy”. In principle, the question of double taxation is easily dealt with if other taxes are in fact reduced with the introduction of pricing. Some equity implications are summarized in Levinson (2010).⁷⁹ Overall the equity issues depend on how the system is implemented. However in a world of GPS, cell phones, EZ-Pass,⁸⁰ and the NSA, few believe that toll collectors with electronic toll collection systems will not be tracking the location of users, even with public sector protestations and the development of privacy-protecting technologies.

The adoption of Electronic Toll Collection to replace Manual Toll Collection is nearly complete. However the deployment of new toll roads is barely faster than the construction of roads overall.

Figure 1 shows the cumulative mileage of toll roads in the US.⁸¹ Based on a logistic curve extrapolated on current growth rates from 1987 forward, it would take until 2331 for half of the National Highway System (75,000 of 150,000 miles) to be tolled. Expectations of new toll road construction with both constrained resources and falling per capita demand, prospects for new toll roads are dim.

This is not to say there will never be a new toll road in Minnesota, but this is unlikely to be significant from a systems perspective. Several opportunities for new toll roads in the past two decades (US 212, Mn610, the St. Croix River Bridge) have been foregone in what were seemingly better circumstances than are expected for new rights-of-way in coming years.

Based on the observed growth of toll roads in the US, and the reluctance to convert existing roads to tolls, a major uptake of toll roads is unlikely. The only untolled roads that have been successfully tolled in the US are HOV lanes, which are on a different trajectory, and generally remain free for HOV users (as discussed below).⁸²

                                                 

⁷⁹     Levinson, David (2010) Equity Effects of Road Pricing: A Review. Transport Reviews 30(1) 33-57. http://nexus.umn.edu/Papers/TransportEquityReviewPaper.pdf  

⁸⁰     Hirose, Mariko (2015) “Newly Obtained Records Reveal Extensive Monitoring of E-ZPass Tags Throughout New York.” https://www.aclu.org/blog/free-future/newly-obtained-records-reveal-extensive-monitoring-e-zpass-tagsthroughout-new-york and Stanley, Jay (2015) “Christie Use of Tollbooth Data and Why Location Privacy Must Be Protected” https://www.aclu.org/blog/free-future/christie-use-tollbooth-data-and-why-location-privacy-must-beprotected  

⁸¹     Data for Figure 1 comes from several sources: 

Miles of Toll Roads http://www.fhwa.dot.gov/policyinformation/tollpage/miletrends.cfm  http://ntl.bts.gov/lib/5000/5700/5741/toll95.pdf Accessed July 6, 2015

Toll Facilities in the United States: February 1999 Publication No. FHWA-PL-99-011 Internet: http://www.fhwa.dot.gov/ohim (http://ntl.bts.gov/lib/21000/21900/21924/PB99148959.pdf) Accessed July 6, 2015 Toll Facilities in the United States: Toll Mileage Trends — 2003 to 2013  (Interstate and Non-Interstate Bridges, Tunnels, and Roads) Miles of Toll Bridges and Tunnels

http://www.fhwa.dot.gov/policyinformation/tollpage/miletrends.cfm Accessed July 6, 2015

TOLL FACILITIES IN THE UNITED STATES Bridges - Roads - Tunnels - Ferries August 2009 Publication No:

FHWA-PL-09-00021 Internet: http://www.fhwa.dot.gov/ohim/tollpage.htm found at http://www.bv.transports.gouv.qc.ca/per/0961949/05_2009.pdf Accessed July 6, 2015

1940-1990 Data from International Bridge Tunnel and Turnpike Association, Washington DC, reported in GomezIbanez and Meyer Going Private p.169. Brookings Institution.

⁸²     Note that HOV occupancy restrictions sometimes do change, for instance from HOV-2 to HOV-3.

NET: Conversion of existing untolled roads to tolls is also likely to be a non-starter both due to collection costs and the expected political pushback, particular if this is done selectively. Users of a road proposed for conversion will complain about why their road is converted to tolls while others aren’t.

7.3       Toll Areas

Cordon or area-based pricing charges users crossing into or traveling within a particular congested area. Prices vary by time of day to a greater or lesser degree of refinement. This has been implemented in several cities, notably Singapore, Stockholm, Gothenburg, Milan, and London. It has been proposed in several US cities, including New York, San Francisco, and Washington DC, which all have natural features (major rivers) that serve as logical cordons. None of those proposals has yet made it off the drawing board.

NET: Prospects for a London-style congestion cordon in Minnesota (for instance in downtown Minneapolis) are unlikely. Many cities which have significantly more congestion have yet to go forward with such a strategy. These systems are expensive to implement on a per use basis, and while they are effective in discouraging through trips in cities, auto trips to cities can be more easily metered with parking charges (at least until autonomous vehicles become widespread, as discussed in Chapter 1).

7.4      Toll Lanes

High Occupancy/Toll (HOT) lanes, including the local MnPASS system in Minnesota are separate lanes on an existing facility that have prices varying by time-of-day or dynamically (in real-time, based on actual traffic levels), while giving a discount (full or partial) to high-occupancy vehicles. These lanes were historically conversions of under-utilized HOV lanes, but many now are new construction.

The brightest spot in tolling today is the growth of HOT Lanes in the United States, as shown in Figure 7.2. From a slow start in the 1990s, growth has picked up in the past decade significantly, with a number of projects recently opened and many more underway. The graph shows projects that are open, under construction, or with contracts signed. The number of projects in planning stages are more still. While still a small share of the total highway mileage in the United States (there are 4 million miles of streets and roads in the US, the National Highway System has 150,000 miles, the Interstates are 46,000 miles of that), it is growing rapidly, and should eventually absorb the US HOV network and then some. Conversion of HOV lanes to HOT lanes itself is too small to make an important difference (this has already taken place in the Twin Cities region). Rather, if the system is to grow, sources of growth will need to come from new capacity.

Applying a best-fit logistic growth curve, at current rates of growth, HOT Lanes, projected at 722 lane miles nationally in 2020 based on current and let projects, could grow to 10,000 lane miles by 2035, 32,000 in 2044, and 50,000 miles by 2050, on pace to maximum of about 64,000 miles. In short, this rate of growth implies there will be HOT Lanes on every mile of urban and suburban freeway, and perhaps on selected arterials. Whether this is due to General Purpose lane conversion or new construction cannot be determined from such simple modeling, and inevitably will fit the context to some extent. To date it has been mostly new construction, but this change needs to be seen in the context of vehicle automation and more efficient use of existing pavement roadspace. This suggests that as lanes are narrowed and new lanes created from existing lanes and underutilized shoulders, some of those lanes will be Express or HOT lanes.

HOT lanes could be added to many regional highways as they are expanded or reconfigured with narrower lanes (See Chapter 1), the current pace of 1 freeway every 6 years in a metro area with so many freeways (2004: I-394, 2010:  I-35W S of Minneapolis, 20156: I-35E N of St. Paul) suggests about 5 decades before the entire Metro area freeway network has express toll lanes available. And that does nothing for surface streets. Yet if the S-Curve is correct, we should expect that growth to accelerate past the current local rate.

Minnesota DOT has developed extensive plans for the deployment of an Express Lanes (HOT Lane) Network. Two sections are already open, and one is currently under construction, while others in various stages of planning.^[62] The value of HOT lanes can be combined with the opportunity to use these lanes for freeway-based Bus Rapid Transit (express buses) now, and as a separate restricted network for Automated Vehicles in the future. 

While currently HOT lane (and toll systems in general) use different electronic toll collection systems, federal requirements are moving towards interoperability, so out-of-region travelers can use transponders to pay for toll route travel.

NET: HOT Lanes are likely, but not significant from a system revenue perspective. The MnPASS system currently serves fewer than 1 percent of Minnesotans daily. Deployment remains slow, only as freeways are rebuilt.